(Columbia Studies in The Classical Tradition 39) Harris, William Vernon-The Ancient Mediterranean Environment Between Science and History-Brill Academic (2013)

The Ancient Mediterranean Environment
between Science and History

Columbia Studies in the
Classical Tradition
Editorial Board
William V. Harris (editor)
Alan Cameron, Suzanne Said,
Kathy H. Eden, Gareth D. Williams, Holger A. Klein
VOLUME 39
The titles published in this series are listed at brill.com/csct

The Ancient Mediterranean
Environment between Science
and History
Edited by
W.V. Harris
LEIDEN BOSTON
2013
Cover illustration: Fresco from the Casa del Bracciale dOro, Insula Occidentalis 42, Pompeii.
Photograph Stefano Bolognini. Courtesy of the Soprintendenza Archeologica di Pompei.
Library of Congress Cataloging-in-Publication Data
The ancient Mediterranean environment between science and history / edited by W.V. Harris.
pages cm. (Columbia studies in the classical tradition, ISSN 0166-1302 ; volume 39)
Includes bibliographical references and index.
ISBN 978-90-04-25343-8 (hardback : alk. paper) ISBN 978-90-04-25405-3 (e-book) 1. Human
ecologyMediterranean RegionHistory. 2. Mediterranean RegionEnvironmental
conditionsHistory. 3. NatureEffect of human beings onMediterranean RegionHistory. I. Harris,
William V. (William Vernon) author, editor of compilation.
GF541.A64 2013
550.937dc23
2013021551
This publication has been typeset in the multilingual Brill typeface. With over 5,100 characters
covering Latin, IPA, Greek, and Cyrillic, this typeface is especially suitable for use in the humanities.
For more information, please see www.brill.com/brill-typeface.
ISSN 0166-1302
ISBN 978-90-04-25343-8 (hardback)
ISBN 978-90-04-25405-3 (e-book)
Copyright 2013 by The Trustees of Columbia University in the City of New York.
Koninklijke Brill NV incorporates the imprints Brill, Global Oriental, Hotei Publishing,
IDC Publishers and Martinus Nijhoff Publishers.
All rights reserved. No part of this publication may be reproduced, translated, stored in
a retrieval system, or transmitted in any form or by any means, electronic, mechanical,
photocopying, recording or otherwise, without prior written permission from the publisher.
Authorization to photocopy items for internal or personal use is granted by Koninklijke Brill NV
provided that the appropriate fees are paid directly to The Copyright Clearance Center,
222 Rosewood Drive, Suite 910, Danvers, MA 01923, USA.
Fees are subject to change.
This book is printed on acid-free paper.

CONTENTS
List of Tables and Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

Notes on Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xix
What Kind of Environmental History for Antiquity? . . . . . . . . . . . . . . . . . . . 1

W.V. Harris
PART ONE
FRAMEWORKS
Energy Consumption in the Roman World . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Paolo Malanima
Fuelling Ancient Mediterranean Cities: A Framework for Charcoal
Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
Robyn Veal
PART TWO
CLIMATE
What Climate Science, Ausonius, Nile Floods, Rye, and Thatch Tell Us
about the Environmental History of the Roman Empire . . . . . . . . . . . 61
Michael McCormick
Megadroughts, ENSO, and the Invasion of Late-Roman Europe by the
Huns and Avars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Edward R. Cook
The Roman World and Climate: Context, Relevance of Climate
Change, and Some Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Sturt Manning
vi contents
PART THREE
WOODLANDS
Defining and Detecting Mediterranean Deforestation, 800 bce to

700ce . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
W.V. Harris
PART FOUR
AREA REPORTS
Problems of Relating Environmental History and Human Settlement

in the Classical and Late Classical Periods: The Example of
Southern Jordan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
Paula Kouki
Human-Environment Interactions in the Southern Tyrrhenian
Coastal Area: Hypotheses from Neapolis and Elea-Velia . . . . . . . . . . . 213
Elda Russo Ermolli, Paola Romano, and Maria Rosaria Ruello
Large-Scale Water Management Projects in Roman Central-Southern
Italy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
Duncan Keenan-Jones
PART FIVE
FINALE
The Mediterranean Environment in Ancient History: Perspectives

and Prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
Andrew Wilson
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
LIST OF TABLES AND FIGURES
Malanima
Tables
1 Energy consumption in the early Roman Empire . . . . . . . . . . . . . . . . . . 17
2 Energy consumption in advanced regions of the West and East
according to I. Morris. 8000bc2000ad . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Figures
1 Dated remains of coal in England 1500ad . . . . . . . . . . . . . . . . . . . . . . . . . 23
2 Oxygen isotopes in the ice carrot GISP2 (Greenland glacier ice
core) 60bc350ad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3 Intensity of precipitation between 400bc and ad 400 (and range
of error) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4 Estimates of forest clearance in Central Europe (Germany,
North-Eastern France) from archaeological wood remains
200bc400ad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
5 Food consumption by modern populations according to age . . . . . . 31
Veal
Figures
1 Factors affecting the wood supply, which underpin the types of
archaeological charcoals found. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2 Examples of charcoals from excavation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3 Modern charcoal stack ready for covering with mud, leaves and
charring residues, which was then set to char by insertion of a
burning log . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4 Summary results of diachronic study of wood fuel of Pompeii c.
third c. bc to ad79 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
viii list of tables and figures
McCormick
Tables
1 Nile Floods: overview of broader qualities as classified by
Bonneau 1971 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
2 Detailed categories of flood qualities of the Nile according to
Bonneau 1971 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3 Recording quality as assessed by Bonneau . . . . . . . . . . . . . . . . . . . . . . . . . 80
Figures
1 Reconstructed precipitation anomalies (mm/day), April, May,
June, 367378ad, northeast France . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
2 Percentages of Nile flood qualities, early vs. later Roman Empire . . 78
Cook
Figures
1 The Dulan-Wulan annual tree-ring chronology from
north-central China and the occurrence of severe droughts
during the times of the Hun-Avar migrations into late-Roman
Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
2 Correlations between DecemberFebruary Nio 3.4 sea surface
temperatures (a measure of ENSO variability) and MarchJune
total precipitation from 1951 to 2003 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3 Two 2,000 year long annual tree-ring chronologies from ENSO
sensitive regions in the Northern and Southern Hemispheres:
Douglas fir from northwest New Mexico and Kauri from the
North Island of New Zealand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
4 Correlations between the Douglas fir and Kauri tree-ring
chronologies and Hadley Centre global sea surface temperatures
(HadISST1) for the winter season: 18712003. . . . . . . . . . . . . . . . . . . . . . . . 96
5 The average (A) and difference (B) of the Douglas fir and Kauri
tree-ring chronologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
6 Correlations between DecemberFebruary average SSTs and the
average and difference of the Douglass fir and Kauri annual
tree-ring chronologies, each with an identified ENSO signal . . . . . . . 98
7 Comparisons of correlation patterns of MarchJune
precipitation with actual and tree-ring ENSO indices . . . . . . . . . . . . . . 99
list of tables and figures ix
8 The Douglas firKauri average interhemispheric ENSO index

with drought inducing La Nia periods indicated around the
times of the Hun-Avar migrations into late-Roman Europe . . . . . . . . 100
Manning
Figures
1a Comparisons of general northern hemisphere temperature
covering the past millennium and the often differing (almost
opposite) precipitation records from the west and east
Mediterranean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
1b A comparison of periods noted in the analyses of Nicault et al.
(2008) of decadal or longer intervals of wetter (more negative
PDSI) and drier (more positive PDSI) for Italy, Greece and the
Levant versus reconstructed winter NAO indices. . . . . . . . . . . . . . . . . . . 111
2 A. The standard radiocarbon calibration curve for the period
3000bc to ad1950 from known-age trees. B. The 14C record per
mille () from Athis is the relative 14C content decay
corrected and normalized. C. The residual 14C record per mille
() after a 1000-year moving average is removed . . . . . . . . . . . . . . . . . . 123
3 Bottom: observed sun-spot numbers (SSN) per year. Top: The
annual 14C record per mille () ad16001900 and an 11-year
moving average of this record. Middle: The 14C record per mille
() from IntCal09 and IntCal04 and two models of 14C
production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
4 Top: A. The residual annual 14C with 2 point (pt) FFT smoothing
from the data shown in Figure 3 from Stuiver et al. (1998)
calculated minus a 22pt FFT smoothing to emphasise the change
around the longer-term trend. B. The residual annual production
of 14C (iterative methodsee Figure 3) with 2 pt smoothing . . . . . . . 125
5 High resolution 10Be data from Greenland for the most recent six
centuries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
6a Comparison of the Total Solar Irradiance (TSI) reconstructions of
Vieira et al. (2011) from the 14C record versus Steinhilber et al.
(2009) from the 10Be record . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
6b Top: Total Solar Irradiance (TSI) reconstruction from ice-core
10
Be records from Steinhilber et al. (2009)see Figure 6a.
Bottom: 14C production from Marmod09 (Reimer et al. 2009) . . . . . . 129
x list of tables and figures
7 Top: detail of the Total Solar Irradiance (dTSI) reconstruction

from ice-core 10Be records from Steinhilber et al. (2009) for the
period 300bc to ad800. Bottom: two 14C production models: (a)
from Marmod09 (Reimer et al. 2009) and (b) the iterative model
from Usoskin and Kromer (2005) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
8 Top: the two 14C production models in Figure 7now not
invertedfor the period 300bc to ad800. Bottom: the SSN
reconstruction from Solanki et al. (2004) for 300 bc to ad 800 . . . . . . 131
9 The main trends of the solar proxy records in Figures 7 and 8 for
the period 300bc to ad700 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
10 Top: Extra-tropical Northern Hemisphere temperature record. A.
50-year smoothed curve; B. A 10pt FFT smoothed curve. Middle:
IntCal09 radiocarbon calibration curve. Bottom: Ring widths of
ANG-7B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
11 Top: the mean tree-ring widths record for the tree-ring series
from Istanbul. Bottom: the reconstructed temperature and
precipitation records from central European oak time-series
shown in Bntgen et al. (2011, Figure 4) are shown for the time
interval covered by the Istanbul tree-ring data at the top of the
figure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
12 Reconstructed precipitation (AprilMayJune = AMJ) in mm
with respect to the instrumental ad19012000 period record from
the Bntgen et al. (2011) study for central Europe and for
Germany . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
13 Reconstructed summer (JuneJulyAugust = JJA) temperature
anomalies with respect to the instrumental ad 19012000 period
record from the Bntgen et al. (2011) study for central Europe and
for Switzerland . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
14 Comparison of the temperature reconstruction for the
extra-tropical Northern Hemisphere for the last 1000 years by
Christiansen and Lungqvist (2011) with the temperature
anomalies reconstruction over the same period by Bntgen et al.
(2011; 2006) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
15 A comparison of the Bntgen et al. (2011) precipitation record
from Central and Northern Europe versus two east
Mediterranean precipitation reconstructions and the PDSI
reconstruction by Esper et al. (2007) from Morocco . . . . . . . . . . . . . . . . 145
list of tables and figures xi
16 Three 18O plots from Mediterranean locations. A: Soreq Cave

from Bar-Mathews et al. (2003); B: Bucca della Renella from
Drysdale et al. (2006); and C. the record from planktonic
foraminifera (Globigerinoides ruber) from the southeast
Mediterranean off Israel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
17 Four 13C records from Mediterranean region speleothems . . . . . . . . 150
18a Comparison of two temperature proxy records derived from
speleothems (from northern Spain and Austria), with the proxy
temperature records for central Europe and for Switzerland from
tree rings produced by Bntgen et al. Top: the Marmod09 14C
production model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
18bDetail from Figure 18a of the summer temperature
reconstructions from (top) the central Alps speleothem from
Spannagel Cave (Mangini et al. 2005) with (bottom) the central
European temperature reconstruction based on tree-rings in
Bntgen et al. (2011) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
19a Two records of past major volcanism relevant to the Northern
Hemisphere from two Greenland ice-cores . . . . . . . . . . . . . . . . . . . . . . . . 155
19bBottom: number of ring-width growth minima in the Bristlecone
pine (BCP) record of Salzer and Hughes (2007) associated within
5 years of a volcanic signal in an ice-core record. Top: Packages of
years where there are notable decreases in Bristlecone pine
growth and ice-core volcanic eruption signals . . . . . . . . . . . . . . . . . . . . . 156
20 Two temperature reconstructions for high northern latitudes of
the northern hemisphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
21 A comparison of a selection of records or events discussed in the
text for the period 300bc to ad800 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
Kouki
Tables
1 The relevant archaeological periods in Jordan . . . . . . . . . . . . . . . . . . . . . 200
2 A reconstruction of the climate history of southern Levant and
the rural settlement in the Petra region . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
Figures
1 The Petra region, southern Jordan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
2 A modern barrage in the Jabal Harun area in September 2011 . . . . . . 201
xii list of tables and figures
3 Environmental zones in the Petra region . . . . . . . . . . . . . . . . . . . . . . . . . . . 205

4 The distribution of rural settlement in different environmental
zones in the 1st century ad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
zones in the 3rd century ad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
zones in the 6th century ad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
zones in the 7th century ad . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
Ermolli et al.
Figures
1 Location map of the study area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
2 Pollen diagram from the C106 core in the Salerno Gulf . . . . . . . . . . . . . 217
3 Three-step scheme of the evolution of the Neapolis area . . . . . . . . . . . 221
4 Pollen diagram from the Neapolis port sediments. . . . . . . . . . . . . . . . . . 223
5 Three-step scheme of the evolution of the Velia area . . . . . . . . . . . . . . . 227
6 The road to Porta V at Velia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
7 Section of the alluvial deposits at Velia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
Keenan-Jones
Figures
1 The Italian Peninsula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
2 The Bay of Naples in the Roman Period showing the Aqua
Augusta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
3 The area of the proposed Tiber flood prevention works of 15 ad . . . . 249
NOTES ON CONTRIBUTORS
Edward R. Cook is Ewing Research Professor at Lamont-Doherty Earth

Observatory of Columbia University and is the Director of the Tree-Ring
Laboratory there, which he co-founded in 1975. His primary research activi-
ties have been oriented around the development of statistical methods of
tree-ring analysis, with an emphasis on time series modeling and robust
estimation of tree-ring chronologies, all with the goal of producing the best
reconstructions of past climate possible. He co-edited and contributed to
a book entitled Methods of Dendrochronology: Applications in the Environ-
mental Sciences (1990). Cook has over 150 peer review publications and was
elected Fellow of the American Geophysical Union in 2011. His most recent
publication in 2012 is a reconstruction of past summer temperatures from
tree rings in East Asia for the past 1,200 years.
Elda Russo Ermolli is a Researcher in Physical Geography and Geomor-

phology and Professor of Quaternary Geology and Geoarchaeology at the
University of Naples Federico II, and also an Associate Researcher in the Pre-
history Deptartment of the Natural History Museum in Paris, specializing
in Quaternary palynology. Her main research interests concern the climatic
and environmental history studied through pollen analysis of marine, tran-
sitional and continental records. The aims of her work are the reconstruc-
tion of the vegetation cover variations, and their connection to natural and
/or anthropogenic causes. She is the author of more than fifty journal arti-
cles as well as of book chapters and many congress reports.
W.V. Harris is Shepherd Professor of History and Director of the Center for
the Ancient Mediterranean at Columbia University. His most recent book is
Romes Imperial Economy (2011), and he was the editor of and a contributor
to Mental Disorders in Classical Antiquity (2013). For his other publications
see http://www.columbia.edu/history/faculty/Harris.html.
Duncan Keenan-Jones is currently a Mellon Postdoctoral Research Asso-

ciate at the Illinois Program for Research in the Humanities and Visiting
Assistant Professor in the Department of the Classics, University of Illinois
at Urbana-Champaign. He is interested in the use of historical, archaeolog-
ical, quantitative and scientific approaches in the study of ancient society,
xiv notes on contributors
technology and environment. He has recently published on the lead con-

tamination in the drinking water supply of Pompeii, and he has in the press
a quantitative study of lead-pipe inscriptions in Campania and a geoarchae-
ological study of mineral deposits in the aqueducts of Rome.
Paula Kouki has recently finished her PhD at the University of Helsinki. She
is currently working as a researcher for the Finnish Jabal Harun Project at
the University of Helsinki, and is the editor of the forthcoming publication
of the FJHP survey results, PetraThe Mountain of Aaron: The Archaeologi-
cal Survey. She has fieldwork experience in Jordan, Greece and Finland. Her
research interests include Near Eastern Classical and post-Classical archae-
ology, landscape archaeology, human-environment relationships, the per-
ception of time in archaeology and archaeological field survey.
Paolo Malanima is the Director of the Institute of Studies on Mediter-

ranean Societies (ISSM) of the Italian National Council of Research (CNR),
based in Naples. He has been Professor of Economic History and Economics
at the University of Pisa (19771994) and University Magna Graecia in
Catanzaro (19942002). He is a member of the editorial board of the jour-
nals Societ e Storia, Rivista di Storia Economica, Economic History Review
and Scandinavian Economic History Review. He is author of The Pre-Modern
European Economy (Brill, 2009), and (with A. Kander and P. Warde) Power to
the People. Energy in Europe over the last Five Centuries (Princeton, 2013).
Sturt Manning is the Goldwin Smith Professor of Classical Archaeology

and Director of the Malcolm and Carolyn Wiener Laboratory for Aegean and
Near Eastern Dendrochronology at Cornell University. His research interests
cover Mediterranean archaeology (especially the Aegean, Cyprus and east
Mediterranean), and archaeological science (especially dendrochronology,
radiocarbon, and climate). For more information on his publications, see
http://cornell.academia.edu/SturtWManning
Michael McCormick is Francis Goelet Professor of Medieval History at

Harvard University. His books include Origins of the European Economy
(Cambridge University Press, 2002) and Charlemagnes Survey of the Holy
Land (Dumbarton Oaks-Harvard University Press, 2011); he is senior editor of
the Digital Atlas of Roman and Medieval Civilizations (http://darmc.harvard
.edu/). He currently co-directs the excavation of a late Roman settlement in
eastern France, and chairs the Science of the Human Past, a new Harvard
research and teaching network of natural scientific approaches to the past
notes on contributors xv
including human health and the environment, and applying computer sci-
ence to the study of ancient texts.
Paola Romano is Associate Professor at the Earth Science Department of

the Federico II University, Naples. She teaches geomorphology and geoar-
chaeology to both undergraduate and graduate geology students. Her main
research interests fall in the field of morpho-dynamic processes in conti-
nental and coastal environments. In the last decade she has been applying
her expertise to the environmental and morphological reconstruction of
ancient and historical human settlements in southern Italy.
Maria Rosaria Ruello is a post-doctoral researcher in geomorphology and

geoarchaeology at the University of Naples Federico II. Her main research
interests are the paleoenvironmental and geomorphological evolution of
natural contexts and their interaction with archaeological sites.
Robyn Veal earned her PhD at the University of Sydney with a dissertation
on the fuel economy of Pompeii (2009), which will appear as Fuelling Pom-
peii (Accordia, London). Her interests include natural resource economics
in the ancient world; archaeological theory and whole assemblage analy-
sis; and GIS and information management systems for archaeology. She has
been the Ralegh Radford Rome Fellow at the British School at Rome, and
is currently the Anniversary Research Fellow at the McDonald Institute for
Archaeological Research at Cambridge.
Andrew Wilson, Professor of the Archaeology of the Roman Empire, Uni-

versity of Oxford. Recent and forthcoming publications include: Quantifying
the Roman Economy: Methods and Problems (ed. with Alan Bowman, Oxford,
2009), Settlement, Urbanization and Population (ed. with Alan Bowman,
Oxford, 2011); The Roman Agricultural Economy: Organization, Investment,
and Production (ed. with Alan Bowman, Oxford, 2013); Maritime Archaeol-
ogy and Ancient Trade in the Mediterranean (ed. with Damian Robinson,
Oxford 2011); and Saharan trade in the Roman period: short-, medium- and
long-distance trade networks, Azania: Archaeological Research in Africa 47,
4 (2012), 409449.
ABBREVIATIONS
A LAnne pigraphique
AJA American Journal of Archaeology
AJPh American Journal of Philology
Annales HSS Annales: histoire, sciences sociales
AO Arctic Oscillation
AP arboreal pollen
AS Anatolian Studies
BGU Aegyptische Urkunden aus den Kniglichen/Staatlichen Museen zu Berlin
CIL Corpus Inscriptionum Latinarum
C.Ord.Ptol. Corpus des Ordonnances des Ptolmes ed. M.T. Lenger
C.Th. Codex Theodosianus
DJF December January February
ENSO El Nio-Southern Oscillation
FJHP Finnish Jabal Harun Project
GPCC Global Precipitation Climatology Centre
IAWA International Association of Wood Anatomists
IGLS Inscriptions Grecques et Latines de la Syrie
ILS Inscriptiones Latinae Selectae ed. H. Dessau
ITRDB International Tree-Ring Data Bank
JFA Journal of Field Archaeology
JRA Journal of Roman Archaeology
JRS Journal of Roman Studies
KNMI Koninklijk Nederlands Meteorologisch Instituut
LIA Little Ice Age
LibStud Libyan Studies
MCA Medieval Climate Anomaly
NAO North Atlantic Oscillation
NSA Notizie degli scavi di antichit
ORom Opuscula Romana
OSL Optically Stimulated Luminiscence
P. Berl. Leihg. Berliner Leihgabe griechischer Papyri ed. T. Kaln et al.
PBSR Papers of the British School at Rome
PDSI Palmer Drought Series Index
P. Tebt. The Tebtunis Papyri ed. B.P. Grenfell et al.
Sammelbuch Sammelbuch griechischer Urkunden aus Aegypten ed. F. Preisigke et al.
SEG Supplementum Epigraphicum Graecum
SHA Scriptores Historiae Augustae
SSN Sun Spot Number
TSI total solar irradiance
ZPE Zeitschrift fr Papyrologie und Epigraphik
PREFACE
Historians and natural scientists are two separate tribes, both of them to
a great extent endogamous, intellectually speaking. It is true that ever since
Auguste Comte (17981857) there have been historians who have wanted his-
tory to be more scientific in the sense that it would conform to a supposedly
higher epistemology, true also that some of the earliest Annales historians
were interested in, for instance, psychology, as well as more obviously in
physical geography. And there have always been historians of medicine, and,
in a sense, of the environment. It also remains true, however, that few mem-
bers of one tribe have been much attracted by the other.
But now a certain change is taking place. It features some extremists, who
can with some inaccuracy be labeled technological determinists, and no
doubt it will leave a great number of historians untouched even in English-
speaking and French-speaking lands (not to mention Germany or Italy,
where genuine environmental history has made virtually no headway to
date among ancient historians). And it is obviously only in certain scientific
fields that anyone is interested in the historical past.
The change in the attitude of historians that I am referring to has noth-
ing to do, be it noted, with the notion that history should become in some
extended sense an experimental science. In so far as the language of exper-
imentation merely serves in such books as Natural Experiments of History
(2010), edited by Jared Diamond and James A. Robinson, as window-dressing
for comparative history it will do no harm;1 not so, however, if it gives anyone
the idea that a new and superior way of doing history is involved.
No, what is changing for the better is that the subject-matter of his-
tory has widened still further in recent decades to include problems that
have also been, and continue to be, the objects of widespread scientific
attention. Actual collaboration between scientists and historians is still
rather rareor at least inconspicuous, not least because scientists with
historical interests commonly suppose, with varying degrees of justifica-
tion, that they can do without real historians. But whatever the future of
historical-scientific collaboration may be, we historians now have on our
1 For a critical review of this book and its significance see Roth 2012/3.
xx preface
agendas all sorts of problems that require us to leap over the traditional
disciplinary boundaries.
The reluctance of historians to engage with the natural sciences has
roots so much out in the open that there is no need to describe them. But
some growing fields of historical interest, the most conspicuous of which is
probably environmental history, are making this attitude obsolete. Scholars
still write books about the ancient environment that are essentially digests
of what Greek and Roman writers said about the environment or about
natural features (rivers or mountains), but if we want to know what the
environment in antiquity was actually like, and why it developed as it did, we
turn to scientific archaeology, to geology, to climate studies, to palynology,
to botany, and so on.
This book is largely the work of others, but it also springs from my con-
viction that ancient historians, and indeed historians more generally, can
benefit greatly at the present juncture from making more use of some of the
natural sciences and of technological expertise. There are many productive
alliances to be formed. The present collection, as I think all contributors to
it would agree, is full of historical and scientific questions that are far from
being resolved. We shall only get closer to resolving them if we make use of
all relevant methodologies and bodies of knowledge.
For ancient history, in particular, there is a serious challenge here. The
field can look inwards, re-hash well-known textual sources and concoct
new biographies of famous Romans. Or it can put its energy into explor-
ing fresher subjects, with fresher methods. Of course the choice is not a
stark as that, and no one deplores more than I do the factand it is a
factthat there are professional Graeco-Roman historians at large who
have a poor knowledge of Greek and Latin (the great anxiety of the tradi-
tionalists) and are prepared to publish books and articles without know-
ing the bibliographies of the subjects that they write about. No one said
that writing good scholarly history was easy. But if we want to deserve
intellectual respect we must confront modern times as well as ancient
ones.
For the most part this book consists of papers that were given in earlier forms
at a conference entitled History and Environment in the Ancient Mediter-
ranean, held in Rome on June 15th and 16th, 2011. I should like to thank first
of all the speakers who are represented here for their commitment and their
willingness to engage in a somewhat unconventional project. Special thanks
go to Duncan Keenan-Jones, a young scholar whose work was not known
to me when I organized the Rome conference but who afterwards kindly
preface xxi
agreed to contribute a paper about water management, a subject that the

conference neglected.
Some of the conference took place at the American Academy in Rome,
and I should like to thank its officials Carmela Franklin, Christopher
Celenza, Corey Brennan and Anne Coulson for making it possible to meet
there in extraordinarily pleasant surroundings. I owe an equal debt to the
director of the Institutum Romanum Finlandiae, Katariina Mustakallio, for
the opportunity to hold one of the conference sessions at that remarkable
institution. Several other people gave useful advice or provided practical
help with this project, and I most of all thank Roger Bagnall (NYU), Saskia
Hin (Rostock), Joe Manning (Yale), and Nate Pilkington (a former Columbia
student who is about to take up a post-doctoral fellowship at Cornell) for
their assistance. But the person who more than anyone has helped to beat
this volume into shape has been my long-suffering collaborator Emily Cook,
ABD in the Art History and Archaeology Department at this university, to
whom I offer my deep appreciation.
Finally, none of this would have been possible without generous funding
provided by the Andrew W. Mellon Foundation, to whose officers I once
again offer my sincere thanks.
William Harris
Columbia University, February 2013
WHAT KIND OF ENVIRONMENTAL HISTORY FOR ANTIQUITY?
W.V. Harris
What sort of fresh potential does environmental history have for our under-
standing of the Greek and Roman world? (Others must answer the sym-
metrical question: what sort of fresh potential do Greek and Roman his-
tory have for our understanding of changes in our planet?). Environmen-
tal scientists, climatologists, and archaeologists, together with historians
economic and otherwise, have recently been converging on a number of
major issues in the environmental history of the late Holocene period, that
is to say, roughly speaking, historical times. The Mediterranean area has
rightly attracted especially intense attention, partly at least because, as far
as ancient times are concerned, it uniquely combines extensive written
records with an ocean of archaeological information, while also offering the
results of numerous palynological and geological studies. This convergence
seems to promise that before long we shall be able to construct an environ-
mental history of the region that will be more satisfying to both scientists
and historians.
One obstacle of course is, to put it bluntly, that no historian knows enough
science and no scientist knows enough history. In both camps, however,
there are outstanding individuals who have made vigorous efforts to cross
the great divide; and there exist ambidextrous archaeologists who keep up
with both fields as far as that is humanly possible.
Not that one should give way to optimism. The technical complexities of
every field grow and grow, and collaboration between humanists and sci-
entists, while it may be on the increase, faces all sorts of barriers. And it is
by no means clear that when we are considering the late-Holocene environ-
ment we are usually pursuing the same questions. This book is somewhat
experimental in combining historical and scientific approaches, and the
experiment reveals disparities of aims and methods that will continue to
require discussion. We need to clarify our agendas.
It is worth recalling here the distinction usefully stressed by Horden
and Purcell between history in the Mediterranean and the history of the
Mediterranean.1 By the former Horden and Purcell meant the history of
1 Clarified in Horden and Purcell 2005, 357.

2 w.v. harris
politics, society, the economy, religion and warfare that takes up most of the
space in Mediterranean histories. But the Mediterranean world has had its
own physical history, it is not just a setting.2 By the history of the Mediter-
ranean Horden and Purcell meant the history of interactions between man
and the environment, hence a history of human use and misuse of natural
resources, a history of cultivars and rivers, a history of landscapes, settle-
ment patterns, demography and connectivity. The Corrupting Sea, which
was the first book to carry out a part of this agenda (but demography and cli-
mate were reserved for Part II), did not attempt to explain contingent events,
indeed it largely ignored them.
Some of the contributors to this volume, by contrastnotably Malanima,
Cook and McCormickaim to fit together climatic change and historical
events of various kinds.3 Malanima takes a Roman economic decline to
have resulted from population growth (which meant that inferior land had
to be used) and the end of the so-called Roman Warm Period (a decline in
temperatures shortened growing seasons).4 This is a coherent story, but the
facts on which it is based are fragile, and there were other factors at work.
If there was an empire-wide economic decline, it probably set in after the
beginning of the Great Pestilence (the Antonine Plague), in other words
in a period of (possibly sharp) population shrinkage, not growth. And the
pestilence itself was probably a major factor in any economic decline that
did occur.5
As for climate change, the scientific study of the ancient climate has
barely begun (McCormick, below, p. 63),6 which means that this is an inop-
portune time for anything like environmental determinism.7 There are now
multiple sources of information about the ancient climate, and the strong
2 Not that one should complain when historians use this term. What is a ground for com-
plaint is the publication, even within the last decade, of general histories of classical antiquity
that pay no attention to the natural world: see Cambridge Ancient History volume XII, for
example (2005). When, on the other hand Chakrabarty 2009, 201, announces that anthro-
pogenic explanations of climate change are spelling the collapse of the distinction between
natural history and human history, he seems to be catching up with Braudel.
3 Here I discuss approaches and methods and do not attempt to summarize any of the
papers included here.

4 The warmth of the Roman Warm Period must have been quite relative, at least in some
places: Plinys description of his Tifernum villa argues strongly for a climate there colder than
the present one (see Letters 5.6.4, with Grove and Rackham 2001, 142).
5 Harris 2012.
6 See also Ermolli et al., this volume p. 213.
7 As is unwittingly demonstrated by the self-described neo-determinists Issar and Zohar
2007.
what kind of environmental history for antiquity? 3
impression I receive from this literature (see especially Finn et al. 2011,
Manning, this volume) is that the Roman Empire was full of micro-climates,
and that no narrow date can be assigned to the end of the Roman Warm
Period. If, as some suppose, that occurred much later than 150 ad,8 it has
to compete with other possible causes of economic deterioration such as
repeated civil war and a severe currency crisis in the 270s. Nonetheless
Malanimas paper implicitly calls into question the viability of the history
in/history of distinction, or rather it firmly places the economic history of
an agrarian society in the history of category. Bresson, similarly, found it
impossible to separate off the exchange of commodities in antiquity from
the history of.9
Malanimas paper sets out without unprecedented clarity the likely
energy needs of the Roman world, and might lead in any of several inter-
esting directions, such as an attempt to calculate the carrying capacities of
a variety of regions.
The fuel history of the Roman world certainly counts as history of too.
One of the material bases of the whole culture was charcoal, which was the
essential means for producing the high temperatures needed for iron- and
copper-working. This topic is not of course new,10 but it is now receiving
more attention from archaeologists, thanks in good part to the enthusiasm
of Robyn Veal. Her paper in this volume raises the question how much char-
coal was employed in other processes and practices. If there is to be a clear
answer, it will presumably come from further excavation reports and the
new methods of analysing charcoal that she outlines. The most intriguing
result she reports, from Pompeii, is the predominance of beech-wood (fagus
sylvatica) charcoal from the third century bc to the date of the eruption,
with other taxa, including fruit and nut woods in small, but then increasing
amounts, becoming commoner in the first centuries bc and ad. Was this
change simply a matter of preference, or were the beech woods thinning
out?11
8 As late as 400 perhaps: see the bibliography in Manning, this volume, p. 135. Mannings
conclusion leaves a supposed end of the Roman Warm Period c. 150 in limbo, dating the
change to some time in the third century. It is not clear whether the third century was drier or
wetter in central-Mediterranean lands: see on the one hand McCormick (p. 70) and Manning
(p. 163), and on the other Keenan-Jones (p. 234).
9 Bresson 2005, 99100. See further Horden and Purcell 2005, 357.
10 The book edited by Fiorentino and Magri 2008 gives an idea of the state of this field.
11 Incidentally I note that beeches, which do not now grow below 800 metres in Italy,
sometimes grew at notably lower altitudes in the past (Theophrastus, Hist.Plant. 5.8.3, with
Grove and Rackham 2001, 142; Ferrari Fontana et al. 2008), which makes it less likely that
Pompeii was running out of them. Beeches can be coppiced.
4 w.v. harris
McCormicks paper complements Malanimas. Having amusingly settled

the question of the exact year in the 370s when Ausonius wrote his poem
Mosella,12 he generously presents the results of his Dumbarton Oaks research
groups work on the Mediterranean climate, 100 bc to 600 ad. This leads
primarily to some specific questions about agricultural production, though
McCormicks scholarly caution and rigor sometimes give way to the more
general hope that climate change can be correlated with imperial fortunes.
The results proposed so far are regional but intriguing: one is a refinement
of data collected forty years ago by Danielle Bonneau about the annual
flooding of the Nile, which appear to show that (evidently because of lower
precipitation in the rivers headwater region) the Nile flooded less well in
the period 155 to 299ad than it had in previous centuries, with the overall
effect of lowering agricultural production. Whether this had any impact on
the grain supply of the city of Rome is unknown, but it must certainly have
made life more difficult for the Egyptian poor.
Another possibility canvassed by McCormick is that the cultivation of rye
(secale cereale) in central Europe, inside and outside the Roman Empire, was
a response to climatic change, either to lower temperatures, as previously
proposed, or to lessened precipitation, as proposed by McCormick himself.
Like the spread of other cultivars, the spread of rye, a grain known to but
looked down on by classical Mediterranean writers, is a matter of contro-
versy; the very concept of domestication is a complex one.13 We seem to lack
an up-to-date account of the diffusion of rye in Europe, but K.-E. Behres
impressively detailed study of 1992 appears to show that it was cultivated in
free Germany as early as the second century ad. The decisive reason for rye
becoming a crop has to be seen in the change of the harvesting method and
not in climatic deterioration, he claimed.14 But it would certainly be tempt-
ing to associate the spread of rye cultivation in central and northern Europe
with a colder climate if a close chronological fit could be established.
McCormick briefly floats the idea, and Edward Cook develops it in full,
that the Hunnic and Avar invasions of Europe in late antiquity were proba-
bly encouraged (McCormick) by aridification in central Asia. Or that arid-
ification may have contributed to the impulse to move westwards (Cook).
What Cook contributes to strengthening this old theory is recent dendro-
chronological information that indicates a megadrought in central China
12 Unless, that is to say, Ausonius description of the physical conditions of the Moselle
valley wasas seems to me quite likelylargely unlocated in time.

13 Cf. Weiss et al. 2006.
14 Behre 1992, 142, q.v.
(present Qinghai province) about 360ad. With his wide knowledge of cli-
mate systems, he is able to delineate the climate change in question with a
new level of detail. One obvious difficulty is that the Qinghai sites are a good
4,000 kilometres from the earliest known home of the Huns, to the east of the
lower reaches of the Volga. Furthermore, Huns are reasonably well attested
further west, to the north and west of the Black Sea, as early as the 370s.15
These two papers raise the wider question whether climatic variation
has ever been a detectable major cause of social or political events. It is
probably fair to say that most historians are congenitally disinclined to
accept such ideas, fond as they are of human agency and specific cases.
Some recent work may make this scepticism harder to defend. In particular,
it has been argued that ENSO (El Nio/Southern Oscillation) (see Cook,
p. 92, for an explanation of this concept) has been one of the causes of civil
conflicts in tropical countries since 1950.16 The very modesty of the claim in
question (ENSO may have had a role in 21% of all civil conflicts since 1950)
is appealing.
The paper contributed by Sturt Manning, which amounts to a compre-
hensive review of the current state of knowledge concerning the ancient
Mediterranean climate, points out some of the reasons why such are
hypotheses may be problematic:17 on the one hand, obviously, lie the com-
plexities of historical causation,18 on the other the present uncertainties of
climate history.
When we talk about the historical effects of climate change, it is essen-
tial that we consider both alternative explanations and the likely effects on
human beingsand these particular human beingsof climate instabil-
ity, aridification, and changes such as short and long periods of decreased or
increased temperatures, and of decreased or increased precipitation. Should
we think of the Huns leaving central Asia in desperate economic straits and
arriving in Europe as, fairly briefly, a highly formidable force? Think of other
large-scale migrations, such as the Arab expansion of the seventh century
15 Orosius 7.33.10. But where exactly Huns were to be found in the 370s and 380s continues
to be debated.
16 Hsiang et al. 2011.
17 Manning, this volume pp. 114115.
18 Manning 2010 associates the large 14C production peak centered on 765bc with popula-
tion growth in Greece and all the complex cultural changes that began in that period. Such a
climate would have resulted in longer and more reliable growing seasons. While well aware of
the difficulties of establishing any causal relationship (42), he concludes that climate change
may have created conditions that actively promoted development and change in human
societies (44). See Danti in the same volume, esp. 140, for a properly sceptical approach.
6 w.v. harris
or the European occupation of the Americas: when we know about their

underlying motives in detail, they concern the obvious material advantages
of conquering your weaker neighbours, together with religious considera-
tions of various kinds. Which is not to exclude the possibility that traditional
historical methods have under-estimated the influence of climatic change.
It can be hypothesized that climatic events have much more potential for
explaining economic and other consequent difficulties in past societies than
for explaining economic, cultural or any other kind of successor migra-
tions. Thus it has been plausibly arguedthough the argument is far from
closedthat a basic cause of the collapse of Classic Mayan civilization was
a period of recurrent major droughts.19 And it may not, on the other hand,
be in any way an anomaly when Paula Kouki discovers that in southern Jor-
dan, in the region of Petra, the intensification of settlement and agriculture
lagged behind improved climatic conditions by at least two centuries.20
Manning mentions several examples of proposed correlations between
climate change and large-scale successes or advances, none of which has
been adequately demonstrated.21 He seems willing to attribute the so-called
Greek Renaissance that began in the eighth century bc to a climatic change
favourable to agriculture; the point, he says, is not who is right or wrong
rather how the same case can be and is differently interpreted. I entirely
disagree with this last claim, because I want to save climate scientists from
chasing historical mirages: we could possibly say that a climate reason-
ably favourable to agriculture was a necessary condition for the Greek
Renaissance, but it has zero explanatory power.22 Axial age or not, this
was a unique succession of events, which can to a considerable extent be
explainedalthough some would say that the great fascination of classical
Greek history is that its successes remain to a considerable extent unex-
plained.
Manning is, however, generally cautious about climatic causes that have
been attributed to historical events and processes. As he notes, several
scholars have argued that climate change was a factor, even a significant
19Hammond 2010.
20Kouki, this volume pp. 205206.
21 Compare the argument of Sallares 2007, 19, based on Speranza et al. 2002, that a decline
in solar activity leading to colder and more humid conditions c. 850bc may well have been
the critical factor underlying the simultaneous development of Iron Age cultures around the
Mediterranean, which is otherwise difficult to explain, an assertion, this last, which seems
crucially to under-estimate Mediterranean connectivity.
22 A historian is not allowed to say that a climatic event marks the beginning of the Greek
Renaissance (Manning, this volume, p. 132).

factor, in the decline and re-organization of the later Roman world,23 but
the nature, scale and chronology of the climate changes themselves are
still most unclear. What difference does it make if average temperatures
change by one degree or by several? How much less precipitation means
difficultiesor disaster? We need more data, obviously, but, more impor-
tantly still, we need better thinking about the possible range of differences
between local climates within the same region and within the Roman
Empire (Mediterranean and otherwise) as a whole; and we also need bet-
ter thinking about the likely effects of climatic change on human, or rather
Roman, behaviour.
Many questions about the environmental history of antiquity seem ripe
for scientist-humanist collaboration.24 Scientists sometimes frame the sub-
ject simply as the effect of mankind and climate on landscapes,25 but it is the
vital interactions that demand attention. What would grow where? Grains
(not only rye) and olive trees may raise the largest problems here, but there
is also much to say about other fruits and vegetables,26 and about the timber-
productivity of various areas from Spain to Egypt. What levels of population,
urban and rural, could the resources of various areas support? The Mediter-
ranean has many desert areas and marshes and mountainshow many of
them were productive? And the sea itself has a multifaceted history. One
scholar has it the Mediterranean is relatively poor in fish27should we
think that this was so in antiquity, or should we allow that many coastal
populations depended heavily on harvesting the sea?28 Then there is the vast
array of the mineral resources, widely studied in the last half-century in par-
ticular, but in the case of metals at least still needing Mediterranean-wide
studies.
What effects, direct and indirect, did humans in turn have on natu-
ral resources, in particular on woodlands, soil, fresh water and wild life?
Sub-topics here include wood fuel, deforestation and erosion, fallow fields,
drainage and irrigation projects, hunting, and the consequences of tax
23 Below p. 158.
24 See also the list of questions in Harris 2005b, 1220, and the bibliography mentioned
there, which I do not attempt to bring up to date in this introduction.
25 E.g. Sadori and Giardini 2008, 229. An obvious difference is that scientists generally
want to deal with longer periods, in this case since the mid-Holocene, i.e. the last 4500 years,
whereas few Mediterranean (as distinct from Ancient Near Eastern) historians want to look
back much beyond 1000 bc.
26 See Sallares 2007, 29, for some brief remarks about the diffusion of fruit trees.
27 Sallares 15.
28 Cf. Horden and Purcell 2000, 576577. See now Boardman 2011.
8 w.v. harris
regimes.29 How large a share of ancient economies was pastoral, and how
did pastoralism affect landscapes? The list can go on: one of the most
contested topics in ancient history at present is osteological: the relation-
ship between surviving human bones and stature, and more generally well-
being.30 This issue leads to the wider question of disease patterns. Hence we
need, among others, agronomists, food scientists, geologists, botanists, met-
allurgists, hydraulic engineers, and physiologists.
Arching over all these questions is the perennial problem of general-
ization, already mentioned in the context of climate. Horden and Purcell
attempted to circumvent this problem by concentrating on four micro-
regions. In retrospect, I believe that this was in essence a good idea but that
we need two or three times as many micro-regions, at least. Many scholars
have devoted themselves to studying Greek regions and Roman provinces
over the last forty years, but unfortunately few of them have pursued envi-
ronmental questions in detail if at all (there have been some splendid excep-
tions).
My contribution to the body of this book is an account of deforestation
that attempts to combine historical and scientific information and comes to
a mixed conclusionquite a bit of deforestation, but not a Mediterranean-
wide crisis.
We then proceed to local studies, that is in a sense to micro-regions.
Paula Koukis study takes us to Petra. Her highly suggestive conclusion
is, in brief, that the patterns of settlement there do not conform to what
the climate data might lead one to expect. When precipitation somewhat
increased in the second and third centuries ad, rural settlement thinned out;
a drying climate in the sixth century was accompanied by the enlargement
of settlements. She considers various possible explanations, concluding that
the notion of a favourable climate is simplistic even in a region that was so
heavily influenced by a single climatic variable, rainfall. In particular, much
will always depend on the plants and technologies available, and indeed on
local traditions of land use.
Our last two studies take us to Italy. The approach of Elda Russo Ermolli
and her collaborators is geoarchaeological. They found that both at Neapo-
lis and Velia it was probably a combination of climatic events and human
neglect that led to sedimentation and urban degradation in and after the
third century ad. In both places the coastline moved forward more quickly
29 On the latter see especially Purcell 2005.

30 See among others Steckel 2009, Wheeler 2012, Pitts and Griffin 2012.
in the third century. The old port of Neapolis ceased to function, and flood-
ing rendered part of the site of Velia uninhabitable. Our authors conclude
that the effects of particular land use conditions were enhanced by extreme
climatic events, creating episodes of severe erosion.31
Roman officials intervened from early times to influence land-use (by tak-
ing control of all coastal woodlands: Cicero, De Republica 2.58), and perhaps
even earlier to exercise some control over water resources. Keenan-Jones
paper considers their ambitions and achievements in the latter department,
concentrating on central and south-central Italy in early imperial times. His
first case is the Aqua Augusta in Campania, a longer construction, taken as a
whole, than any of the aqueducts of the capital itself. Whoever diverted the
water from the springs at the head of the River Sabato (see the map on p. 241)
to the towns to the north and west of Naples as far as Misenumit was prob-
ably Augustus who was responsible for most of itdid a great favour for
their permanent and temporary residents, including of course members of
the imperial household and of the senatorial-equestrian elite. In so doing,
the person or persons responsible also did a disfavour to the inhabitants
of the basin of the Sabato itself. It would be worth speculating further how
grave this damage is likely to have been, and whether the decision is likely to
have been an entirely political and social one. For Frederiksen, the feeling
behind [Augustus] massive corrections of nature in Campania is not in the
least utilitarian; rather it was a matter of spectacular ostentation.32 Since
the aqueduct served the interests of productive towns such as Puteoli, how-
ever, its net economic effect may well have been positive. And we have been
reminded by the Hadrianic Bronze of Agn that Roman authorities often
took thought for utilitarian irrigation schemes as well as spectacular aque-
ducts.33
Keenan-Joness second case concerns the aborted project of 15 ad to make
the waters of the River Clanis flow northwards into the Arno instead of
southwards into the Tiber, to prevent the Tiber from flooding again. It could
probably have been done (since it was in fact done by 1700), but as why it
was not, we are at the mercy of Tacitus. The most interesting aspect of the
matter, it seems to me, is the inability of Romes famed hydraulic engineers
to find a solution that would have prevented serious floods in the capital.
31 It might be tempting to associate this erosion with local deforestation, but the pollen
data from the port sediments at Naples, which show an increase in cabbage cultivation during
the third century (Ermolli et al., this volume, fig. 4) and the spread of wild vegetation, seem
to exclude this.
32 Frederiksen 1984, 334.
33 Beltrn Lloris 2006, with bibliography on other cases (166167, 192193).
10 w.v. harris
These case studies, together with those incorporated in the papers by

McCormick and Veal, have shown how much hard work is necessary to
produce results in this field. All the more reason for scholars in diverse fields
to attempt to collaborate. But an even more pressing conclusion is that all
of us who are concerned with such questions about the past as these need
to debate our objectives and consider carefully which questions most need
answering.
PART ONE
FRAMEWORKS
ENERGY CONSUMPTION IN THE ROMAN WORLD*
Paolo Malanima
Economic development has been supported, over the last two centuries, by
a technical revolution in the use of power and energy. The introduction of
modern machines, able to deliver huge quantities of work per unit of time
on the one hand, and the availability of cheap fossil energy sources on the
other, have enormously increased productive capacity. Both changes were
the necessary although not sufficient conditions for the notable discontinu-
ity in the economic history of the human populations and were the main
determinants of a huge increase of output. The scarce availability both of
mechanical power and energy set a limit to the growth potential of pre-
vious agricultural economies from the 5th millennium bc until the start
of modern growth two centuries ago, and was the direct determinant of
phases of decline or collapse. We cannot but agree with the view presented
by E.A. Wrigley on pre-modern agricultural or organic societies. His opin-
ion is that societies before the Industrial Revolution were dependent on the
annual cycle of plant photosynthesis for both heat and mechanical energy.
The quantity of energy available each year was therefore limited, and eco-
nomic growth was necessarily constrained.1 This was the main reason why
decreasing returns to labour prevailed in past agricultural civilisations, as
the English classical economists maintained.
The topic of energy consumption as a whole has been only marginally
investigated in the case of the Roman world (though there has been some
attention to particular energy sources such as wood). Previous attempts
to quantify energy consumption do not allow one to understand the pro-
cedures followed.2 It is obviously impossible to present definite figures of
energy consumption, since local conditions and the relations between
human beings and the environment differed so much within the Roman
* I thank Elio Lo Cascio for his comments on a previous draft of this paper. I also
thank the participants in the conference Growth and Factors of Growth in the Ancient
Economy, January 2829, 2011, held in Chicago (with the support of the Federal Reserve Bank
of Chicago), and particularly Alain Bresson and Joel Mokyr, for their comments.
1 Wrigley 2013, 1. See also Wrigley 2010 on the same topic.
2 See the Appendix.
14 paolo malanima
Empire. It is possible, however, to present plausible data and plausible con-

fidence intervals around the figures. This is a first step towards a comparison
of energy consumption within past societies and between past societies and
the present world.
The purpose of the present work is to focus on energy consumption in the
early Roman Empire; and, in particular, to identify the energy sources ( 1),
to quantify their exploitation (23), and their constraints to the growth
potential (45). The last section (6) will be devoted to the dynamics of the
ancient energy systems, that is the innovations in the technical exploitation
of energy and its availability. The Appendix will present the procedure
followed in the quantification of energy consumption in the Roman Empire
and discuss alternative estimates.
1. The Input of Energy
Often it is not completely clear what actually were the sources of energy
in past agrarian civilizations.3 The consequence is that any quantification
becomes imprecise or, indeed, quite impossible. Although certainty is unat-
tainable on the subject, a plausible order of magnitude is not out of reach.4
There were three main inputs of energy in pre-modern agrarian civiliza-
tions from about 5000bc until 1800ad: food, firewood and fodder for working
animals.5
Food has been the primary source of energy since the beginning of the
human species. A second source, firewood, began to be exploited as fuel
between 1,000,000 and 500,000 years ago. From then until the Industrial Rev-
olution it was the main provider of heat.6 The third source, fodder for draft
animals, began to supply mechanical work in the agricultural civilizations
between 5000 and 4000bc, that is since the exploitation of animal power
on a wide scale in agriculture and transportation. These were still the main
3 Here I refer to the energy sources with a cost (often an opportunity cost). Solar light
is important for our survival, but is free and then excluded from our calculations. The same
holds true for the vegetation of a forest, when not exploited by the humans. Water and wind
power, when exploited through mills and sails (expensive to build), is included, while it is
excluded when not exploited for some productive activity. See, however, the Appendix for
more information on the subject.
4 I have discussed this topic in greater depth in Malanima (forthcoming). See the follow-
ing Appendix on the quantification of energy consumption in the early Roman Empire.
5 I have examined the transitions among energy systems in greater depth in Malanima
2010.
6 Perls 1977; Goudsblom 1992.
energy consumption in the roman world 15
energy carriers of ancient Mediterranean civilization. The discovery of fire

on the one hand and the exploitation of draft animals on the other, marked
two main changes in the history of technology. The most recent change has
been the spread of thermal machines over the last two centuries. In the
long period between the first exploitation of animal power in agriculture
and the steam engine, so for almost seven millennia, no radical change, or
macroinvention,7 occurred in the exploitation of energy, although several
minor changes took place.
Food consumption has not changed so very much during the long his-
tory of mankind, at least in term of calories. Even in the case of ancient
Greek and Roman civilizations, we can assume a daily average consump-
tion of 23,000 calories;8 as recent estimates indicate. In particular, the diet
of the Mediterranean region with its high population density was probably
marked by much lower overall meat consumption.9 Pork meat was a promi-
nent food of the urban high-income strata of society, whereas the poorer
ancient Roman population consumed primarily vegetarian food.10 Although
within a wide geographic area such as the Roman Empire differences in diet
were remarkable, the intake of calories was necessarily similar.11
Regional variations in firewood consumption were much wider and de-
pended on two main variables: temperature and industrial demand. In
Mediterranean civilizations the amount of 1kg. of wood (that is about 3,000
kcal.) per head per day can be assumed as the lower margin of a likely range,
given the relatively high temperature in this area of the world. Calculations
of industrial consumption by metallurgy and other industries (such as pot-
tery, glass and tile production) and services (such as baths) suggest that
another half kg. could be added to this daily amount, at least in regions
with widespread industrial activity. This half kg. more is, however, a rela-
tively high estimate, based on what we know on early Modern Europe.12
For the early Roman Empire only rough estimates on wood consumption
by metallurgy are possible.13 Differences in firewood consumption certainly
existed within the Roman world and derived from the regional differences
7 I use here the word macroinvention following Mokyr 1990.

8 Here I use the terms of kilocalorie (kcal.) or calorie as synonyms, although they are not.
Actually, a kilocalorie (the correct unit of measure when we speak of food or heat) is 1,000
calories.
9 Koepke and Baten 2008, 132.
11 See Jongman 2007b.
12 Kander et al. 2013.
13 See the Appendix.
16 paolo malanima
in temperature and industrial development. A range between 1 and 2 kg.,

that is between 3,000 and 6,000 calories per head per day, seems plausi-
ble.14 According to a calculation of biofuels consumption on a world scale
about 1850, that is when wood was still the main fuel, the per capita aver-
age was 2.3kg. and this average was far lower in the South.15 When taking
into account the high temperatures in the Southern Mediterranean and the
existence of regions with poor industrial activity, a lower estimate of fire-
wood consumption of about 3,000 kcal. per head per day, that is 1 kg., seems
plausible for the Roman Empire. A consumption of 6,000 kcal, equivalent
to 2kg. of wood, could however have been reached in cold regions, in the
mountains, or in areas with relatively high industrial activity.16
As to the contribution by draft animals to the energy balance, an estimate
can be based on the ratio between their consumption of fodder (expressed
in some energy measure) and population. We follow, in this case, the same
procedure we use today to establish the average consumption of oil in a
country: that is, dividing the oil consumed among the population. The only
difference being that in pre-modern agrarian civilizations, we are mainly
dealing with biological converters and that their fuel is food intake. From
the available information on the size of ancient working animals17 and the
draft animals-population ratio,18 we then estimate how much energy was
consumed per head dividing the calories of fodder intake by the population.
The range of a plausible consumption is 1,0002,000 kcal. per head per day.
The only energy carriers not provided by the land through photosynthe-
sis in ancient agricultural civilizations were wind, used to drive sailing ships,
and water, exploited for mills as from the 3rd century bc.19 An estimate of
the consumption of the energy of wind and water is difficult.20 We know,
however, for the early Modern Age, that their contribution to the energy bal-
ance hardly represented more than 1 percent of the total energy consumed.
It seems plausible to assume that watermills and sailing ships were not more
numerous in the Roman Empire than in medieval and early modern Europe.
14
The article by Harris 2011a is important for the quantification of firewood consumption.
15
Fernandes et al. 2007 (see the auxiliary material for the article in http://onlinelibrary
.wiley.com/doi/10.1029/2006GB002836/suppinfo). The consumption of biofuels in the Medi-
terranean regions was lower than the average.
16 See the lower energy consumption proposed by Smil 2010, reported in the Appendix to
this paper.
17 On the topic see in particular Kron 2000, 2002, and 2004. See also Ward-Perkins 2005,
Ch. VII and Fig. 7.3.

18 This ratio is hard to establish for ancient economies. See, however, the Appendix.
19 Wilson 2002a and 2008b and Lo Cascio and Malanima 2008.
20 But see the Appendix.
In mere quantitative terms, the role of wind and water in pre-modern agrar-
ian societies was negligible, although they were very important from the
technological viewpoint. Actually, sailing ships and watermills were the
only engines whose mechanical work did not derive from the metabolism
of food.21 Together these engines provided 100 percent of the mechanical
energy by non-biological converters.
2. A Quantification
Table 1 presents a likely consumption range for the ancient Mediterranean

in the age of the early Roman Empire, that is the 1st century and the first half
of the 2nd, up until the Antonine Plague. As we see, energy consumption
is comprised between 6,000 and 11,000 kcal. per capita per day (or 9.218.4
Gigajoules per year). We see also that half of consumption consisted of food
for humans and draft animals, the other half of firewood.
Table 1. Energy consumption in the early Roman Empire (in Gj. per capita per year
and kcal. per capita per day).
Gj/year Kcal/day %
Sources of energy Min. Max. Min. Max. Min. Max.
Food for humans 3.1 4.6 2,000 3,000 33 27
Fuel 4.6 9.2 3,000 6,000 50 55
Fodder for animals 1.5 3.0 1,000 2,000 17 18
Total 9.2 16.8 6,000 11,000 100 100
Sources: see text and Appendix.
Today World energy consumption is 50,000 kcal. per capita per day or 76.5
gigajoules (Gj.) per year. In Europe it is notably higher: 100,000 kcal. per day
(153 Gj. per year). At the beginning of modern growth, in the early decades
of the 19th century, World average consumption per capita was 710,000
kcal. per day (1015 Gj. per year) and the European 15,000 kcal. per day
(23 Gj. per year).22 Around 1850, consumption per head of the three main
sources of energy (food, firewood and fodder) in Northern Mediterranean
21 Technical change in maritime technology was continuous and certainly contributed to
enhance the exploitation of wind power, although, in mere quantitative terms, the energy
consumed by sailing ships remained modest. See now, on changes in maritime technology,
Harris and Iara 2011.
22 Malanima 1996 and 2010.
18 paolo malanima
countries (Portugal, Spain, France, Italy) ranged between 11,500 and 13,500
kcal. per day.23 A Mediterranean average including Northern Africa and
the Near East (for which we have no data until 1970) would certainly be
lower.24 Thus, a plausible result is that per capita energy consumption in
the ancient Roman world was 56 times less than the World average in
2000 and 10 times less than the European average at the same date. It was
also a little lower than that of the Northern Mediterranean countries at the
beginning of industrialisation. The Roman Empire included many Southern
regions, where the consumption of firewood was certainly lower than in the
Mediterranean countries of Europe at the start of industrialisation.
It is hard to specify the impact of the production of energy on the envi-
ronment in the early Roman Empire. If we assume that food production
required half a hectare per capita,25 firewood half a hectare of forest and fod-
der for draft animals another half hectare, then per capita requirement was
1.5 hectares. This estimate is nothing but a plausible average (based mainly
on late medieval-early modern European examples, where the productivity
of fields, meadows and forests was quite similar to that in Roman antiq-
uity).
In around 165ad, the Roman Empire measured 3,800,000 km2.26 Accept-
ing the previous calculations regarding consumption and soil per head, to
provide energy for the 70 million inhabitants living in the Empire 1,050,000
km2 were necessary, which is 2530 percent of the total. If we assume a pop-
ulation of 100 million, plausible as well for the middle of the 2nd century ad,
the need of soil to support energy production becomes 1,500,000 km2, which
is 40 percent of the Empire. If we exclude the mountains (lands more than
600 metres high), which in the Mediterranean regions cover 2025 percent
of the total area and were hard to exploit, the extent of the agrarian soil
in the Roman Empire becomes about 3,000,000 km2. In this case, accord-
ing to the two previous population estimates, the share covered by fields,
exploitable woods and meadows becomes respectively 33 and 50 percent of
the total area. These shares naturally rise if we subtract from the total extent
not only the mountains, but also hilly lands hard to cultivate, marshes, lakes
and urban areas.
23
Kander et al. 2013.
24
For these countries the series elaborated by IEA (International Energy Agency) start
only from the 1970s.
25 Fallow land is not included.
26 I take both the extent of the Empire and the inhabitants from Scheidel 2007, 48.
3. Efficiency and Energy Intensity
Only a part of energy input is actually transformed into useful energy (or
energy services, that is mechanical work, light and useful heat). How great
this share is depends on the efficiency of the converters of energy, that is
labour (L) and capital goods (K ). The thermodynamic efficiency () of the
system of energy can be represented through the following ratio between
the energy services (Eu) and the total input of energy (Ei):
=
Eu
Ei
Today, in our developed economies, this ratio is about 0.35; that is 35 percent
of the input of energy becomes actual mechanical work, light or useful heat.
In past agricultural civilizations, the efficiency was much lower. A plausi-
ble calculation is easier for the past, when biological converters prevailed,
than for the present. Today, in fact, the variety of machines, with diverse
yields, make any estimate hard. The ratio between useful mechanical work
and input of energy into biological converters, such as humans and working
animals, is around 1520 percent.27 Part of the intake of energy in the form of
food is not digested and is expelled as waste, whilst the main part is utilized
as metabolic energy in order to repair the cells, digest and preserve body
heat. A human being or animal consumes even when inactive. The use of
firewood is even less efficient. The greater part of the heat is dispersed with-
out any benefit for those who burn the wood. Its yield is about 510 percent.
Overall, the efficiency of a vegetable energy system based on biological con-
verters, such as that of ancient civilizations, was around 15 per cent at the
most: that is 1,0001,500 kcal. were transformed into useful mechanical work
or heat; the rest was lost. Thermal machines are much more efficient than
biological converters such as animals and humans.
Another measure of efficiency in the use of energy is the ratio between
the energy input and output, that is GDP. It represents the energy intensity,
or the quantity of energy we need to produce a unit of output (Y ):
i=
Ei
Y
27 See the useful Herman 2007.

20 paolo malanima
This ratio depends on the efficiency of the converters, but, contrary to

the previous ratio, it also depends on the structure of production, that
is the relative importance of the different sectors and subsectors within
the economy. Some sectors (e.g. industry and especially heavy industry)
consume much more energy per unit of output than others (e.g. some
services). If there is a change in the relative importance of any specific
sector, energy intensity changes as well, even without any change in the
thermodynamic efficiency of the converters. It is apparent that the impact
of energy use on the environment depends both on the amount of energy
exploitation and on energy intensity; higher intensity implying a higher
impact on the environment. In past agricultural civilizations, for any unit
of GDP (e.g. 1 dollar), the expense of energy was higher than today. Around
2000, in Western Europe, energy intensity was 78 Megajoules per dollar.28
In past agrarian economies it was at least twice as much, since mechanical
converters of energy are more efficient than biological converters. In 1800
Western Europe, that is before the start of industrialization, it was 1214
Megajoules per dollar. Assuming that in the early Roman Empire energy
intensity was the same as in pre-modern European societies, the level of per
capita GDP would be about 1,000 dollars (1990 intern. $ Purchasing Parity
Power).29
4. The Energy Constraints
Vegetable energy carriers, such as those exploited in past pre-modern civ-

ilizations, are reproducible. The suns energy enables a continuous flow of
exploitable phytomass and the circulation of water and wind. Although the
availability of these carriers was and is endless,30 and the energy system
based on them was and is sustainable, their increase was hard and time-
consuming. A large part of working time in pre-modern economies was
aimed at providing energy. All in all, the expense31 for energy (food, firewood
and fodder) could represent 6070 percent of the average income. In pre-
modern economies consumption represented, at least, 80 percent of GDP.32
28International 1990 Geary-Khamis dollars Purchasing Parity Power.

29See on the topic Lo Cascio and Malanima 2009; forthcoming.
30 Actually, it is not endless, but the Suns light will still reach the Earth for 5 billion years.
31 Including the opportunity cost when a source of energy is provided directly by the
consumer himself.
32 Malanima 2009, chap. VII.
Although this 80 percent was not devoted completely to providing energy,

the expense for food and firewood was remarkable.
Since all sources of energy came from the soil and soil is not endless, the
consequence during epochs of demographic rise was a fall in soil per worker
and then decreasing returns to labour. The main change taking place from
the start of modern growth has been the elimination of the dependence
of the energy system on the soils constraint. When demand increases, it
is much easier to provide coal, oil or natural gas, than the vegetable carri-
ers utilized in past agrarian economies. Since in pre-modern organic veg-
etable energy systems, the transformation of the Suns radiation by plants
into phytomass, thanks to photosynthesis, was central and climatic condi-
tions can heavily influence the output of energy, climatic phases marked
the past history of mankind. Short-term deviations from the average tem-
perature or precipitations resulted in dramatic increases or falls in energy
availability: the well-known years of plenty and the frequent famines of the
agricultural economies. Long-run changes were much less felt or were even
unnoticed, although they influenced agricultural production, thus the over-
all availability of energy, and, consequently, total output and population
trends.
The second important constraint of all pre-modern energy systems was
the low power of the converters, which resulted in a low working capacity
per unit of time. The high standard of living of modern societies is the result
of the higher output per unit of time or higher labour productivity. The
power of a man in everyday work is the same as a 40-watt lamp, or 0.050.07
Horse Power (HP). The power of a horse is 1520 times higher. In pre-modern
civilizations, the most powerful engines were watermills, whose power was
about 3 HP, and sailing ships, which could even reach 50 HP.33 To clarify
this central point about the differences between past and modern energy
systems, we must remember that the power of an average car (80 kilowatts)
is equal to the power of 2,000 people and that the power of a big generating
electric station (800 megawatts) is the same as that of 20 million people. The
electric power of a medium sized nation such as Italy in 2000 equals 80,000
megawatts, which is the same power as that of 2 billion people. Today, a
nuclear plant or a nuclear bomb can concentrate millions of HP, or the work
of many generations of humans and draft animals, into a small space and a
fraction of time.
33 I neglect here the employment of power for military purposes. A catapult was an
ingenious concentration of power.

22 paolo malanima
While the adoption of new energy carriers in the past two centuries has
greatly expanded the quantity of energy at our disposal, an equally key
development has been new technology (machinery) able to concentrate
large amounts of work in particular locations in order to carry out specific
tasks. This concentration of work allows humans to accomplish tasks that
were barely imaginable just a few lifetimes ago. It was the first step toward
a new control of the natural forces at a level inconceivable in past agrarian
civilizations.
5. Innovations
The progress of technology in the ancient Mediterranean world did not

reveal interruptions or declines:34 the use of machines was more widespread
in ancient Greece and Rome, together with ancient China, than in any other
civilization until certainly the 12th or perhaps the 14th century A.D. in West-
ern Europe.35 On the other hand, looking at the problem of technical inno-
vation from the viewpoint of energy, Roman technology consisted primarily,
as J-P. Vernant wrote, in the application of the human and animal force
through a variety of tools, and not in the utilisation of the forces of nature
through the use of machines.36 The introduction of new tools, that is micro-
inventions, was continuous. In a sense this flow of innovations made human
work more efficient, although this increase in efficiency, from the specific
viewpoint of energy and power, was modest indeed.
As suggested by A. Bresson, in the 1st century ad,37 Heros work demon-
strates the knowledge of all the main elements for constructing a steam
engine, such as the conversion from rotatory to alternating movement, the
cylinder and piston, non-return valves and gearings: the main technical
elements embodied in the Newcomen engine were, if not in function at
least well known in the Hellenistic age.38 We can wonder, however, how
widespread this knowledge actually was. With the exception of Heros work,
no other mention of the use of steam is available in ancient literary texts or
archaeological remains.
We know that in England coal began to be used on a wide scale from
the 1st century ad both for domestic usage and for the melting of metals.
34 Greene 2000; Schneider 2007.

35 Wilson 2008b, 362.
36 Vernant 1957, 207.
37 Described in Pneumatica 2.11.
38 Bresson 2006, 72.
Fig. 1. Dated remains of coal in England 1500ad (% of the total dated remains every
50 years). Source: based on data in Smith 1997.
Coal has been recovered from 70 archaeological sites in England and Wales.
Its chemical analysis has allowed these remains to be dated (Figure 1).39
Although we cannot quantify the level of consumption, we can specify the
chronology of its exploitation. When the Romans conquered England, coal
was already exploited. Its utilization spread and attained a maximum level
from the 2nd until the 4th century. At least until the 5th century ad, coal
continued to be used on a wide scale. Later it almost disappeared.
Coal, however, is very unevenly distributed across the globe, and, apart
from Australia, is almost entirely found in a few parts of the Northern
hemisphere, that is, North America, North-Western and Eastern Europe,
Russia and China. The centres of ancient civilizations and especially the
Mediterranean regions are not comprised in the geography of coal. The high
price of firewood on the one hand and the lack of coal on the other did
not allow the transition towards a new energy system in a Mediterranean
civilization.40
39 The decline of the curve in Figure 2 coincides with economic decline in Britain. See the
trend of the British economy described by Ward-Perkins 2005, Ch.V.

40 Bresson 2006, 77.
24 paolo malanima
6. An Energy Crisis?
It is still hard to quantify the rise in population during the millennium

spanned by ancient Mediterranean civilizations. While historians do not
agree on the figures, they do agree, on the trend of population. In 800 bc,
some 20 million people lived around the Mediterranean Sea, whereas in
150ad the population of the Roman Empire numbered 70 million,41 although
the estimate of 100 million could be equally plausible, given the uncertainty
of any estimate for that period. Such a level of population was again attained
by the European continent (without Russia), only in the early modern cen-
turies. Although a calculation of the carrying capacity of the Mediterranean
world is risky, the estimates proposed above regarding the extent of land
necessary to support the population in energy sources do suggest that the
rising population put pressure on resources. Data on decreasing returns to
labour are, however, scanty and uncertain.
It has been suggested that body size diminished in Western Europe from
150ad, after a period of rise.42 On the topic, however, there is no certainty at
all. Koepke and Baten write that during Roman times we have more or less
stagnating heights.43 If stature actually diminished, probably it diminished
later in Central and Northern Europe (e.g. Germany) than in the Mediter-
ranean regions.44
A wider knowledge begins to be available on climate and we can start to
speculate on the possible influence of climatic changes on the availability
of energy sources. On this topic as well, the evidence is still contradictory,
however.
For a long time the rising pressure of population was supported by rising
temperatures in the Mediterranean and the whole of the Northern hemi-
sphere, during the Ancient Climatic Optimum.45 Historians agree on the exis-
tence of a Roman Warm Period.46 Research on ice carrots from Greenland
ice core and the ratio of two oxygen isotopes (18O/16O) provides a record
of ancient water temperature and then climatic oscillations. On this basis
changes in temperature have been reconstructed over several million years.
Annual changes from the 1st century bc are represented in Figure 2.
41 Scheidel 2007, 47.

42 This is the opinion expressed by Jongman 2007a, based on data collection by Geertje
M. Klein Goldewijk. See also Kron 2005 and 2008.
44 Koepke n.d.
45 Haas 2006, 147150.
46 Sallares 2007, 19. See also the long-term view in Blender et al. 2006.
Fig. 2. Oxygen isotopes in the ice carrot GISP2 (Greenland glacier ice core) 60bc
350ad. Source: Rossignol 2012, 97.
We can see that the two centuries bc were favourable from a climatic
viewpoint. Temperatures were high during that period and remained so
until the middle of the 2nd century ad. Some historians suggest that, after
150ad temperatures diminished remarkably, as the curve in Figure 2 shows.
Very little, however, is known about the evolution of climate in the Mediter-
ranean.
Rossignol has claimed that a remarkable worsening of the climatic condi-
tions occurred from about 150ad. The middle of the 2nd century witnesses
the end of a warm period during which the ratio of the oxygen isotopes
had attained levels which would only be reached again in the 20th cen-
tury.47 The presence in the ice carrots of sulphuric acid, dated between 153
and 162, reveals the influence of volcanic eruptions on the fall in tempera-
tures.48 Higher temperatures mean that the season for harvesting vegetables
is longer; that land can be cultivated at higher altitudes and further North.49
Soil per worker rises when temperatures are milder.
The opinion expressed by S.W. Manning is more cautious: A range of
records indicate that a stable and reasonably positive (warm, and in a num-
ber of areas or cases also mainly moist) climate regime was in place for the
period from about the 2nd century bc through the 2nd century ad. This
47 Rossignol 2012, 96. See also Manning this volume, Fig. 8.

48 Rossignol and Durost 2007.
49 See also Weinstein 2009.
26 paolo malanima
Fig. 3. Intensity of precipitations between 400bc and ad400 (and range of error)
(mm. per year). Source: Bntgen et al. 2011, 581.
unusual status, reducing some of the typical variability, uncertainty and

risks of the Mediterranean climate regime for farming, would have been con-
ducive to the growth of the Roman world. It was also an especially favourable
time (warm, moist) for both agricultural and demographic expansion in
central and northern Europe.50 According to Manning, the stability of the
previous several centuries ended; agricultural uncertainty and bad years
would have increased. It is hard, however, to specify the turning point
towards decreasing temperatures. The 2nd century does not reveal, in his
opinion, a clear declining trend.
Precipitation has been reconstructed for the region of Israel51 and for
Germany and Switzerland.52 We know that it diminished and the climate
became drier when the temperature was falling (Figure 3). In Central Eu-
rope, precipitation peaked in 100bc, but from then on diminished, reaching
a minimum in ad300 (100 millimetres less than in the second century bc).
The climate became increasingly dry.53 According to Manning, the 2nd to
50 Manning this volume.

51 Orland et al. 2009.
52 Bntgen et al. 2011, 581.
53 Schmidt and Gruhle 2003a and 2003b.
5th or 6th centuries ad seem to be relatively arid in several areas of the

eastern Roman empire, and the indications of less favourable climate condi-
tions further East into central Asia may have been one of the forcings behind
the movements of populations that led to invasions/migrations into the late
Roman world.54
The pressure of population on the energy resources both to provide food
(and then widen the arables) and firewood resulted in a decline of the
forested areas.55 By the end of the Republic, most of the areas of Italy that
were accessible to Rome had lost most of their stands of tall trees, but except
for some metal-working centres, most places had stabilized their fuel sup-
plies. Patches of eroded land continued to multiply, however, all the way
through the high-imperial period of prosperity.56 In Spain, climate dete-
rioration would have hampered vegetation recovery after fire and exacer-
bate[d] human impact (deforestation) in general.57 In such cases, because of
the need to meet the inelastic demand for food, the livestock and meadows
diminish (although for the ancient world nothing certain can be said on the
matter). Intensification occurred in agriculture and convertible husbandry
spread to support the demographic rise at least in Italy.58 For a comparison,
in Europe, between 1500 and 1700, the 40 percent rise in population, from
80 to 120 million,59 resulted in a 20 percent decrease in agricultural product
per capita (that is energy, since the greater part of energy came from the
fields).60
Population pressure on the energy sources diminished certainly after the
Antonine Plague, that spread between 160 and 170ad,61 as archaeological
wood remains from Central Europe seem to suggest (Figure 4).
By themselves, neither population rise nor climatic changes are necessar-
ily connected to phases of economic decline. Their coincidence can, how-
ever, deeply influence the economy and provoke destructuration and finally
collapse.
54 Manning this volume.

55 See, however, the reconstruction by Kaplan et al. 2009. See also Ruddiman and Ellis
2009.
56 Harris 2011a, 139.
57 Kaal et al. 2011, 172.
58 Forni and Marcone 2002.
59 Russia is included in these estimates of population.
60 Kander et al. 2013.
61 See especially Lo Cascio 2012.
28 paolo malanima
Fig. 4. Estimates of forest clearance in Central Europe (Germany, North-Eastern

France) from archaeological wood remains 200bc400ad (decadal data; any point
of the diagram represents the intensity of the felling). Source: Bntgen et al. 2011,
580.
Conclusion
The energy system of ancient Mediterranean civilizations was the same

as that of all agrarian societies. Despite the increase in useful knowledge
and the extensive development of the agrarian energy basis, supported by
a favourable climatic phase, this system was finally unable to support the
increasing needs of the rising population (as always in agrarian civilisa-
tions). If we follow the economic approach by the classical economists,
rephrased by E.A. Wrigley with particular reference to energy, an increasing
pressure on the resources by the rising population would have been followed
by decreasing returns and then diminishing energy availability, after some
centuries of rising population. Data showing a clear economic trend for the
first centuries of the Empire are almost entirely lacking, but an unfavourable
climatic phase, beginning probably, but not certainly, in the second half of
the 2nd century ad, contributed to a decline.
Much later, during the Little Ice Age, in the early modern centuries, the
reaction to a similar crisis was a much wider use of coal.62 This main change
developed in England since the 16th century. Then, in the 18th century, the
steam engine began to interact with the new, rising input of energy. This
62 The topic is discussed in Malanima 2010 and 2011.

interaction initially began to involve the Central and Northern European

regions and subsequently also the regions far from the centre of the great
change then in progress. The combination of changes in power and energy
was the basis of modern growth. Just as in many other pre-modern societies,
the structure of the energy system prevented ancient Mediterranean civi-
lizations from following a similar path. Ancient growth found in its energy
basis a main constraint to its further economic progress.
30 paolo malanima
Appendix
Estimates of Energy Consumption in the Early Roman Empire

A wider analysis of per capita energy consumption in the early Roman
Empire is presented in Malanima (forthcoming). The topic of energy con-
sumption in pre-modern economies is also discussed in Malanima 1996,
2006, 2009, 2010, 2011 (www.paolomalanima.it).
As seen above (12), sources of energy of pre-modern, agricultural econ-
omies are the following:
1. food;
2. fuel (almost always firewood);
3. fodder for working animals;
4. water and wind power.
1. Food
Food consumption has always been the most stable energy carrier ever
exploited since the beginning of the human species. In the following dia-
gram (Figure 5), I report the series presented by Jongman (2007b, 599), on
calorie consumption in present day populations. Taking into account the
age structure in Roman antiquity, with more young people than today, the
range of 23,000 calories seems plausible. Considering yields per hectare,
to cover the needs of a family of 5 people, about 5 hectares were necessary,
including fallow lands. Thus a family needed between 2.53.3 hectares of
cultivated land (excluding fallows): i.e. from half to two-thirds of a hectare
per person (for data on yields, and soil per capita necessary to satisfy food
demand, see Forni and Marcone 2002, on agriculture in Roman Italy).
2. Firewood
As said in 1, firewood consumption depends on temperature and industrial
use. One kg. of wood can be seen as the lowest possible level of consumption
(as also stated by Harris 2011a; see also data in Pireddu 1990, 27). Although
hard to quantify, firewood consumption was low where temperatures were
high and high where temperatures were low (see, for instance, data in Warde
2006, referring to early modern Europe). If, to simplify, we assume that
in a Mediterranean climate, each individual consumed 1 cubic metre of
wood per year, that is 625kg., including industrial uses as fuel (1.7 kg. per
day), this amount of wood could be provided by the yearly growth of half
a hectare of forest (Chierici 1911, 232233). Assuming that the population
of the Roman Empire in around 165ad was 70 million inhabitants for an
Fig. 5. Food consumption by modern populations according to age (kcal.) Source:

Jongman 2007b, 599.
area of 3,800,000km2, and that every inhabitant consumed 1 cubic metre

of firewood (including wood from prunings), then the total requirement
was 70 million cubic metres. It could be provided by a wooded area of
350,000km2, or 910 percent of the total inhabited surface of the Empire.
With a population of 100 million inhabitants, the wooded area rises to
500,000km2, or about 13 percent of the total. A city such as Rome, with 1
million inhabitants in the age of Augustus, needed 50 km2 of forest to cover
its needs.
As to industrial consumption, we can only provide some calculations
from what we know about the output of metallurgy. Let us assume that iron
production was between 80,000 and 160,000 tons per year (cf. Harris 2011a)
and, at the lowest, a consumption level of 30kg of firewood (transformed
into charcoal) per kg. of iron (Smil 1994, 144156). Charcoal, known in Egypt
as early as the 3rd millennium bc, was widely used in Greek-Roman antiq-
uity (Wikander 2008, 138). For the production of 80,000 tons of iron, the
quantity of firewood would thus be 2,400,000 tons (converted into charcoal).
In cubic metres, the requirement was 3,840,000 (assuming 625kg. per cubic
metre, and then dividing 2,400 million kg. by 625). With a yearly productiv-
ity of half a cubic metre per hectare of forest, in order to produce 3,840,000
cubic metres, 1,920,000 hectares or 19,200km2 were necessary. Assuming iron
output being twice as high, the need amounts to 38,400 km2 of forest. This
32 paolo malanima
area is only 510 percent of the total forest required by the population for
heating and cooking. As said before (1), other industries (such as pottery,
glass and tile production) and services (such as baths) exploited wood. An
estimate is, in this case, impossible. Fuels different from firewood repre-
sented a negligible share of the total. Thus, our estimate for a Southern,
Mediterranean civilization such as the Roman Empire is between 1 and 2 kg.
of wood, that is 3,0006,000 calories per capita per day.
3. Fodder
The estimate of fodder consumed by draft animals is more complex. From
the viewpoint of energy, an ox or some other working animal is like a
machine. It metabolizes vegetables to accomplish a task. In order to estab-
lish the average consumption in energy sources per head, the input of energy
by a draft animal must be divided by the family members that exploit it.
We know that improved fodder management and nutrition determined a
remarkable increase in the size of animals during Graeco-Roman antiquity.
Ley farming and meadows supplied animals with better fodder than in the
late Middle Ages and early modern times (Kron 2000). Oxen were taller and
heavier than in Medieval and early modern Europe: about 400 kg instead of
2300 (Kron 2002 and 2004).
We can establish a ratio between working animals and population in
ancient Mediterranean civilizations from the technical relationship sug-
gested by ancient agronomists between land and working animals. In the 1st
century bc, Varro recalls the opinions of Cato and Saserna about the need
of a yoke for every 80100 iugera (2025 hectares) (On agriculture 1.2122).
Since a yoke is composed of two oxen, the relationship is therefore a work-
ing animal per 1012.5 hectares. A century later, Columella tells of two yokes
of oxen for a farm of 200 iugera (or 50 hectares) (On agriculture 2.12.17).
Again we find a ratio similar to that suggested by Varro and relatively close
to the animal-land ratio found in early Modern Europe. Since a peasant fam-
ily required a farm of about 35 hectares to support its living (as shown in
1 of this App.), we could divide among the 1015 members of two average
families endowed with a farm of 35 hectares each, the calories from fodder
consumed by oxen (2530,000 kcal per animal per day) and we would obtain
the result of 1,7003,000 kcal. per head. We would have to add to this esti-
mate horses (on which see Vigneron 1968), mules, donkeys and camels, and
we would also have to include urban inhabitants (excluded from the previ-
ous draft animals-peasant families ratio) in the denominator of our ratio. All
things considered, a range of 1,0002,000 calories per day per capita seems
plausible.
4. Wind and Water

The only possibility of estimating the consumption of water and wind power
is to start from power (work done per unit of time -1 second-). In the case of
a large sailing ship, with a carrying capacity of 400 tons, a rare example in
the ancient world, where the majority of sailing ships were below 100 tons
(Greene 1986, 26), a relationship existed between tonnage and power. The
power of such a ship (400 tons) was about 50 HP (Malanima 2006). Assuming
(absurdly!) that this power was exploited fully for 24 hours and 365 days per
year, energy per year would be 438,000 HPh (Horse Power hour is a measure
of energy), that is, 770,000 kcal. per day. We would now need a plausible
ratio between ships and boats on one hand and population on the other.
Even assuming the ratio existing in early modern Europe to be correct, the
result would be less than 1 percent of the entire energy consumption per
capita.
The watermill was the most powerful engine existing on land. Generally
its power did not exceed 23 HP, although examples of big mills (Munro
2003) or the combination of several mills in powerful sets of engines are not
lacking (Brun 2006; Wikander 1979, 2000 and 2008). The mechanical work
produced by a watermill endowed with the power of 2 HP is about 64,749
kj. (15,000 kcal.) per day, and since a man consumes 2,5503,000 kcal. per
day as food, consumption of gravitational energy by a watermill is 6 times
the energy consumption of food per capita. In late medieval and early mod-
ern Europe, a ratio existed between watermills and population: 1 watermill
every 250 people. Otherwise stated, any small village of 50 families had its
own mill (on the topic Makkai 1981 is important). If we divide a mills energy
consumption by 250, the result is 60 kcal. Certainly, the use of mechanical
energy to grind cereals was a remarkable achievement of ancient civiliza-
tion. Its contribution to the energy balance was, however, modest in mere
quantitative terms. Although we do not know the inhabitant-watermill ratio
in the ancient world, and even allowing for the existence of the same late
medieval ratio, which seems too high for antiquity, as early as the first
centuries of the Roman Empire, the result is that the contribution to the
energy balance was indeed modest (Reynolds 1983; Lo Cascio and Malan-
ima 2008).
Let us consider that previous calculations on mills and ships assume
full-time work (24 hours per day), which is implausible. Contributions to
the energy balance assuming more realistic working time imply a reduction
of the available energy per head.
34 paolo malanima
The Estimates by Ian Morris

Different estimates of energy consumption have been provided by Ian Mor-
ris (2010a and 2010b). According to Morris (2010a, 28), the sources to be taken
into account for a calculation of energy consumption (including the ones
used in modern economies) are the following:
Food (whether consumed directly, given to animals that provide labour, or
given to animals that are subsequently eaten);
Fuel (whether for cooking, heating, cooling, firing kilns and furnaces, or
powering machines, and including wind and waterpower as well as wood,
coal, oil, gas, and nuclear power);
Raw materials (whether for construction, metalwork, pot making, clothing or
any other purpose).
We can see that there is a similarity between this list and the sources taken
into account in this paper. However: 1. I do not include feed given to ani-
mals that are subsequently eaten, since it is already included in the 23,000
kcal. of food for humans (and it would be a duplication of the same source
in our calculations). These animals certainly put a high pressure on carrying
capacity. If agricultural produce is not consumed directly by the population,
but consumed by animals which are then eaten by the humans, the pres-
sure on land is higher. In any case those animals are only used as food and
are not exploited in agriculture or transport. They are part of human food;
which enters the energy balance. As a consequence, I include only feed for
working animals; 2. it is not clear how Morris computes the contribution by
wind and water power; 3. raw materials cannot be considered as energy car-
riers and are not included in my estimates (or in those of the International
Energy Agency or the US Energy Information Administration). Morris fol-
lows, however, Cook 1971, who includes vegetable fiber, which brings solar
energy into the economy through photosynthesis (134). See also Cook 1976,
51 and 135. Raw materials, however, are not used as providers of energy. Fire-
wood, is also generated by photosynthesis, hence when used as an energy
carrier I include it in my calculations. When timber is used as raw material
for construction, it is not included, despite being produced by photosynthe-
sis. It is not an energy carrier in this case.
The results by Morris are quite different from those presented in the
previous pages. In the following Table 2 some data are reported from two
series presented by Morris (2010b, 628).
Table 2. Energy consumption in advanced regions of the West and East according
to I. Morris. 8000bc2000ad (thousands of kcal. per capita per day).
West East
(000) (000)
2000 230 104
1900 92 49
1800 38 36
1700 32 33
1500 27 30
1000 26 29.5
200ad 30 26
1ad 31 27
200bc 27 24
8000 6 5
Source: Morris 2010b, 628.
Around 2000, the average world energy consumption was 50,000 calories per
day. According to Morris estimate, in 200bc some parts of the World already
exceeded this level even without fossil fuels.
In both works by Morris (2010a and 2010b), previous data (reported in
Table 2) for the year 2000 actually refer to the most advanced countries
in the West (USA) and in the East (Japan). In addition, data for previous
years refer to the most developed core within the West (Morris 2010b, 42),
whose borders, however, are not clearly defined. In any case, Morris results
are too high. In 1800, according to recent research, energy consumption in
Western Europe (a highly developed part of the globe) was not 38,000 kcal.
(as maintained by Morris), but about 15,000 (average for Sweden, Norway,
The Netherlands, Germany, France, Spain, Portugal and Italy) (Kander et al.
2013 and data published in Gales et al. 2007). In 1900, for the same countries
of Western Europe, the average was 41,500 kcal. per day per capita, and not
92,000 (as in the previous Table 2). In England it was 95,000. Morris estimate
for 1900 is only plausible if by West we refer only to England. As we see, data
for the Roman Empire are also quite different from ours. Even if we take the
most advanced part of the Roman Empire, Italy, in 1861, that is, the year of
the Unification of the country, energy consumption per capita was 1112,000
calories (Malanima 2006), less than half the estimate proposed by Morris for
the West (31,000) in 1ad.
Energy intensity represents the ratio between energy consumption and
GDP. In Western Europe from 18001820 it was 1215 Megajoules per 1 dollar
(1990 international Gery-Khamis dollars), when per capita GDP was 1,200
36 paolo malanima
dollars (according to the series by Maddison 2007, in 1990 international

Geary-Khamis dollars PPP). If we assume the very high estimate of 1,500
dollars for Roman Italy (taking into account that recent estimates hardly
exceed 1,000 dollars, as shown in Lo Cascio and Malanima 2009 and forth-
coming), the resulting estimate of energy intensity, taking Morris estimate
of 31,000 kcal. per head per day (and then 11,315,000 kcal. per year, or 47,342
Gigajoules), is 32 Mj. per dollar, and thus more than twice that ascertained in
1800 for Western Europe. With a GDP per capita of 1,000 dollars in the early
Roman Empire, the implied energy intensity becomes 47 Mj. per dollar. For
a comparison, in 2000, World energy intensity was 11.5 Mj. per dollar (1990
Geary-Khamis int. dollars) and in Europe it was 5.5 Mj. per dollar.
Vaclav Smil (2010, 107113) proposed estimates of energy consumption in
ancient Rome that are far lower than those by Morris. Here is the comment
by Morris on Smils views: Roman total energy capture would be some-
where between 4,600 and 7,700 kcal/cap/day [according, that is, to Smils
calculations]; if we assume that roughly 2,000 kcal/cap/day of this was food
(which means ignoring the archaeological evidence for relatively high levels
of expensive calories from meat, oil, and wine), that leaves just 2,6005,700
kcal/cap/day to cover all other energy consumption. To justify this estimate,
Smil suggests that Roman fuel use was just 180200 kg. of wood equivalent
per capita per year, or roughly 1,7502,000 kcal/cap/day. Smils estimate
of firewood consumption certainly seems too low. On the whole, however,
Smils estimates are closer to mine than are those by Morris.
FUELLING ANCIENT MEDITERRANEAN CITIES:
A FRAMEWORK FOR CHARCOAL RESEARCH*
Robyn Veal
Introduction
Fuel in the ancient Mediterranean has to date received little detailed anal-
ysis. Humans in the Mediterranean consumed fuel in socio-culturally con-
ditioned ways (i.e. history in the Mediterranean); but that they could con-
sume fuel at all, and which fuels were available in which areas, is very much
a history of topic.1 Quantitative and qualitative studies of the economy
have focused on production and trade of goods and slave labor, but the
fuel economy has been difficult to trace in the historical sources, mentions
being more incidental than material. The most important archaeological
evidence, i.e. that of the archaeological charcoal, is not yet routinely col-
lected by all excavators. This is an omission that begs attention, as ancient
settlements could not function without fuel. The gathering of wood for fuel
occasionally resulted in dramatic changes in the environment when over-
exploitation occurred (for example, on islands), while in other places, more
sustainable practices appear to have occurred. (Wood was not the only fuel
in many parts of the Mediterranean: animal dung and agricultural waste
such as chaff and olive lees were also consumed.) Geology, topography,
and climate determine which trees may grow in a particular location; but
* The author thanks the conference organizer, W.V. Harris for the invitation to speak and
to contribute this paper; and also for his helpful comments upon the draft. All conference
participants are thanked for their enormous warmth and intellectual generosity. Access to at
the time unpublished manuscripts by both W.V. Harris and P. Malanima is also gratefully
acknowledged. I also thank A. Wilson for helpful discussion. A fellowship at the British
School at Rome provided the optimal environment for writing this offering, which is also
based on work completed at the Australian Centre for Microscopy and Microanalysis, and
the Department of Archaeology (University of Sydney).
1 Cf. Horden and Purcell 2000. Fuel as a natural resource neatly straddles the nuanced
distinction Horden and Purcell make between examining the (natural) history of the Medi-
terranean, as opposed to human history conducted in it. We may examine these from pre-
history to the modern period using all data at our disposal for one any period: scientific,
archaeological and historical.
38 robyn veal
politics, land ownership, cultural mores and agricultural practice moder-

ated the physical factors.
This contribution provides a framework for examining ancient Mediter-
ranean cities fuel supplies. Archaeological charcoal is at the heart of this
approach but aspects of the historic sources are also considered and a case
study of Pompeiis fuel economy c. third c. bc to ad 79 is briefly overviewed
in line with the methodology suggested. New scientific techniques beyond
simple charcoal identification as to wood type have started to appear and
are discussed here in terms of their usefulness for examining forest man-
agement and consumption.
Further aims of this contribution are to encourage researchers to collect
charcoal, and to show the detailed ways in which it can now be used to
examine a citys fuel supply. In time, with sufficient further research, it may
be possible to synthesize regional patterns of supply and consumption for
the Mediterranean (and the ancient world as a whole). Indeed the relevance
of studying ancient wood fuel remains appears to have become greater today
as we consider modern problems of climate change, and the potential of
pelletized wood (at perhaps 70% of the calorific value of coal)2 as a part of
our fuel future.
Studying the Fuel Economy:

Modern and Ancient Difficulties
Fuel is a central part of most production processes, and as such forms part
of a city or states, economic consumption. An economys size may be esti-
mated by its Gross Domestic Product, or GDP. Two methods are routinely
used to calculate GDP: i) the income method, or ii) the expenditure/con-
sumption method. For the income (of households) method, GDP is defined
as labor income + capital interest + rent. Under the expenditure/consump-
tion model, GDP is equivalent to consumption + investment + government
spending + net exports. In the ancient world, the latter is often expressed
in terms of grain equivalents.3 These two methods implicitly include some
sort of value for fuel. They are meant to arrive at the same value, and they are
2Wynn 2011. See also Collaborative Partnership on Forests 2011.

3The range of modern scholars calculations of the size of the ancient Roman GDP has
been summarized recently by Lo Cascio and Malanima forthcoming. See also Scheidel et al.
2007.
fuelling ancient mediterranean cities 39
usually expressed on a per capita basis.4 Besides estimating population,

other difficulties in measuring both modern and ancient GDP include the
fact that both methods omit unpriced goods, typically natural resources
such as wood and water (or they may be underpriced). Both methods also
omit non-market activities, for example: barter and villa type non-
monetized activities (operating outside regular markets), patronage, and
euergetism. In ancient studies, a third method for calculating GDP has been
more recently described, the so-called new institutional economics.5 This
method more explicitly addresses the problems of calculating economic val-
ues in a system that is so moderated by social, religious and political mores.
It is not the intention to review ancient economic methodology in detail
here, but merely to work towards understanding the part in the ancient
GDP fuel might have constituted, since, as already intimated, it appears to
have been overlooked. In the twentieth century, data for the United States of
America provide some food for thought. Energy as a proportion of GDP var-
ied between 8% and 14% between 1970 and 1998, dropping back to roughly
9 % in 2006.6 So, in recent history, there has been a more efficient use of
energy per GDP dollar over time. Our modern energy sources are mostly
nuclear and fossil fuel (petroleum) based (i.e. high calorific fuels by weight).
In the ancient world wood and wood converted into charcoal predomi-
nated, although there were many alternatives, such as peat, coal (in limited
amounts), and the agricultural wastes mentioned in the introduction. Of
these, olive lees and pits provide a very high calorific potential fuel, while
peats and chaff provide much lower rates of return by weight.7 Many of these
alternatives were often dried, pressed and even pre-charred prior to use, as
they are even today in some locations. Charring increases calorific, i.e. heat,
potential, which is an indicator of the quality of a fuel.
How can we estimate fuel energy as a percentage of GDP in the ancient
world? In the ancient economy, agricultural activities dominated, but are
rarely thought to have been greatly efficient.8 Manufacturing also operated
4 Here lies one of the first problems of such estimates, which population figures should
be used in ancient calculations? Also see Scheidel and Friesen 2009.

5 Bang 2009.
6 Institute for Energy Research 2010.
7 Modern work on biochar (any kind of charred organic material) is providing data
for comparison with ancient equivalents (Sohi et al. 2009). At Leptiminus (Mediterranean
Africa), amphorae of pre-charred olive waste were found outside a kiln and olive pits present
elsewhere in some areas of the site were interpreted as fuel waste (Smith 2001, 434435).
Smith quotes several other examples of this phenomenon in the region.
8 Cf. Spurr 1986 and White 1970, 4752, for Roman agricultural efficiency.
40 robyn veal
at a basic level compared with the twentieth century, so if fuel is currently

nearly 10% of the equation today, and relative efficiency in the ancient world
is anything to go by, we might conclude that the value of fuel inputs might
be higher than 10%. Perhaps as high as 20%, or more. In assessing the value
fuel has in the modern world we may refer to extensive modern pricing data
that explicitly allow a comparison between the cost of fuel and the cost of
other inputs to the economy. In the ancient world, we have little data on the
actual volumes of fuel consumed, and even less on the prices (and in any
event these will likely have been underestimated, as previously noted).9 We
still lack the data to make more than an educated guess, but nothing will be
lost if we attempt to qualify and quantify the fuel supply.
Sources of Information about Fuel
Most historical information about wood (written, epigraphic, sculptural)

relates to timber for construction: of buildings, ships, and war machines,
to name a few of the largest uses. Meiggs overview, Trees and Timber in
the Ancient Mediterranean World, synthesizes the ancient historic sources,
and the archaeological data available at its publication date, and focuses
substantially on wood use for timber. At that time, charcoal analysis had
been established as a discipline, but was as yet a fairly obscure specializa-
tion. Meiggs discusses fuel consumption mostly in relation to exceptional
cases, for example, Delos, where all wood fuel for temple use was imported
from about the 3rd c. bc.10 The amount of forest consumed proportionately
for fuel (as opposed to timber) is, and was, relatively high. In modern day
developing countries that are still substantially wood dependent (much of
Africa), over 90% of harvested wood is used for energy, while about 10 %
is consumed for timber and other purposes.11 The focus on large timbers in
the ancient world is unsurprising since the evidence for long distance trade
9 Diocletians Edict (with all of its incumbent difficulties, not least of which is its fairly
late date) provides some useful data that tells us of the relative values of timber for building,
as opposed to raw fuel wood, or cuttings (kindling) (Graser 1959). Timber has the highest
value, while kindling is rather highly valued. Relative transport costs are also provided. Of
course it is not valid to assume the relative values of these commodities remained constant
through time. Other small inferences about pricing can be gleaned from the historic sources,
but these do not provide anything more than a broad indication.
10 Meiggs 1982, 441457; see also Reger 1994.
11 Food and Agricultural Organisation 2011. We are not able to calculate this proportion for
the ancient world at this point but the volume of wood used for construction in any current
developing society has rarely been much greater than 20%.
Fig. 1. Factors affecting the wood supply, which underpin the types of archaeological
charcoals found. The factors are named in general terms in the top left-hand part
of the diagram, while at the bottom right, described are the ways in which these
factors may be exemplified.
in this commodity is well documented. The large cedars of Lebanon which

were exported all over the ancient world to build large palaces and temples,
stir the imagination, while the remains of fuel in the archaeological record,
i.e. those small black fragments that often actually fall through an archaeol-
ogists sieve, seem plain and insignificant. However, charcoal remains, and
their analysis, are the keys to unlocking the wood fuel economy.
Factors Affecting the Wood Supply
Physical Factors
Figure 1 depicts a diagrammatic representation of the factors that affect the
wood supply (and consequently the charcoals we might find in an archae-
ological setting). At the broadest level, landscape characteristics (geology/
soils, topography), and climate are the first determinants of which trees
may grow in a location. These are moderated in the longue dure by human
agency, which can improve or damage the broader environment, making it
more, or less, hospitable for trees. Land clearance for agriculture in partic-
ular can lead to soil erosion. Over use of water for irrigation leads to aridi-
fication of an environment ultimately changing the types of trees that can
survive. The interaction between man and the environment, especially in
relation to the wood fuel supply, is both iterative and complex.
42 robyn veal
Woodland Use and Management

Woodland use and management also of course altered the landscape in
various ways. In the historical sources, dramatic examples of state or impe-
rial command for a forest to be cut for one purpose or another12 suggest
whole-scale deforestation might have been occurring in parts of the Roman
period, but except for specific localized instances, the pollen record shows
no large scale deforestation until the Mediaeval period.13 Pollen studies are
our main source of scientific information about forest cover. They con-
tribute to knowledge of the palaeo-environment through evaluation of
archaeological soil samples from excavations (i.e. quite local data), and/or
longer records taken from bogs, ponds, lakes or other damp environments
which provide a more regional signature.14 A renewed interest in pollen the-
ory and method has meant that the reliability of pollen studies for landscape
reconstruction is now being re-evaluated.15 Scale is an issue in examining
pollen evidence. Views of short term historical periods (e.g. the Roman
Imperial period) are difficult to find in long scale pollen studies (e.g. the
Holocenefrom 12,000 years ago until today). Also, we know broadly that
continual reduction of forest cover occurred from the beginning of agricul-
ture, and accelerated with metal smelting and smithing, especially of iron,
but this is not always obvious in a pollen diagram.
The more commonplace management of woodlands belonging to vici
being carefully husbanded for generations rarely attracts comment in the
12 For example the building of the Roman navy to confront the Carthaginians in the First
Punic War would have taken considerable resources (Polybius, Hist. I.20).
13 Deforestation requires definition, see W.V. Harris, this volume, and also Harris 2011a,
108109. Harris moves to a more nuanced view on a regional basis. Grove and Rackham 2001,
Chs 1011 provide an overview of the history of forests in the Mediterranean and the factors
which led to changed forest cover over time. Meiggs 1982, 377, suggests real deforestation
could not have occurred until railways were built in the 19th c. in Italy. This is rather later
than the pollen evidence now suggests. For pollen, for example in south-east Italy, see Russo
Ermolli and di Pasquale 2002. Modern Europe is now afforesting at a slow rate (while Asia
and South America are losing forest cover).
14 Dincauze 2000, 377380.
15 Giesecke et al. 2010 overview research being done to evaluate inter-annual variabil-
ity of pollen fall, the effect of weather, and adequacy of sampling methods, among other
variables. The results suggest that absence of a pollen type from a diagram does not nec-
essarily mean absence in the environment, and concentrations of just one pollen type can
vary wildly, even when multiple samples are carried out in the same area. For an accu-
rate regional (i.e. Mediterranean-wide) view then, we must carry out many pollen studies
over time and space and compare these. The European Pollen Database (available at www
.europeanpollendatabase.net), was established in 2007 just for this purpose but not all stud-
ies completed are yet recorded here.
historical sources, as this type of sustainable management was the norm.

Local communities had to preserve their woodlands for fuel or potentially
perish. Imperial consumption might have been, from time to time, on a large
scale, but as Grove and Rackham point out, forests can grow back, and will
do so providing soils remain, however species patterns and diversity may
change as colonisers (i.e. those tree types which are stronger competitors)
move in. Over-use of forests depends not just on the presence of the nor-
mally suspect heavy fuel use industries such as smelting, but more particu-
larly on a range of factors including the carrying capacity of an area, and the
woodland management practices.16 It should be noted however, that decid-
uous forests usually recover relatively spontaneously, while conifer forests
may not recover without planting seed.
In areas of marginal environment, overexploitation has always been a
problem for the survival of broadleaf deciduous trees, which were the most
desirable and among the most commonly used types used as fuel in the
Mediterranean.17 Such marginal areas include those with poor or thin soils,
those with low rainfall, or both conditions. Areas, especially islands, which
were in early history verdant with forest, became marginal or deforested
more quickly if they were commandeered for special purposes (such as
smelting). Some areas were never heavily wooded (much of Greece, for
example), but occasionally scholars assume the lack of woodland means
the environment has been denuded.18 Taken together, the ecological and
socio-political factors, which we may call macro factors, likely influenced
the wood fuel supply (and the resultant archaeological charcoal remains)
in some places more than others. But in considering a strategy for exam-
ining the fuel supply in any one part of the Mediterranean, consideration
of the macro factors helps to frame the way in which the work may begin:
the type(s) of fuel that might be expected, as indicated by the physical envi-
ronment (and thus what reference materials are needed); the historical evi-
dence that exists for woodland use and management (and political control),
and the types of archaeological approach required to find the fuel remains.
Demographics and fuel consumption levels are key to subsequent quantita-
tive modeling.19
16 Grove and Rackham 2001, Ch. 11.

17 The low growing heat tolerant shrubs of the macchia being the other type. We dont
know in what portions (yet) these two basic woodland types provided fuel for the Mediter-
ranean; nor indeed, the proportions used for alternative fuels.
18 Greeces woodland cover, both past and present is discussed by Rackham 1982.
19 In some areas, wood fuel remains are not the only ones we should be seeking, in
particular in Africa.
44 robyn veal
Local Silvicultural Practices

On a more local scale, silvicultural practices further modified woodlands.
The particulars of Greek and Roman forest management strategies in Greece
and Italy can only be broadly inferred from the ancient sources. What
happened in the more distant parts of the Graeco-Roman world is even
more obscure.20
Building and military requirements were the highest priorities, and for
these purposes the state owned or commandeered resources during the
Roman period. The indigenous forerunners to the Romans, and the Romans
themselves, had a sophisticated knowledge of cropping strategies for the
various fuel products required, and of timber performance for various uses.
Coppicing (cutting of standard sized round-wood at regular time intervals
at ground level), and pollarding (cutting above animal grazing height, i.e.
about 23m), were well-known strategies. Cutting cycles are attested in the
Roman period as ranging from 5 to 7 years, for chestnut, and from 7 to 10
years for oak,21 though the charcoal evidence varies greatly on this point.
Both methods of harvesting wood increased a trees life. Other methods of
harvesting were known.22 Because the physical requirements for different
tree types have not changed in time,23 we can use modern scientific studies
on forest management to assist our interpretation of the ancient data. We
now know, for example, at what altitude in a given climate a particular trees
photosynthesis is maximized (so we can scientifically infer a trees favored
growing conditions).24
Cultural Uses of Wood and Wood Charcoal Fuel
In considering how much and what type of fuel might be employed in an

ancient city or town, the range of cultural activities employing fuel need
20 A detailed account of the ancient writers advice on raising various tree types for fuel is
provided in Veal 2009, I, 1724, to be published in Veal (forthcoming).

21 Columella, De Re Rustica 4.33.1; Pliny, NH 17.147.
22 Grove and Rackham 2001, 48 discuss different methods of producing small, so-called
round-wood (i.e. of c. 1030 mm diameter). In the archaeological record, small branches

produced by different cropping methods are indistinguishable from each other, and in fact,
from cropping of whole trees, where the larger sized branches and trunk have been sorted
and removed for other purposes (i.e. building). See also Visser 2010.
23 See, for example, Scagel et al. 1969, 26.
24 In modern Italy, beech (Fagus sylvatica) photosynthesizes optimally between 1200
1500 m when light and water are sufficient (Pignatti 1997, 487490), although the trees range
is 02,000 m in all of Italy. Lack of water is the factor that severely hampers photosynthesis.
to be reviewed, together with other aspects of cultural influence, and the

practicalities of making and sustaining a fire. Historical evidence, when
available, is a starting point, but the analysis of archaeologically collected
fuel-remains from firmly identified cultural contexts provides proof (if not
always conclusive proof). Processes that in general need only a regular fire
temperature in (more or less) an open fire, with the possibility of varying
the temperature, e.g. general cooking, fulling, and tanning, more often used
wood. Processes which required higher temperatures, and which obtained
these by means of containing the heat in a kiln or oven type structure
usually also used wood: ceramics manufacture (pottery, brick and tile),
lime-slaking, glass-making. Roman bathing falls somewhere between the
two, in that hypocaust systems were designed to trap and conduct heat
efficiently, even though the heat required was only for heating water, as
well as heating the ambient air in the rooms. On the other side of the
scale, processes requiring high sustained temperatures of c. 1100C or more,
required the use of charcoal. The industry consuming the most charcoal fuel
was iron-smelting (and smithing, although temperatures required are lower
for smithing).25
Many of the cultural activities named above as using wood, could con-
ceivably have used charcoal as well as wood (and also fire accelerants such as
olive oil, and/or the alternative fuels already mentioned). Most fires would
have been started with kindling (if embers were not retained). Addition-
ally to be considered are public or private rituals, especially cremation, and
the heating of rooms in closed braziers. Lower temperature metal-smelting
and smithing activities (for copper, bronze, silver and gold), could conceiv-
ably have used wood, but charcoals cleaner burning properties (and more
constant temperature) suggest its likely dominance in any metal processing,
since obtaining as pure a product as possible would have been easier with
charcoal fuel. Other considerations might include the possible avoidance
or inclusion of some wood types, e.g. for religious reasons, or because of the
odour or toxicity of smoke.26
25 Sim 1998, 7.
26 Many of the woods of the Rosaceae family have a pleasant perfume/smoke when burnt.
Other uses of some woods would have limited the amount available for burning: e.g. lau-
rel (Laurus nobilis) was used extensively in wreaths, for celebrations, medicinally and in
cooking. Cuttings for these purposes would keep any laurel tree naturally trimmed, with
little waste available for burning. Some woods have a noxious smoke: e.g. oleander (Ner-
ium oleander). Other examples of fragrant woods identified by the author in association
with burials or ritual include Cistus sp., juniper (Juniperis spp.), and myrtle (Myrtus commu-
nis).
46 robyn veal
Fig. 2. Examples of charcoals from excavationi) a typical assemblage from one

context containing dust and charcoals of mixed sizes; ii) a partial small branch:
heart and bark are visible, and even ancient cropping marks. Photos: the author;
and Jennifer Stephens, respectively.
Which Fuel: Raw Wood or Manufactured Charcoal?
Throughout this discussion, reference has been made to both (raw) wood
fuel, and wood charcoal fuel. In the archaeological record, these appear
to be the same thing, but discrimination between the two will ultimately
be essential. Figure 2 shows some examples of archaeological charcoal,
including a typical assemblage from one context of study in Pompeii, and
an example of a partial small branch.
There is no accepted test currently available to discriminate between
archaeological charcoal that is the result of burning raw wood, and archae-
ological charcoal that arises from wood converted into charcoal that has
been re-burnt, although a new method called reflectance is under study
(discussed in further detail later).27 Peoples of the ancient Mediterranean
used both types of fuel. Charcoal was made in charcoal heaps or pits of suit-
ably cut or collected logs and branches, which were then covered in soil and
leaves or cereal residue and sealed with a mixture of ash and soil (to exclude
oxygen). The wood heap was then slowly charcoalified rather than burnt.
Modern ethnographic studies28 help to illuminate this process, which is also
documented in the ancient sources.29 Charcoal making typically occurred in
the forest where woods were sourced, and charcoals were then bagged up
and transported to private homes or markets nearby. Figure 3 shows a mod-
ern charcoal heap that has been prepared (by careful stacking of branches
of equal length on a flattened mound of charred residue). Charcoal making
piazze had to be located near water so the charcoal maker could dampen
down the mound if necessary to prevent its full combustion (or indeed acci-
dental burning of the surrounding forest).
Naturally, the extensive use of charcoal (as opposed to raw wood) in a
society has a much greater impact on that societys woodland resources. To
gain a full understanding of forest consumption for fuel, we need therefore
to understand how much charcoal was being consumed, and also, how
efficient a conversion process was in place for making charcoal from wood.
This is one aspect of the fuel economy for which archaeological science
and ecological modeling together may provide some answers. Ethnographic
data suggest that conversion from raw wood to charcoal can be as efficient
as 3 or 4 portions of wood to make one portion of charcoal, to a ratio of
27 Work has been undertaken by a number of authors, only a few of which are listed here:
McParland et al. 2009; Scott 2005; McParland et al. 2010; Scott and Veal 2010.
28 Veal 2009, I, 142.
29 See for example Theophrastus, Hist. Pl. 5.9.24.
48 robyn veal
Fig. 3. Modern charcoal stack ready for covering with mud, leaves and charring
residues, which was then set to char by insertion of a burning log. The charcoal
maker intermittently checks progress by inserting a thin stick progressively through
the mound: white gases indicate water vapor escaping; blue indicate emission of
organic volatile gases. After some time no gas or vapor is emitted and the charcoal
maker knows the process is finished. Charring of a stack this size takes about one
week. (Photo: the author, Borgo Pace, Le Marche).
as much as 10 or even 15:1.30 Some of these data arise from tropical and/or
very dry climates so their relevance to the Mediterranean needs careful
interpolation.
Charcoal Analysis
Charcoal is analyzed by identifying charred wood cellular structures under

reflective light microscopy.31 The structures closely resemble those of wood
in its natural state and so identification is sometimes possible to species
30 For a full review see Veal 2009, I, 142146.

31 Leney and Casteel 1975.
level. Where this is not possible, identification is made to the most refined
level possible (genus, sub-family, or family level). Knowledge of regional
vegetation from pollen and phytolith studies and from flora guides can also
contribute. Identification is made by comparison with wood atlases,32 and
modern reference charcoal.33
Charcoals results are computed either by fragment count (per wood
type), or by weight. Both methods have drawbacks: charcoal keeps on frag-
menting post-excavation with any type of handling, while weights can be
biased substantially by the amount and variety of inter-cellular mineral
inclusions, and any attached soils (which often far outweigh the charcoal
itself). The process of recovery by flotation34 helps avoid the problem of
weight bias, but it also tends to fragment the charcoal further, increasing
laboratory analysis time for little return, and occasionally resulting in the
loss of smaller wood types, although this rarely affects major trends. Frag-
ment count is now the mode most often used, as it is the most convenient
and efficient.
Methodology in charcoal analysis has been fairly strongly influenced by
pre-historic archaeological method, where flotation has led to collection
and identification of charcoal fragments over 2mm.35 Hand collected char-
coals and those collected over dry sieves of 45 mm are also examined. In
urban environments, city-wide syntheses, other than that for Pompeii (dis-
cussed below), are rare.
32 The IAWA handbooks for hardwood and softwood identification (Wheeler et al. 1989;
Richter et al. 2004), are used as the basis of nomenclature for recognizing the possible
macro and microscopic wood structures. Schweingrbers 1990 atlas is the main standard for
European charcoal identification. Many other publications relating to localized flora have
been used. An online database that is being constantly augmented is likely to become the
flagship point of first reference: Schoch et al. 2004.
33 Difficulties can arise if the modern environment is substantially different (i.e. lacking
the ancient wood types). The use of reference collections maintained in the northern part of
Europe to identify Mediterranean assemblages is also an issue. Climate affects growth and
cellular structural habit. Collection of locally curated specimens (certified by a herbarium)
is highly desirable, and from differing parts of a tree, as well as (ideally) from different
altitudes.
34 This is a method of archaeological recovery where a fixed amount of soil is washed over
a fine mesh and the floating material (in the light fraction) is recovered separately from
the sinking material (in the heavy fraction). This method complements regular dry sieving
of soils and/or hand collection during excavation, and allows collection of small and fragile
materials such as seeds, charcoal and small bones.
35 Asouti 2004a provides a good overview of the history of the discipline. Methodology
continues to be refined, see for example, Asouti 2004b and more recently, Thry-Parisot et al.
2010.
50 robyn veal
Emerging Techniques in Charcoal Analysis
Charcoal analysis is still a young discipline. Charcoal remains are not always
collected by excavators, sometimes for reasons of cost, and sometimes be-
cause its usefulness is not fully appreciated. Even when collected it may
reside in storage and be the last artifact type to be analysed, with seeds and
other macro-remains being privileged over charcoal. However, in the last
twenty years or so, significant theoretical and experimental work has been
carried out. Much of this work has related to fragmentation rates and repre-
sentativeness of archaeological charcoals, and has provided a stronger base
from which we may now be more confident about the reliability of charcoal
reports. The discipline now needs to move forward with the use of newer
scientific techniques to characterize the quality of fuels employed, and the
relative ratios of wood to charcoal fuel. An overview of progress and emerg-
ing tools follows.
General Studies on Fragmentation Rates and

Representativeness of Charcoal Assemblages
Aspects of fragmentation patterns, and the representativeness of collected
archaeological assemblages have been under discussion for some time, and
are now being specifically tested. Whereas in the past focus has been on col-
lection of charcoals sized in classes from 12 mm, 24 mm and above 4 mm,
recent work has begun to demonstrate that in the very small classes (i.e.
12mm) oak can be especially over-represented (in the European flora).36
This work is paving way for greater confidence in examining the >4 mm
sized charcoals, i.e. those which may be collected by dry sieving (and not
only flotation). Larger charcoal fragments are faster and easier to fracture for
identification and also generally provide a greater level of security of identi-
fication, although much is possible even with 2 mm fragments. Other work
suggests the required number of fragments per archaeological level is far
fewer than had previously been the accepted practice, although the partic-
ularities of any one site have always to be considered.37
36 Chrzavzez et al. 2011.

37 Typically a quantity of 350450 fragments per level has been recommended (Chabal
et al. 1999, 66), based on extensive work in pre-historic excavations. A level is a notion that
requires definition according to the type of archaeological contextual environment. Charcoal
work in the UK and Ireland is simplified by lower floristic diversity by comparison with the
Mediterranean. Even so, recent work demonstrates that fewer fragments (6080) may often
be sufficient to ascertain major trends (Veal 2009, I, 87; OCarroll and Mitchell 2011; Py 2006,
41). These developments mean charcoal analysis is becoming a more cost-effective operation.
Sampling of sites for charcoal is regularly limited to hearths or kilns and

other areas of concentrated charcoal deposits. However this strategy merely
provides a view of the last (or last few) burn episodes(s), and often means a
restricted number of wood types are identified. Random sampling and col-
lection over all context types, will, in the long run, provide a greater view
of wood diversity (and thus the breadth of the fuel economy).38 Collection
of soils and analysis from off-site areas may complement the archaeologi-
cal site charcoals, providing data to assist vegetation reconstruction (since
they will more likely reflect ambient woodland). On-site urban charcoals
will reflect those woods from the environment specifically selected for fuel.39
Sometimes the two goals of environmental reconstruction and analysis of
the fuel economy are improperly conflated. Rarely can both goals be met by
the same dataset because: (1) environmental reconstruction from wood frag-
ment identification (only) is incomplete (non woody plant forms are rarely
preserved); and (2) selection of wood for different economic uses cannot be
ascertained if the archaeological traces of different wood consuming activi-
ties are not observable in the archaeological record. In pre-historic studies,
the process of human selection of woods for burning cannot necessarily be
considered to have ranged across all woods available in the environment
(as has often been traditionally accepted).40 In urban environments, on the
other hand, with specialized industries and various options for transport,
more selective use of woods for different purposes appears probable. In
some cases we have sufficient data to confirm such trends but few studies
are yet able to attempt the level of integration required of the archaeological
record with the charcoal results to provide regular proof.
Tree Ring Counting and Curvature

Dendroanthracology or the science of tree ring analysis specifically in the
service of charcoal research has arisen as the new standard for estimation of
tree cropping indicators. Methods for the measurement of fragment diam-
eters (which permit us to infer the raw woods metrics) include (1) place-
ment of a charcoal fragment on the familiar ceramics rim and base mea-
surement template, and (2) estimation of fragment size and diameter using
38 Py 2006, 41.
39 They may also reflect timbers chosen for construction (or furniture), if a holistic burn
event is identified, which of course, muddies interpretation as far as fuel analysis goes.
40 Gelabert et al. 2011 provides a detailed review of the principle of least effort (PLE) for
firewood collection by pre-historic peoples and offers evidence for the need for a much more
nuanced approach.
52 robyn veal
trigonometry, both manually and digitally.41 The former method is fast and
economical, but potentially more error-prone, the latter is more expensive
and time-consuming but less error-prone. Having estimated diameters of
charcoals (and having allowed for their c. 15 % shrinkage from raw wood),
one can construct a picture of the sizes of the branches used and of cropping
strategies. We can complement this method by estimating ring curvature,
which falls broadly into three groups: flat (indicating use of mature large
branches or trunk wood); moderately curved (indicating use of medium-
sized branches); and highly curved (indicating use of small branches or
twigs).42 Not all fragments in an archaeological assemblage may lend them-
selves to this type of analysis (they may be too small), but major trends are
usually discernible. Observation of repeated use of small woods (in conjunc-
tion with a robust presence of the same type of wood) is usually interpreted
as being indicative of sustainable forest management. Use of larger scale
woods, especially trunk wood, indicates exploitation at a higher opportu-
nity cost (since whole trees, or even very large branches, take longer to
regrow than small wood). This may, or may not be a less sustainable practice,
depending on the cropping cycle and time allowed for regrowth (which can-
not always be inferred from the charcoal). Reference to nearby pollen stud-
ies (which show large scale changes in woodland makeup), and/or large-
scale changes in wood types in the charcoal over time (e.g. from large decid-
uous trees to smaller colonizing or macchia types), together may be the
markers of over-exploitation.
There are many other types of observation that allow inferences to be
made about whether or not freshly cut wood, or older, aged wood has been
used, about stresses that the growing tree may have been under, about
fungal or insect infestations, and about aspects of the combustion process.43
However, these usually relate to occasional observations about individual
fragments, and can rarely provide statistically valid information.
Heat Values of Different Woods

The calorific potential (or heat value) of different woods is a useful tool for
examining wood utility, and also for comparing the heat potentials of alter-
native fuel types. Heat value is not a fixed value for any one wood type since
it will vary with moisture conditions, size and shape of faggots burnt, and
41 Thry-Parisot et al. 2010, 145. Marguerie 2011 has more recently tried to qualify and
quantify limits to dendro-anthracological methods.

42 Marguerie and Hunot 2007.
43 Carrin 2006, 86.
other ambient factors in combustion.44 However, specific gravity (at a spe-

cific moisture content) is a rough proxy for expected heat values.45 Average
specific gravities for dry wood, for most of the common broadleaf trees of the
European flora, range from around 0.5 (for riparian wood types, i.e. water-
loving species like alder (Alnus spp.)), to about 0.75 for beech (Fagus sylvat-
ica) and oak (Quercus spp.).46 The implication is that if inferior heat-value
woods are consumed, then a greater volume will be required to achieve the
same temperature in a fire or kiln. The naturally higher volume of water
in a riparian wood must first be driven off (and this consumes calories).
Consumed as charcoal, fuel heat outputs of different woods become more
similar, and charcoal in general provides about double the calorific potential
(by weight) of wood. However, in the process of charcoalification of wood,
the water content is removed (which constitutes part of the weight of the
wood), and so for the riparian types, again, a greater weight or volume of
wood is required to make an equivalent weight of charcoal than for a denser
wood. Only potential calorific value can be considered, as actual heat value
is dependent on a range of different factors, and the calorific values reached
in a fire never achieve their theoretical maxima.47 The matter is further com-
plicated by the presence of resins in some woods. These tend to be highly
combustible, and tend to increase heat potential. Also, so-called lower heat
potential woods (often called soft woods) may be advantageous if they
ignite and burn quickly (for example, in the process of initiating a fire).
Measuring the Absolute Burn Temperature

of Charcoals: the Reflectance Technique
The reflectance technique is a method borrowed from coal studies that
allows the estimation of the absolute temperature to which a charcoal
fragment has been subjected. Experimentation by L. McParland on various
common (modern) wood types charred at various temperatures and for
various times has proved the linear relationship between reflectance and
burn temperature, although the effects of taphonomy48 and the usefulness
44 Lyons et al. 1985.

45 Bootle 2004, 202.
46 Macchia vegetation types, which by form are scrubbier and denser, usually have quite
high heat values. However, they can be more difficult to crop and manage (being hard to cut
and sometimes thorny).
47 Lyons et al. 1985.
48 Taphonomy is the archaeologists term to describe every type of process that happens
to an artifact after deposition, whether environmental or anthropogenic. See a recent study

54 robyn veal
of this method for archaeological charcoal are just now being tested.49 It may
provide a complement to other temperature estimation methods, and it is
already revealing more about the taphonomy of archaeological charcoal in
various contexts.50 Reflectance can also provide an indication as to whether
raw wood, or charcoal fuel, has been employed. Measurements of wood fires
will tend to exhibit low reflectance of a range of temperatures (often on a
Bell curve) of c. 100C to c. 450500C, while charcoal-only fires will show
temperatures starting from c. 350C upwards (the approximate temperature
at which charcoal starts to form) and will lack low temperature readings.
Of course, the matter is complicated by the fact that both fuel types may
have been used in a fire. Studies are continuing on a range of archaeological
charcoals associated with particular cultural uses.
Charcoal Quality Analysis: Measuring Purity of Charcoal

Modern fuel economists spend much time on estimating wood charcoal
consumption in developing countries where wood is still a major fuel. In
these economies, different types of charcoal are produced for different pur-
poses. In particular charcoal destined for domestic use (manufactured at a
seemingly good conversion ratio) is only lightly charred in order to leave
in some of the organic volatiles (so combustion later on is not too difficult);
while charcoal destined for industrial use (especially iron smelting) is heav-
ily charred in production in order to produce as high a quality product as
possible. The difference between the two types may be measured by their
absolute carbon content.51 For domestic use, charcoal of about 65 % carbon
is desirable, while for industrial use, charcoal of about 8085 % carbon will
provide a hotter fire. The charring process in the latter instance is longer, and
(Thry-Parisot et al. 2010) which attempts, confusingly, to redefine taphonomy as all pro-
cesses associated with the chain of events that leads to an archaeological assemblage, includ-
ing human selection of the wood.
49 McParland et al. 2009. More recently the method has been used to test cremated bones
with less success (Veal et al. 2011) and iron smelting charcoal, with good success (these results
will be reported shortly).
50 Taphonomic effects in charcoal typically increase fragmentation, or even reduce it
to dust, thus rendering it difficult to separate from the soil matrix, let alone identify it as
to wood type. Early trends are suggesting that, especially for charcoal remains associated
with industrial processes, the charcoal collected may well sometimes only be unconsumed
charcoal fuel, and will reflect the temperature of formation of the charcoal (in charcoal
production), rather than that of the industrial process under study.
51 A further difference is that woods charred for a lower period will produce some smoke
(when the remaining organic materials are burnt), while highly charred woods will produce
very little. Both produce far less smoke than raw wood.
the conversion rate seemingly poorer, i.e. a lower volume of charcoal is pro-
duced using the same volume of wood, but the product has a higher calorific
potential.52 It is likely the same differentiation occurred in the ancient world,
and work has begun on charcoals identified from industrial and domestic
environments in Pompeii to test this hypothesis.
Pompeii As a Case Study
The author has completed a city-wide synthesis of Pompeiis fuel supply

from the third century bc to ad79, which is published elsewhere in sum-
mary form and will be published in detail shortly.53 It is useful to review here
how this study fits the model described above. Pompeii was a coastal site
on fertile soils, with a climate conducive to the growth of a large range of
broad-leaf deciduous and evergreen trees in the city, in the plain and in the
nearby hinterland. Baths, bread baking, and iron-smithing, to name only a
few activities, consumed fuel on a large scale. Domestic consumption at the
elite level would likely have included both wood and especially charcoal,
for cooking and for heating braziers, among other purposes. Charcoal fuels
capacity to burn at a higher and more consistent temperature, with less or
no smoke, was likely useful both in the kitchen and the triclinium. Recipes
from Apicius suggest both fuels were used in the kitchen,54 as do the ranges
of ceramics and stove and oven types found in the city. In the study about
4,000 charcoal fragments were examined. These were dated by known rela-
tive typologies for ceramics and coins from the third century bc to ad 79, and
52 Schenkel et al. 1998.

53 A preliminary investigation of the House of the Vestals (Veal and Thompson 2008),
provides an overview of the approach; my doctoral thesis (Veal 2009) will appear shortly in
revised form (Veal forthcoming). A summary of some of the major trends appears in Veal
2012.
54 Apicius (cf. the Grocock and Grainger 2006 edition) provides a large number of verbs to
describe cooking methods including baking (in the oven, or possibly in special closed baking
dishes on the stovetop called clibani); boiling, frying, steaming, and smoking. Most recipes
describe cooking in a pot over a fire, and in these instances we cant tell precisely if a wood
or a charcoal fire is intended. Some give us more information, for example, his recipe for
Lucanian sausages (II.iv) requires them to be hung up ad fumum. Smoking requires fresh
(usually fruit) wood, for appropriate flavours to be imparted. But his recipe for pisa farsilis
(V.iii.2), a complicated recipe for peas layered with cooked meats, is finally finished in a
baking dish over a slow fire (lento igni imponis). We infer this to be charcoal, as raw wood
smoke would be very undesirable in an oven, or even over a closed baking dish on the open
stove. This is discussed further by Renfrew 2004, 23.
56 robyn veal
Fig. 4. Summary results of diachronic study of wood fuel of Pompeii c. third c. bc to

ad79.55
they were excavated from four separate locations by three different exca-
vation teams. The excavations were carefully documented, and recorded by
these teams using substantially similar methods of recovery. Detailed exam-
ination of the associated archaeological records allowed close integration of
charcoal analysis with the archaeology permitting a very nuanced view of
the results on a century-by-century basis. The architectural history of these
sites suggests that as many as eleven different property owners, from a range
of socio-economic classes, may have been responsible for this material. The
results showed a clear use of one dominant fuel, beech (Fagus sylvatica),
through the whole period, with secondary use of oaks (Quercus spp.), maples
(Acer spp.) and hornbeams (Carpinus spp.), and a smattering of conifer,
riparian and other types of woods. The beech constituted 5075% of the
assemblage depending on time period and location, and it diminished over
time to be replaced in part by the secondary woods, but also by some fruit
and nut woods (e.g. Prunus spp., Castanea sativa, and Vitis vinifera). Fruit
and nut woods were observed in small, but then increasing amounts in the
1st century bc and 1st century ad, i.e. from about the time of Roman colo-
nization. A summary graph is provided in figure 4. Work is ongoing in two
55 From Veal 2012.

further locations in the city, and the fuel economy trends appear to be the
same, and so, now, with confidence, we may speak of the citys fuel economy
on a holistic basis.
The primary analysis of the charcoal data led to economic modeling of
the quantities of wood and charcoal fuel consumed in the city, using a lim-
ited range of population figures, over various relative proportions of fuel
types, and ranges of charcoal making efficiency ratings.56 Factoring in the
preferred growth niches for the wood types has allowed for basic assessment
of forest growth areas on mountain and plain required to support the fuel
quantities inferred. Subsequent analysis also explored cost/benefit possibil-
ities for different types of transport (road, river, sea).57 A picture of a large and
complex fuel supply system emerged where the city was greatly dependent
on its somewhat distant, (i.e. 1525km) hinterland as the major provider of
its fuel. The social and economic structures that supported and sustained
the wood fuel economy can in part be inferred. Work is ongoing to refine
this model through (especially) estimating the ratio of wood to charcoal use
(through reflectance). Tests to compare absolute carbon content are also in
train, both on the archaeological charcoals, and some obtained from a con-
trolled modern charring experiment.
Conclusions
With respect to Mediterranean studies, W.V. Harris has called for a natu-
ral history, articulated through periods,58 and he names fuel as one of the
three physical necessities that specifically require study. It is to be hoped
that this discussion points the way forward. Recent scholarship on energy
consumption in the ancient Mediterranean has examined data holistically,
using a top down approach (cf. Malanima, this volume). Wood fuel, con-
stituting perhaps 50% or more of the total energy that fuelled the ancient
economy, has been broadly examined using assumptions about average
wood consumption and average calorific return from wood. This top down
approach has, at least, shed light on the importance of fuel. These aver-
ages were naturally moderated by considerations of temperatures required,
extent of industrial activity, and other cultural concerns, but at present we
have not yet qualified these factors in detail.
56 The base economic model was presented at the conference from which this paper arises
and may be examined in Veal 2012. This is a base model, and is currently being refined.
57 The economics of the transport options will be discussed in Veal forthcoming.
58 Harris 2005b, 12.
58 robyn veal
Consumption of wood and wood charcoal is often discussed in terms

of metal smelting and smithing, and the fuelling of baths, but many other
industrial activities used wood and charcoal. Some were great consumers,
such as lime-slaking, glass-making, and tile- and brick-making. A closer
examination of the different fuel consuming processes is needed, together
with a much closer understanding of the ratio of raw wood to charcoal
consumed, and the productivity achieved in charcoal making, for different
cultural purposes. Charcoal identification, together with the developing
associated scientific studies, are the key to revealing the finer details about
raw wood and wood charcoal fuel consumption. The variability in the supply
system, in terms of quality of fuel and charcoal, was potentially much greater
than the averages we have been able to consider to date.
Pompeii offers the unique example of a well-documented city that has
provided a base for moving forward to analyze other citys fuel supplies,
but the differing geological, climatic, political and cultural frameworks of
settlements around the Mediterranean need to be analyzed in order to
accurately examine their fuel supplies. Charcoal offers the potential to view
ancient forest husbandry. It provides a bottom up view of the actual fuel
consumption patterns of a city or town. With time, and future studies, the
potential to provide a much more nuanced view of Mediterranean-wide
practice may be realized.
PART TWO
CLIMATE
WHAT CLIMATE SCIENCE, AUSONIUS, NILE
FLOODS, RYE, AND THATCH TELL US ABOUT THE
ENVIRONMENTAL HISTORY OF THE ROMAN EMPIRE
Michael McCormick
Recent scientific advances are transforming our understanding of how, why,

and how fast climate systems change. That understanding is still imper-
fect. Yet what has become alarmingly visible today invites us to explore how
societies and environments interacted in the past. Modern climate scien-
tists seek signals in natural scientific proxy data such as tree rings and ice
cores that testify indirectly to past climate conditions. A rich if incomplete
memory of climate change also lies buried in the written records and archae-
ological remains of earlier civilizations. In some real sense, a human being
who attests a climate event in a particular place and time constitutes the
ultimate proxy testimony to past climates. But because this testimony is
subject to all the complexities of human recording, historians and archaeol-
ogists must take the lead in the urgent study of this human proxy data.
The human record of western Eurasia is deep, incomparably rich and
well-studied; it has deservedly attracted the attention of climate scientists
working on the early modern period.1 The medieval European records also
offer much to the climate scientist, and a project to assemble and ana-
lyze the abundant Arabic evidence for medieval Middle Eastern climate is
underway.2 While valuable work has been accomplished in assembling the
1 See in particular Euro-Climhist. A Data-Base on past Weather and Climate in Europe
and its Human Dimension elaborated at the University of Bern, Abteilung fr Wirtschafts-,
Sozial- und Umweltgeschichte (WSU), a new version of which is supposed to be available
online in Spring 2012: http://www.hist.unibe.ch/content/institut/abteilungen/wsu/index
_ger.html. A good overview of the evidence and research to that date for Europe from the
Middle Ages forward is Brzdil et al. 2005. See also the next notes.
2 Pioneering studies of the European Middle Ages came from H.H. Lamb and Emmanuel
Le Roy Ladurie; see the important summations of their work: Lamb 1995 and Le Roy Ladurie
2004. The most valuable printed collections of climate events for western Europe are Alexan-
dre 1987 and Buisman and van Engelen 1995. Vogt et al. 2011 describe a study from medieval
Arabic sources of some 5,000 reports of weather phenomena between 800 and 1500ad, but
make no reference to plans to make the data set more broadly available. For two important
collections of climate evidence from the late antique and medieval Byzantine records, see
the next note.
62 michael mccormick
data from late antiquity and the Byzantine Empire, there is so far no exten-
sive and critical corpus of ancient climate evidence.3 To fix that, a small
group of scholars at Harvard has begun creating a first geodatabase of writ-
ten evidence on the Roman and post-Roman climate.4
Yet, by themselves, the written records rarely suffice. As the real possibil-
ity of accurate reconstructions of past climates takes shape, we must pick
up the challenge of leveraging the independent testimony of the natural
archive of scientific proxy data by comparing it to the existing historical
and archaeological evidence. Of course the different sorts of evidence are
immensely complicated. How many climate scientists can scan the meter of
Ausonius allusive poetry to determine the ambivalent grammatical case of
a particular Latin word, and so understand his phrase correctly? How many
philologists can assess tree rings testimony on precipitation patterns or the
18O records from multiple ice cores that testify to temperatures in central
Greenland? The solution must be for scientists and scholars to collaborate
and to prepare, explain, and share our evidence in ways that are compre-
hensible and useful to specialists working in very different fields. Only then
can the archaeologist and historian consult, for instance, the results of den-
droecologists analyses, and the dendroecologists use those of the scholars.
That is a huge challenge, but it is only the beginning. Once we have identi-
fied the historical and natural scientific evidence, it is essential to compare
and connect them, an undertaking that presents its own problems. Then
comes the most delicate yet crucial task: to identify possible cases of climate
change and human response from the combined evidence of the natural
scientific, historical, and archaeological records. This paper explores these
issues through a few case studies. The first considers the long-standing and
intricate debate on the date of a famous Latin poem in the light of new
palaeoclimatic data. The implications of new scientific data for a traditional
philological problem illustrate the complexities and potential of coordinat-
ing the different kinds of evidence. Next, a summary of the present, prelimi-
nary state of knowledge about climate conditions across the Roman Empire
will introduce some reflections on pressing areas for further investigation as
well as pitfalls to avoid when interpreting the data that is already available.
Finally, we will explore some potential cases of human response to climate
change in the late Roman and early medieval archaeological record. It is
almost needless to add that this essays case studies are offered in a spirit
3 Climate reports from written sources in the Byzantine Empire, beginning in 300ad, are
comprehensively catalogued by Teleles 2004 and Stathakopoulos 2004.

4 McCormick, Harper, et al. 2012. See further below.
the environmental history of the roman empire 63
of experiment and exploration. Scientific study of the ancient climate has

barely begun; new data arrive every day with the potential to revise and
deepen our knowledge on every aspect of the phenomena discussed here.
1. The Poet and the Tree Ring:

Dendroecological Dating of Ausonius Mosella
In the later fourth century, the Gallic teacher and statesman Ausonius
(ca. 310ca. 395) composed one of the most exquisite poems ever penned
about the environment and Roman civilization. Mosella describes the way-
farers wonder as he rides from the mighty new fortifications along the
Rhine across the Hunsrck, the hilly plateau that separates the Rhine from
the Moselle River, and reaches the Moselles terraced vineyards.5 Water
courses throughout the poem. The traveler hears the water-powered saw-
mills, whose screaming teeth slice the stone (vv. 361364) likely intended
for splendid buildings rising around the imperial capital at Trier; he muses
on the river fishermen and, famously, on their succulent fish (vv. 75149).
The proud Roman breathes finally free in the bright sun that shines across
the Roman space, civilized now that he has moved away from the trees that
darken the light, not to mention that air fouled by forests (vv. 1022). So
finely detailed is the poem that, imaginary or not, it has proven possible to
track Ausonius trip along a well-known Roman road. Station by station we
can follow him through what the courtier depicts as Roman Gauls renewed
prosperity.6
Ausonius talks about more than just rivers. He notes water conditions
along the way. As he crosses the small settlement known today as Kirch-
berg, Ausonius remarks I pass Dumnissus, drought-stricken (arentem) with,
all around, its fields parched: praetereo arentem sitientibus undique ter-
ris / Dumnissum. Striking juxtapositions pervade Ausonius artistry, and he
continues here by contrasting dry Dumnissus with the next place he tra-
verses: Tabernae, watered by an ever-flowing spring, riguasque perenni
5 Ausonius, Mosella, ed. Green 1991, 115130.

6 Opinions have varied on whether in lines 122 Ausonius is describing an actual or
an imaginary trip. For instance, Green 1991, 463 (but cf. 451) left open the possibility of
an imaginary trip or an actual one retouched, but later Green 1997, 214, reckoned it wiser
to assume it bears no relationship to any historical event; Shanzer 1998b, 228230, seems
inclined to think it a real trip. The new climate evidence suggests that whether reimagined
or more or less recorded, Ausonius description reflects a specific experience in a specific
year.
fonte Tabernas. In fact, Kirchberg sits on a waterless ridge, and so relied on

rainfall for irrigating its fields. Its parched or, literally, thirsty fields seem
to have been suffering from a drought.7 The modern identity of Tabernae is
less certain, but the proposed location at the place known as Heidenptz,
25km west of Kirchberg, fits the ever-flowing groundwater of the poem.8 The
very name Heidenptz seems to me to confirm the identification, for the
local dialect word Ptz derives from Latin puteus and designates a deep well
from which water could be drawn.9 And Heiden, in the sense of pagan
is a common element of German place names. When the word occurs in
toponyms in areas such as the Hunsrck where the Germanic newcomers
culture took root inside the Roman Empire, it typically refers to structures
or ruins inherited from the Romans. So the locals identified this site as the
Roman well.10
For centuries, classicists have sifted through Ausonius masterpiece and
weighed the potential chronological implications of every allusion. It refers
to, and so surely was composed after a victory over the barbarians celebrated
by Valentinian I and his son Gratian. This very likely was the campaign of 368
or 369 in which both emperors participated according to Ammianus Mar-
cellinus.11 The poem was clearly written before Valentinian I died on Novem-
ber 17, 375 since, even if he is not named, this emperor and his doings are
7 Ausonius, Mosella, 511, Green 115116. See Wightman 1970, 130, who describes the geog-
raphy in the light of the Peutinger Table and reports the waterless character of the site.
Kirchberg does indeed sit on a ridge which the Geological Service of the German federal
state of the Rhineland-Palatinate maps as the boundary between two groundwater zones or
aquifers: see the website Rheinland-Pfalz, Landesamt fr Geologie und Bergbau, LGB Hydro-
logische Karten http://mapserver.lgb-rlp.de/php_hydro/index.phtml selecting the layer for
Grundwasserkrpergrenzen, and setting the search for the Gemeinde of Kirchberg.
8 Wightman 1970, 131, notes that it was still rather marshy in 1970. The Rhineland-
Palatinate Geological Service mapping site cited in the previous note is less revealing for the
two leading possible sites of Tabernae, identified as either Belginum near Wederath or the
site about two km southwest of Wederath known as Heidenptz; cf. Green 1991, 465.
9 Rheinisches Wrterbuch ed. Mller et al. 19281971, vol. 6, 12481251, s.v., consulted
online, November 30, 2011, Universitt Trier, http://woerterbuchnetz.de/RhWB/; cf. Bach

19521956, vol. 2, 1.287.
10 Bach 19521956, vol. 2, 1.356357.
11 See in general on Ausonius career, Jones et al. 19711992, vol. 1, 140141, Ausonius 7,
and the detailed study of Cokun 2002a. Ausonius summons from Bordeaux to the imperial
court and the Moselle region to serve as tutor to the young Augustus Gratian has been placed
somewhere in the mid-360s: e.g., Green 1991, xxviii. Cokun 2002a, 3740, concludes it could
have occurred as early as 364, but happened more likely after August 367 and the summer of
369. Ausonius manifestly wrote Mosella after he had come to the imperial court as tutor: see,
e.g., on his expression mea maxima cura (v. 450, Green 129), below, note 15. The campaign:
Ammianus Marcellinus, Res gestae, 27, 10, 616, ed. Seyfarth, 2, 1978, 5154; cf., e.g., Seeck 1920,
5, 3738.
mentioned repeatedly and unambiguously.12 Within this seven-year span,

over the last two decades alone, four different dates have been proposed
for the poems composition. The later fourth centurys rich documentation
allows good arguments for all of them. The enduring debate also reflects
the finely wrought wording and allusive quality of Ausonius poetry. Thus
he clearly alludes to the high honors recently awarded to specific individ-
uals, but Ausonius does so in a way that leaves exquisitely ambiguous the
exact person or persons to whom he is referring. As his most comprehensive
editor has observed, this could well be part of the sophisticated authors cal-
culation: Ausonius aimed to flatter a particular highly placed minister, but
chose to do so in such a way that more than one such official might delight
in recognizing himself in the pattern of praise.13
The two earliest datings (368 or a little earlier; late 369 through mid-370),
have won little support.14 One or more allusions that might seem inconsis-
tent with an origin in 370371 have been adduced in favor of the late date, in
375, or possibly, in favor of a slight retouching in 378 or 379 of a poem origi-
nally composed at the earlier date.15 The most substantial of these supposed
12 Mosella alludes (vv. 420426, Green 128) to the triumphs the imperial father and son
celebrated together at Trier for the victories achieved beyond the Rhine and the Neckar, wit-
nessed by the personified Moselle; the laurel draped letter of victory had arrived recently,
and more are predicted to come soon: sed Augustae veniens quod moenibus Urbis / spec-
tavit iunctos natique patrisque triumphos / hostibus exactis Nicrum super et Lupodunum /
et fontem Latiis ignotum annalibus Histri / haec profligati venit modo laurea belli. On these
victories, see, e.g., Green 1991, 506508, and below, n. 15. On the litterae laureatae announc-
ing imperial victories to the population in this period, see McCormick 1990, 3943, and, in
general, 190196. On Valentinian Is death as an unavoidable terminus ante quem, see, e.g.,
Mondin 2003, 189190.
13 Green 1978.
14 Sivan 1990s early date of 368 has been carefully and comprehensively refuted by
Green 1997, Shanzer 1998a, and Shanzer 1998b. Even if one rejects with him the common
identification of the mysterious figure alluded to in vv. 409411 as Petronius Probus, consul in
371, Cokun 2002bs learned arguments that Ausonius could have finished the poem by the
middle of 370 are suggestive rather than compelling. In favor of the identification with Probus
see, e.g., the subtle and persuasive observations of Shanzer 1998b, 216228. See further for a
date in 370371, Mondin 2003.
15 Drinkwater 1999 argues with some reason for hidden complexities in Ammianus ac-
count of the military campaigns with which Mosella is usually connected, and therefore
concludes for a terminus post quem of 370. He adduces in favor of his late date of 375: Auso-
nius expectation of a consulate, on which see further below; the success of Valentinian Is
border campaigns implied by the statement that castles are becoming granaries (vv. 456457,
Green 129: addam praesidiis dubiarum condita rerum / sed modo secures non castra sed
horrea Belgis); and Ausonius laying claim to tutoring Valentinian II. No specific evidence is
adduced for the granaries assertion; lacking that, I see no way of dating such a propagandistic
claim early or late within this period. Drinkwaters assessment is not universally shared that it
inconsistencies comes when Ausonius refers to the insignia of the consulate

in connection with the imagined future time of his retirement (vv. 448453,
Green 129). These words have been interpreted to mean that he announces
here his expectation of receiving the ordinary consulate, the supreme honor
of the senatorial class. Not injudiciously, such an announcement has been
considered implausibly tactless for an accomplished courtier if the emperor
had not already committed to designating Ausonius as ordinary consul. That
designation in turn has been viewed as unlikely, given that Ausonius actu-
ally held his ordinary consulate only in 379, i.e., some eight or nine years
after the possible original composition of the poem ca. 370 or 371. Hence, it
is thought, the allusion must have been added to the poem only at the time
when Ausonius was sure he was going to be consul, presumably in 378.16
On the other hand, since at least the nineteenth century, some have
observed that Ausonius words do not explicitly specify the ordinary con-
sulate. The poem mentions only the insignia of consular rank, the fasces and
the chair of state, the sella curulis shared by various high-ranking officials,
including consuls. Thus Ausonius might be referring only to the consular
honors that were routinely conferred on higher ranking officials and which
entailed the use of these insignia.17 Such consular honors were prestigious
but fell well short of the ordinary consulate, whose holders paid for and led
the New Years celebrations in the capitals, gave their name to identify that
year ever after, and rose to the highest levels of precedence available to a
non-emperor.18 In fact, a law of 381 cites precisely these same insignia as sym-
bolic of the proconsular rank of the head imperial notary.19
would be farfetched for Ausonius to project himself in 371 as tutor to a yet to be born or new-
born imperial child, as would be implied by the plural nati (vv. 448453, Green 129: ast ego,
quanta mihi dederit se vena liquoris, / Burdigalam cum me in patriam nidumque senectae /
Augustus pater et nati, mea maxima cura, / / mittent emeritae post tempora disciplinae, /
latius Arctoi praeconia persequar amnis.) When that distant day comes when Ausonius will
be dismissed by the emperor to his retirement nest in Bordeaux, he foresees that he will
have accomplished his maxima cura of educating the imperial children (cf. Venus in Aen.
1.678 and Ausonius own use of the expression in reference to his imperial student Gratian in
Cento nuptialis 8, Green 134). Green 1991, 511, finds it natural that Ausonius should expect to
teach a new child as well as the older sibling. One could go further and see in this statement
a quiet assertion by Ausonius of his ambition to do so.
16 See, e.g., Sivan 1990, Shanzer 1998a and Shanzer 1998b.
17 de la Ville de Mirmont 1889, 130132; Green 1991, 511; further Mondin 2003, 191192;
Cavarzere 2003, 178, ad locum, who notes nevertheless that Ausonius uses almost exactly the
same language about his ordinary consulate: Praefationes, 1, 3738 (Green, 4): et, prior indep-
tus fasces Latiamque curulem, / consul, collega posterior fui. See also however below, n. 19.
18 Bagnall et al. 1987.
19 Codex Theodosianus, 6, 10, 3: Notariorum primicerium in numero proconsulum habe-
But was it in fact so unlikely that Ausonius could have been promised an
ordinary consulate seven years ahead of taking the office? When celebrating
his own consulate in 379, he seems to refer to Valentinian Is promise of the
appointment; that necessarily occurred before the latters death in 375.20
We know relatively little about how far in advance consulates might be
decided, but several contemporary instances identify an individual as consul
designatus a year in advance. In one extreme case, a future consul, Lollianus,
is called ordinarius consul designatus before May 337, when actually he only
became consul eighteen years later.21 In fact, the games sponsored by the
consuls cost a fortune; holders of the lesser office of praetor were designated
ten years in advance to allow them to assemble the wealth necessary to
pay for their praetorian games.22 Ausonius might well have appreciated a
substantial lead time to prepare to finance his own ordinary consulate, a
celebration which, in the sixth century, cost 2000 Roman pounds of gold.23
He was by this time well-to-do, but he had not been born to immense
wealth and surely was not close to the front ranks of the wealthiest Roman
senators.24 So, whether we take the consular insignia Ausonius imagined
in his future as referring to the ordinary consulate or merely to consular
honors, those lines need not have been written near 379.
Ausonius reference to Kirchbergs parched fields offers an opportunity
to consider a moment in the environmental history of the Roman Empire in
the light of climate science, and to add a new consideration to the debate
about dating the poem. In a small but tangible way, it exemplifies the new
evidence from the natural sciences that illuminates the ecology and history
of the ancient world. European dendroecologists have recently produced
a remarkable research resource, a record of spring and early summer pre-
cipitation in western central Europe. Its power derives from the fact that
differentpotentially competinglaboratories came together and pooled
their data of 7284 oak tree ring records collected in northeastern France
and northeastern and southeastern Germany.25 With a database of this size
and quality, the scientists could subject the growth records to statistical
mus, tamquam comitis ei semper fasces cum curulibus dederimus. Constantinople, Decem-
ber 13, 381.
20 Gratiarum actio, 22, addressing Gratian about why he might have deserved the con-
sulate, includes seu fideicommissum patris exsolvis; Green 149.31.

21 Bagnall et al. 1987, 1920; Jones et al. 19711992, 1, 513, Lollianus 5.
22 Bagnall et al. 1987, 18.
23 Procopius, Historia arcana, 16, 13, ed. Haury and Wirth 1963, 159.2427.
24 Green 1991, xxvi.
25 Bntgen et al. 2011.
procedures in order to reduce the noise arising from phenomena such as

the differing growth patterns of young and old trees, varying sample sizes
in different periods, etc.26 Because the annual growth patterns of oak trees
in the forests they studied reflect first and foremost how much rain the
trees got in April, May and June, the new data base allowed the group to
reconstruct the early summer precipitation in these areas for each year from
398bc to 2000ad.
The link between spring precipitation and tree ring growth was deduced
in the usual way, by comparing with actual ring widths the record of pre-
cipitation established since the advent of reliable instrumental recording
in these regions, in this case over the period from 1901 to 1980. Some have
challenged this kind of extrapolation from modern data, given the dramatic
changes the Industrial Revolution has introduced into the modern ecosys-
tem. Such concern proves unfounded here. A study of extreme excess or
shortfall in precipitation deduced from the tree ring records between 1000
and 1504ad confirmed the precipitation events from contemporary histori-
cal records in 32 out of 34 extreme years.27
The tree rings analyzed in northeastern France come from the region
immediately adjacent to the one that Ausonius describes.28 Figure 1 displays
the daily early summer precipitation as reconstructed from tree ring growth
for the years during which the poem must have been written. The new den-
droecological database shows that, in the region immediately adjacent to
the Hunsrck, two of the years in play, 371 and 375, were marked by sharp
shortfalls of spring and early summer rain, indeed by drought conditions.
If the thick web of contemporary allusions that Ausonius has woven into
his poem extended to the weather conditions of the journey he depicts,
then these two years should lead the candidates for the date of composi-
tion, regardless of the other deductions drawn from the poems content.
Since the strongest arguments have been adduced for composition in the
period 370371, in my eyes the new dendroecological data decisively tip the
balance toward 371. Quite unexpectedly, completely independent new data
from climate science reinforce arguments derived from the poems internal
characteristics and Roman political history that the work refers to and likely
stems from 371.
This convergence of ancient Roman poet and tree rings analyzed by
sophisticated scientific study represents a fine example of what E.O. Wilson
26 See Bntgen et al., Supplementary Online Materials.

28 See the map of sample locations in Bntgen et al. 2011, Figure 1.
Fig. 1. Reconstructed precipitation anomalies (mm/day), April, May, June, 367

378ad, northeast France. Source: Bntgen et al. 2011 with Supplementary Online
Materials, 9, and Fig. S4. Data on deposit at NOAA Paleoclimatology: http://www
.ncdc.noaa.gov/paleo/pubs/buentgen2011/buentgen2011.html.
calls consilience, a term coined in the 19th-century. Consilience occurs

when evidence of two completely different origins jumps together. Epis-
temologically distinct, the two different pieces of evidence come together
because the reality from which they stem is one. The independent find-
ings of dendroclimatologists and of classicists converge and reinforce one
another. Remarkable though this one case is, it is likely only the first of its
kind. The promise for future investigations on time scales useful to histori-
ans and archaeologists is even more exciting. But beyond that early summer
of 371, what light does todays climate science shed on more general environ-
mental conditions under the Roman Empire?
2. What We Know Today about Climate

Conditions under the Roman Empire
A group of climate scientists, archaeologists and historians met at Har-

vard Universitys Dumbarton Oaks in Washington D.C., to discuss what we
can know about climate conditions in the Roman and post-Roman world.
The result was a kind of white paper, a first synthesis of conditions across
the Roman space as they could be established from the latest available
scientific research.29 It used eleven high-resolution (that is, relatively pre-

cise in chronological terms) multi-proxy indicatorsindependent natural
archives such as tree rings, ice cores, pollen deposits in lake valves, and iso-
topic signals in speleothems exploited by modern climate scienceand our
new geodatabase of ancient written evidence about climate events to delin-
eate broad patterns of change and continuity. Here I will only summarize a
few salient features that our group detected in three broad phases between
ca. 100bc and 600ad. Although some have been claimed before, none have
ever been demonstrated with this rigor and relative comprehensiveness.
Even so, this is just a provisional balance sheet from an area of scientific
inquiry that is now undergoing exponential development.
First: an age of stability characterized Romes maximum expansion. Ex-
ceptionally steady climate conditions prevailed over the territory of the
Roman Empire from ca. 100bc to 150ad. Precipitation was fairly even and, in
the Levant, wetter than usual. The northwest provinces were exceptionally
warm, enjoying conditions perhaps comparable even to those of the 1990s.
Nile floods were unusually favorable, as we will see in more detail below.
Second: instability followed by partial recovery characterized 200 to
400ad. Stability began to dissipate starting about 150 ad. Broader cooling
occurred in the western provinces, as glaciers in the Alps ended a couple of
centuries of retreat and began growing again. Extreme volcanic eruptions
with their risk of rapid climate forcing and volcanic winters and summers
became more frequent, peaking ca. 250290.30 Spring and early summer pre-
cipitation declined sharply in northeastern Gaul from about 235 to 310 ad. In
the Empires eastern provinces, precipitation also declined significantly in
the third century, to judge from some written evidence and Dead Sea lev-
els. Then, as imperial fortunes recovered in the fourth century, Gaul became
wetter and, later in the century, its climate warmed; in the east, wetter con-
ditions returned sometime between 300 and 400 ad.
Two major climate developments that originated outside the Roman
Empire likely worked important and negative consequences inside it. Ex-
ceptionally dry climate conditions developed in Central Asia, possibly under
the influence of the ENSO (El Nio-Southern Oscillation) phenomenon.
These conditions probably encouraged in the later fourth century the migra-
tion into Europe of the nomadic federation known as the Huns. The move-
29 McCormick, Bntgen et al. 2012, where the details and substantiating materials will be
found.
30 Using the same GRIP-2 data, Rossignol and Durost 2007 have come to similar conclu-
sions on volcanic forcing.

ment into Europe of course triggered the Gothic crisis on the Danube.
This ultimately led to the destruction of the Roman army and emperor at
Adrianople that permanently established inside the Empire large groups of
Goths who escaped from imperial control.
The Niles annual floods reflect precipitation outside the empire in east-
ern Africa. We will return to the details in the next section, but a tabulation
based on Danielle Bonneaus study of Nile flood levels between 261 bc and
299ad argues that the early and later Roman Empire experienced different
qualities of Nile floods.31 Written records indicate that unusually favorable
climate conditions for Egyptian food production prevailed over the first two
centuries of the Roman Empire, while the conditions underpinning food
production appear to have been consistently less good from 156 to 299 ad.32
In a third phase, instability returned between 400 and 600 in ways that, on
todays data, appear to differ between the western and eastern parts of the
empire and its margins. In the west, signals are mixed but seem to indicate
fluctuating temperatures interspersed with periods of warming and cool-
ing. Two developments may have had particularly negative consequences
for agrarian production and the economy. The fifth century may have been
nearly as volcanically active as the third, implying considerable potential for
disruptive volcanic winters and summers regardless of the overall temper-
ature trend. Secondly, in northeast France and in northeastern and south-
eastern Germany, early summer precipitation was very wet until about 450.
Then it shifted to very dry and continued in this mode for the next two cen-
turies.33
In the eastern Roman Empire, on the other hand, generally humid condi-
tions resumed no later than about 400, as the eastern provinces approached
their political and cultural apogee.34 Dead Sea levels testify to a steep in-
crease in Levantine precipitation. That changes dramatically in the sixth
31 Bonneau 1971.
32 So far, Nile flood records are easily available or can be deduced for only eight years out
of the next three centuries that the Romans retained Egypt (300618ad): see McCormick,
Harper, et al. 2012 (on which see further, below, n. 44). Reconstructing for these centuries the
kind of Nile flood record that has been developed for the earlier period should be an urgent
priority.
33 For the precipitation record, see Bntgen et al. 2011.
34 Bookman (Ken-Tor) et al. 2004 differ in their datings from Migowski et al. 2006. The
discrepancy in the proposed starting date of the wetter conditions (and therefore of the
onset of the subsequent dry conditions) at ca. 300 or ca. 400ad lies well within the range
of the two-hundred-year accuracy of the radiocarbon dating of the samples at two standard
deviations (see especially Migowski et al. 2006, Appendix A), but both studies concur that
drying set in sometime between ca. 500 and 600 ad.
century when mentions of eastern droughts and heat events equal or exceed
precipitation reports in the written sources; Roman water works in Palestine
appear inordinately concentrated in the first half of the sixth century. The
summer water shortages that affected Constantinople in the 520s could also
reflect a decline in precipitation. More decisively, the two centuries of favor-
ably wet Levantine conditions documented by Dead Sea levels came to an
end at some point in the sixth century and, according to both key studies,
dry conditions persisted through the eighth century.35
Finally, and most spectacularly, written sources across the empire docu-
ment the seriousness of a veiling of solar radiation in 536 and 537 that caused
crop failures in different areas.36 The scientific proxy records appear to con-
verge with the historical evidence of the 536 event. Northern tree rings,
Greenland sea ice, oxygen isotopes all signal cooling that peaked around
540; summer temperatures dropped in the Alps, and a significant glacier
advance is proven in the Swiss Alps for the sixth century.37 Overall, the sixth
century also looks cooler in the post-Roman west, and difficult in the late
Roman east.
In sum, the initial indications suggest that the early centuries of the
Roman Empires existence occurred against the backdrop of remarkably sta-
ble and favorable climate conditions. Around 150 ad the stability ebbed from
a broad spectrum of climate indicators. Agricultural difficulties could well
have attended more fluctuating conditions, even as less favorable tempera-
ture and precipitation conditions became more common in some regions of
the empire. In the west, the fifth century may have been particularly unset-
tled, while the sixth century looks less positive in the east. In a general way,
this provisional sketch appears to fit the overall trend of imperial fortunes.
But it is only a first delineation at a time when new data and advances in
understanding the climate evidence and the mechanisms they imply appear
nearly weekly. Even though this description relies on the most comprehen-
sive survey of the available evidence, those data are sparsely scattered across
the territory of an empire that sprawled across three continents and cli-
mate zones. Further work and more data are urgently needed to correct and
improve this initial sketch by adding regional and chronological detail. The
appearance of very general correlations between imperial success or diffi-
culties and general climate trends does not yet decide the essential question
35For more details on these developments, see McCormick, Bntgen, et al. 2012.
36Gunn 2000; Arjava 2005.
37 For the tree ring evidence, see Larsen et al. 2008; for the rest, McCormick, Bntgen, et
al. 2012.
of cause and effect, of specific types of climate change in specific times and
places, and specific human responses. Let us turn next to what work and
data we need most and follow with two case studies of what we already have:
Nile flood qualities; and archaeological signs of human response to climate
change in the empires northwestern provinces.
3. Opportunities and Challenges for

the Study of the Ancient Climate
Whats next? First we need more, and more refined, data, preferably high res-
olution in chronological terms, and preferably from the underrepresented
imperial heartland: Italy, Spain, North Africa, the southern Balkans, Greece,
the Danubian provinces, and Asia Minor. In Italy, a wonderful first step
would be for the dendrochronological labs operating there to bring their
precious evidence to the quantitative and qualitative levels demanded by
international science by working together and pooling their data. That alone
will create the preconditions in Italy for the kind of environmental and
archaeological breakthroughs that are now occurring in northern European
dendroecology.
Valuable work has been done on lake deposits, but much more is needed,
and that work needs to supply for these centuries very high resolution
limnological data in order to be helpful to the environmental history of
the Roman Empire.38 Hopefully ongoing high-resolution work from Lake
Van and the Dead Sea will yield insight into the detailed climate situa-
tion in the Levant.39 If the chronological resolution issues can be addressed,
speleothems hold considerable promise. They have already yielded some
insights into temperature and precipitation conditions e.g., in Spain, the
Alps, Asia Minor, and the Levant, albeit with very uneven coverage for our
period; potentially very important work for the Roman Empire is ongoing
in Africas Atlas Mountains.40 Ice cores are another potential source. The
Greenland ice cores GISP2 and GRIP have given precious high-resolution
38 See, e.g., Allen et al. 2002.

39 For an introduction to the various kinds of palaeoclimatic proxy evidence mentioned
here, see McCormick, Bntgen, et al. 2012, Appendix.
40 For an overview of the Alpine, Turkish, and Israeli data and further references, see
McCormick, Bntgen, et al. 2012; although the published chronological resolution is at pres-
ent unsuitable for integration into Roman climate history, see for first indications from Spain,
Railsback et al. 2011, and for a summary initial report on ongoing work in the Atlas Mountains,
Wassenburg et al. 2010.
climate proxy data at the local and hemispheric scales. It would be very
desirable to obtain similar data from glaciers closer to the Roman Empire,
although there are of course precious few peaks over the ca. 4,000 m required
for permanent ice preservation in mid-latitude glaciers. Nor should we
neglect the climate data that still lies concealed in the written sources.
The Geodatabase of Historical Evidence on Roman and Post-Roman Cli-
mate now contains over 700 environmental events documented in writ-
ten sources from 100bc to 800ad. As our publications have appeared, we
have put the Geodatabase up on our free, internet-based Digital Atlas of
Roman and Medieval Civilizations (http://darmc.harvard.edu), so others can
use, add to, and improve it. We anticipate that a considerable number of
potential new pieces of evidence will emerge from climate-minded scrutiny
of ancient letters, scientific treatises, inscriptions (recording e.g., repairs of
flood damage), and the papyri.
But data alone do not take us very far. We need understanding too. Cli-
mate scientists and specialists of ancient environments face two challenges.
The first concerns chronology and dating; the second, understanding. Schol-
ars must pay careful attention to the dating methods and their precision
in various climate science investigations. For the tasks of historical anal-
ysis and explanation, high-resolution data are indispensable. If we really
do not know whether a particular climate event or shift began before or
after some human phenomenon that may be connected to climate develop-
ments, we cannot begin to claim causality. Historians work in years, months
and days. Climate scientists, however, often do important work on geologi-
cal time scales for which even millennia might be a flash in the pan. What is
for the historian low resolution data is prevalent in climate science studies,
and dangerous for historical reconstructions of ancient climates.
On the spectrum of precision, the kinds of proxy evidence climate scien-
tists use range from precise to very imprecise in historical and archaeologi-
cal terms. Examples of the former are yearly tree rings, absolutely annually
dated varve-countedcounting back from the year when the sample was
taken, as is possible in rare cases, or from some indubitably dated deposit
eventlake deposits, extremely well-preserved ice cores, or unambiguously
dendrodatednot radiocarbon datedglacier movements. Examples of
imprecise data are any phenomena which rely on radiocarbon dating, in-
cluding geomorphological data such as lake levels, soil transport and so on,
notwithstanding their potential contribution to multiple debates, such as
the controversy over Roman-era soil erosion. These must be considered with
caution.
Scientific methods such as calibrated A.M.S. (accelerated mass spectrom-

etry) radiocarbon dating offer a high level of reliability within the parameters
of their accuracy. This entails what is often, for historians, an inherent level
of imprecision that is very high. In most cases presently available, the high-
est confidence level attaches to radiocarbon datings expressed within two
standard deviations (the measure of dispersion of individual results from
the average result), that is to say, within an error margin that, for our period,
runs typically to over a hundred years. A confidence level of 95 % for a two
standard deviation calibrated AMS dating of 419591ad, for instance, means
that we can be assured that there is a 95% chance that the object sampled
dates to any single year within the 172-year period in question. 172 years is
not usually a date range that historians will find helpful. A newer method,
uranium-thorium dating, is generally credited with greater precision for the
calcium carbonate materials that are central to speleothems, but it does not
appear to be widely available.41
It is true that under certain circumstances, the radiocarbon date range
can sometimes be narrowed considerably, but here we must be wary of the
effects of what we might call the economic imprecision that occurs in some
studies using scientific dating. Radiocarbon dating is relatively expensive,
about $400 or more for a typical sample. I cannot even approximate the
cost of uranium-thorium dating. This leads scientists to limit the numbers
of samples that are actually dated scientifically, and to arrive at more pre-
cise dates by extrapolating precise dates of undated samples taken between
dated samples by calculating the physical distance between the dated sam-
ples. This works, but yields dates which can be very approximate, for it
assumes a regular, uniformly linear growth rate of the deposits that con-
stitute the speleothems, bog layer, or lake bottom (in the absence of dat-
able annual or seasonal varves). When one observes a climate phenomenon
dated only approximately in these ways, it is sometimes tempting to try to
integrate it into the much more precisely dated historical record by assum-
ing a kind of coherence which quickly can lead to circular reasoning and to
correlations or even assertions of cause and effect that amount to a house
of cards. These problems can be overcome with money, good samples, and
good methods, but it is well to be aware of them. We should receive with
circumspection the sometimes extravagant claims based on important and
interesting data whose chronological resolution will not sustain the claims.
41 This radiometric dating method measures age on the basis of the relationship of the
radioactive isotope thorium-230 and its parent isotope uranium-234. As of this writing, I am
unaware of any commercial service that will perform this dating.
We need not only to multiply the dating precautions, we need to multi-

ply the types of independent proxy evidence on which we base our climate
reconstructions. Multiple independent data sets of proxy evidencemulti-
proxy in the jargonare the best guarantee in these early days of fast devel-
oping science that the phenomena we detect are real climate developments
rather than artifacts of proxy evidence formation. In climate science as in
history, conclusions based on consilience, on multiple, independent lines
of evidence, are more robust.
As high resolution climate data from multiple sources and places rele-
vant to the Roman Empires environmental conditions become more abun-
dant, we shall have to join with our colleagues in climate science to under-
stand the new data qua climate data: how does each emerging piece of
the puzzle fit into bigger schemes of global and hemispheric climate sys-
tems, change, stability and instability? How does it fit into the regional
picture? How does it fit into the micro-regional picture? In coming years
we may even hope to define polygons across the Roman Empire whose
essential climate features and much of whose weather should share similar
features at similar dates. Yet the greatest challenge lies beyond establish-
ing and verifying the accuracy of our understanding of climate mechanisms
and proxy data. It comes with the daunting question of climate conditions
and human response. Let us turn to two such cases, one based on the his-
torical record of Nile flooding, and one which compares to archaeological
materials the new dendroecological record of northwestern European pre-
cipitation.
4. Nile Flood Qualities under

the Roman Empire: A Closer Look
Environmental conditions that were capable of affecting the Mediterra-

neans breadbasket must have weighed heavily on the imperial enterprise
and the Roman economy. More abundant floodsmore land inundated
with the Niles fertilizing watersgenerally mean more abundant harvests.
As we noted above, the Niles productivity may have undergone a subtle but
hitherto unnoticed change in the second century ad. At least that is what
emerges from a tabulation based on Danielle Bonneaus study of Nile flood
levels between 261bc and 299ad as evidenced by the papyri, coin issues, and
other records.42 Bonneau was able to identify records which allowed her to
42 Bonneau 1971. On Bonneaus oeuvre, see Bernand 1993.

Table 1. Nile Floods: overview of broader qualities as classified by Bonneau 1971.
Total docu- Good to Poor to

Period mented floods average floods % bad floods %
30bc155ad 112 72 64.3 40 35.7
156299ad 87 56 64.4 31 35.6
Source: McCormick, Harper, et al. 2012.
classify the quality of floods for 199 of the years from the Roman annexation
of Egypt in 30bc down to 299ad. Her treasure trove of data has never been
used for environmental history.43 Naturally imperfections are inevitable in
any such effort. Nevertheless a simple quantification of her data produces a
highly interesting series of observations that invite more intensive investiga-
tion and testing by specialists of the intricacies of the Egyptian documentary
record and of global climate systems. The data also suggest the extraordi-
nary untapped riches the papyri hold for the environmental history of the
ancient world.
At first blush, the proportion of positive floods compared to negative
ones as assessed by Bonneau looks virtually identical on either side of 155 ad
(Table 1).44 However, closer scrutiny of specific flood levels suggests a more
complex picture. Viewed in this way (Figure 2), the data (Table 2) indicate
that under the later Roman Empire a larger proportion (34.5 %) of annual
floods reached normally productive levels than had been the case (25 %)
before 156ad. Across the entire period under review, somewhat better than
normal floods occurred about once every five floods, that is, with the same
frequency before and after 155. The most significant differences concern the
best and worst floods, and therefore the best and worst years of agrarian pro-
ductivity. Between the annexation in 30bc and 155 ad, the very best level of
floods also occurred about 20% of the time: conditions allowing a superbly
43 McCormick, Bntgen, et al. 2012.

44 For the original data for the Roman period, see: Bonneau 1971, 231258. She starts her
record in 261 bc (221). Prof. Kyle Harper (University of Oklahoma), Dr. Alex M. More (Harvard
University), and I created a geodatabase of Nile floods from her flood data and, for the period
after 299 ad, from other materials. Since the studies founded on this material have been
published, we have made the Nile geodatabase available for others to use and improve as
part of McCormick, Harper, et al. 2012, at McCormick et al. 2010, at http://darmc.harvard.edu.
The descriptive values Bonneau assigns to the floods in these years have been numerically
coded as follows: M/mauvaise (bad): -3; F/faible (low): -2; Md/mdiocre (mediocre): -1;
N/normale (normal): 0.06 [this arbitrary value was assigned to make normal floods visible
in the graphing functions of Excel 2010]; B/bonne (good): 1; TB/trs bonne (very good): 2;
Ab/abondante (very high): 3. For her group of forte (strong, sudden), see next note.
Fig. 2. Percentages of Nile flood qualities, early vs. later Roman Empire.
See Table 2 for details.
Table 2. Detailed categories of flood qualities of the Nile according to Bonneau

1971.
Total Best floods Better than normal
floods (values 23) % floods (value 1) %
30bc155ad 112 22 19.6 22 19.6
156299ad 87 7 8.0 19 21.8
Normal Worst floods Sudden rise at

floods % (values -1 to -3) % some point (-0.5) %
30bc155ad 28 25 24 21.4 16 14
156299ad 30 34.5 27 31.0 4 4.6
Source: McCormick, Harper et al. 2012.
abundant harvest were recorded on average every five years. However, under
the later Roman Empire, similar conditions occurred only in eight percent of
recorded years, that is, on average, extremely favorable conditions for cereal
production occurred more than twice as rarely, only once every twelve and
a half years. Thus the early Roman emperors unwittingly enjoyed a positive
advantage in the conditions of Nile food production, quite independently
of the overall more favorable climate conditions that we have detected.
The early imperial advantage was probably only enhanced by the different
patterns of poor or failed floods that prevailed on either side of 155 ad.45 The
least favorable floodsand therefore poorest harvestsoccurred about as
often as extremely favorable floods down to 155ad, that is, about once every
five years. But after 155, poor or failed floods recurred almost every third year.
The new observation is highly interesting. It appears to accord with
the other, entirely independent evidence both for greater climate stability
in a basically favorable regime under the earlier Roman Empire, and for
increased climate instability in the first centuries of the later Roman Empire.
This raises questions of potential pattern shifting in global climate regimes
that should be of interest to climate scientists. The Nile datas apparent
evidence for a changing food supply for the Roman Empires cities and
armies holds considerable explanatory power. Precisely for this reason, it
will require detailed scrutiny from the vantage point both of the written
records and whatever proxy data and potential climate mechanisms can be
developed. In particular, I would like to spell out a little more explicitly than
perhaps is customary among historians the caveats that should attend this
proposition.
Two in particular will need more consideration before so consequential
an environmental change can be accepted as established, and integrated
into analyses of the Roman economy. The first caveat arises from the uneven
quality of the written record over this period. Bonneau classified the quality
of the records and the deductions she drew from them at three levels: those
in which she considered the evidence so explicit that the flood quality
can be considered as completely certain; the floods that she has deduced
with a level of confidence acceptable enough to leave her classification as
unqualified; and those for which the documentation left enough uncertain
45 One final group of entries has been left out of this discussion. That is Bonneaus category
Fo/forte ( quelque moment) (Strong [or sudden] (at some point)), the values for which
are supplied in Graph 2 and Table 2. These sudden floods could have been damaging: see
Bonneau, 1971, 6676. But this is not necessarily the implication, particularly if they did
not last long or dissipated quickly: Bonneau, 76. Thus the very sudden and strong flood
of 90 ad seems to have been reckoned a very good one: Bonneau, 238 with 156, n. 759
and on 263, Graphique IV; the sudden flood of 125 ad was commemorated by a Nile coin
celebrating an abundant grain supply: cf. Bonneau 1971, 242 and Bonneau 1964, 330331;
in 131ad, although the flood may have been satisfactory, the strength of the flood caused
destruction at Oxyrhyncus: Bonneau 1971, 243; in 141 ad, despite some signs of difficulties,
Elephantine recorded an excellent level, as did Alexandria: Bonneau, 245; cf. 183184. Out of
an abundance of caution, I classified such floods as negative in Table 2, at the value of -0.5.
They were more abundant in the earlier period (16 occurrences; 14%) against 4 (4.6%) in the
later period. Could this decline in sudden floods be another indicator of the phenomenon of
diminished floods that seems to prevail in the later period? The turning point around 155ad
seems clear from a graph of the floods: see McCormick, Bntgen, et al. 2012, Figure 10.
Table 3. Recording quality as assessed by Bonneau.

Years Some
Period documented Certain % Unqualified % doubt %
30bc155ad 112 28 25.0 33 29.5 51 45.5
156299ad 87 2 2.3 29 33.3 56 64.4
that she recorded doubt.46 As is not uncommon in ancient history, the top
level of certainty is always the exception (Table 3); as is also not uncommon,
that first level is more substantial in the earlier imperial period. Nearly a
quarter of all the earlier floods (28; 25%) are attested explicitly and therefore
with complete certainty, whereas that is the case only for a small minority
(2; 2.3%) of the later Roman floods. For about a third of both data sets,
Bonneau was satisfied enough with the evidence to leave her deduction
without further qualification, even if it fell short of the exceptional certainty
of her first group, as is, again, not unusual in ancient and medieval history.
Finally, as in any rigorous examination of ancient evidence, a substantial
portion of the deductions leaves room for some doubt: nearly a half of the
early imperial floods, and nearly two thirds of the later ones fall into this
category. The superior floods of the period down to 155, in other words, are
evidentially more secure than the poor ones after that date. But there is real
evidence for both phenomena.
The second caveat also pertains to the nature of the written record. Over
part of the later period in which the best quality floods appear rarer, and
poor floods more common, a new kind of document emerges, the declara-
tion of unflooded land (apograph abrochias). This new type of document
appears around 150ad and continues to crop up until 245. Why it appears,
and why only in this period, is not known. Papyrologists speculation on its
origins has mostly focused on a change in the mentality of documentary
practice or the possibility that such a document could procure tax relief.47
Could the appearance of the new type of document explain the quantitative
difference between early and late periods with respect to poor floods?
46 See Bonneau 1971, 217218; cf. ibid. 1416. Nevertheless the detailed presentation of her
cautious classification method inspires confidence in the overall reliability of her results:
Bonneau 1966.
47 Former explanation: Bonneau 1971, 183187; she also considers that it may have been an
effort by the emperor to make the farmer feel that he was participating in Egypts economic
life; tax relief: Praux 1963; good list but no explanation: Parassoglou 1987. I am grateful to
Roger S. Bagnall for his thoughtful comments and guidance on these records. Naturally he
bears no blame for what I make of them.
Probably not. Bonneau adduced this type of document for only eighteen
years between 157 or 158 and 239; six of those years had floods that Bonneau
rated as normal.48 Another flood she rated as strong or sudden, and certainly
so.49 Of the eleven years for which Bonneau uses the declarations to charac-
terize poor floods, all except two are documented by other types of records
as well, and one of the two also preserves a coin issued in connection with
the flood, usually a sign of an at least partially successful flood.50 At most
then, two additional poor floods of the total of 27 from this period could be
due exclusively to the appearance of a new type of record. This would not
change the overall picture. In fact, I cannot help but wonder if it were not,
on the contrary, a change in the environmentthe increased frequency of
poor floodsthat itself elicited the new type of document, as the tax author-
ities of Egypt were confronted with increasing appeals from farmers whose
fields had not received the flood necessary to produce their usual taxable
crop.
In sum, even with careful caveats, it appears that considered as an aggre-
gate, the data on Nile floods carefully assembled by Bonneau signal an
important shift in the reliability and overall productivity of the Roman
Empires Nile granary after 155ad. This certainly merits closer scrutiny by
papyrologists and by climate scientists considering the precipitation pat-
terns over eastern Africa and their possible global teleconnections; this new
insight may also shed light on the increasing role of Proconsular Africa in
supplying the city of Rome in the later empire. But these are questions for
another day. Let us turn, rather, to what may be Roman responses to climate
change documented in the archaeological record.
5. Is Tile More Comfortable Than Thatch? Climate Change

and Human Response in the Archaeological Record
Within the tangled, incomplete web of what we know about ancient and
medieval civilization, it can be terribly hard to observe and distinguish cause
and effect for events big and little. This is true whether we approach the
problem with the tools of the historian or of the archaeologist, or both. It is
true in spades when we turn to the complexities of climate change.
48 See Bonneau 1971, 247248, 249, 252253, 254, 255, respectively for the floods of ad157,
163, 189, 201, 203, 225, 239.

49 Bonneau, 252, ad 194.
50 Bonneau 1966, 384385.
The words climate change actually designate different types of change

that it is important to distinguish. Compared to short-term rapid change,
gradual long-term environmental shifts seem a priori more likely to gen-
erate graduated and successful human responses. Nevertheless, we must
recognize that context is everything: gradual change toward cooler temper-
atures may affect agrarian practices very differently in temperate areas of the
Roman Empire compared to marginal areas. For instance, at higher altitudes
even a slight shift in temperature might change definitively the growing sea-
son or the hardiness range. One easily imagines that rapid change such as
volcanic cooling could have had more immediately disruptive effects. Never-
theless, when those effects arose from short-term developments, they could
well have been cushioned by a society that had the capacity to seek alternate
food supplies until conditions returned to normal. We can expect that the
well-organized Roman transport sector will have been capable of address-
ing and overcoming some climate-induced famines, especially in years in
which the Nile produced abundant harvests in response to climate systems
outside the Roman Empire.
Distinguishing between trend and fluctuation captures another set of
differences. Trends may be faster or slower and we may surmise differing
consequences for slow versus swift shifts to new patterns, especially if the
changes were lasting. Fluctuation is different: it is repeated change back
and forth and it can be quick. One could imagine that for many kinds of
farms, short-term fluctuation would have affected planting, growing and
harvesting. This could have represented the most difficult type of climate
change for a society. To explore the historical problem of societal response
to climate change, we need to proceed on at least two levels. We need to
select examples of specific, securely demonstrated and understood climate
change with high chronological resolution in specific geographic regions,
and analyze in their light relevant historical and archaeological data, for
instance, about food production or water supply.51 Let me suggest some
questions for future research. Regarding exceptionally good or bad Nile
flood levels, can we detect responses within Egypt in the surviving, well
dated papyri: explicit responses such as price changes, labor shortages,
surges or declines in land transactions, and the like? Beyond Egypt, can
we observe direct or indirect signs of supply difficulties, for instance in the
capitals?52
51 See for example Schmidt et al. 2005; for a possible exploitation of climatic variation by
Roman engineers, see Schmidt 2010.

52 See the brief but suggestive comments of Bonneau 1966, 394395.
Another approach would be to investigate the impact of different types of

climate change on farming practices as documented by written records, the
archaeology of ancient farming technology and food preservation and stor-
age, and archaeobotany. Do the surviving agrarian calendars give any hints
of such adaptation? It might be worthwhile to examine in this light the fifth-
century Palladius in the western Empire, or the late antique agrarian writers
preserved in the Byzantine Geoponica.53 In any case, actual agrarian produc-
tion systems as they developed in antiquity and the Middle Ages must have
reflected the changing environment just as they are acknowledged to have
reflectedand inflectedchanging social, cultural and economic struc-
tures.54
Because it is destined to keep growing, the archaeobotanical evidence is
very promising. For example, that declining early summer precipitation in
northeastern France and Germany from about 230 to at least 315, and again
from around 425 to 650, affected the most sensitive season for many food
plants.55 Now palaeobotanical research has made clear that cultivation of
rye, normally sown as a winter crop, expanded in the late Roman period.
Scholars have suspected a link between ryes cold hardiness and possible
climate cooling; rye in fact needs a hard frost to be able to sprout, lessening
its value in warmer climates.56 However the new dendroecological evidence
shows two long periods of reduced spring precipitation. Could it have been
this particular abrupt and lasting climate change that elicited this human
agrarian response? In fact it has not been sufficiently appreciated that rye
53 Palladius Rutilius Taurus Aemilianus, Opus agriculturae, ed. Rodgers 1975; Geoponica,
ed. Beckh 1895.

54 Henning 2009.
55 Under modern climate conditions, the temperature and precipitation sensitivity of
winter grains peaks in western and central Europe during the heading phase that occurs
in May and June: United States Department of Agriculture, Foreign Agricultural Service,
Monthly Crop Growth Stage and Harvest Calendars, http://www.fas.usda.gov/pecad/
weather/Crop_calendar/crop_cal.pdf, consulted January 4, 2012. Cf. Food and Agricultural
Organization of the United Nations, Land and Water development division, Crop Water Man-
agement, Wheat, Water Supply and Crop Yield: http://www.fao.org/landandwater/aglw/
cropwater/wheat.stm#supply, consulted January 4, 2012.
56 For Roman rye and cold hardiness, see Henning 1987, 100. See in general the compre-
hensive and critical review in Behre 1992 and, on hard freezing and sprouting, ibid. 145.
More recently, for instance, a survey of carbonized fruits and seeds identified from rescue
excavations in Frances Aisne valley from the Neolithic to the early Middle Ages yielded
rye only in deposits dated to the third-fourth and fifth-sixth centuries ad: Bakels 1999, at
74, Table 1. Rye was found stored in the Roman fort at Dichin, Bulgaria from the destruc-
tion which occurred ca. 480. I am most grateful to Andrew G. Poulter for his kind shar-
ing of details of his excavation in advance of publication. In the meantime, see Grinter
2007.
enjoys a special ability to withstand spring drought. The new precipitation

record suggests to me that this drought hardiness may have driven ryes
emergence as a food crop at least as much as cooling (whose exact parame-
ters remain unclear) in late Roman times.57
I can easily imagine how such a shift might have occurred. Rye grains
first show up on archaeological sites in very small quantities and associ-
ated with other cereals.58 Palaeobotanists interpret this as signifying that rye
was present as a weed plant. Some farmers would surely have noticed if a
spring drought killed off all the wheat they had sown, but another adven-
titious cereal weedryecontinued to thrive in an otherwise devastated
field. Dire necessity could well have encouraged them to experiment with
the food preparation and cultivation of a new crop if they survived the first
bad harvest.59 Continuing archaeological investigation should deepen and
reinforce the link between changing precipitation patterns and a shift to cul-
tivating rye over more delicate forms of wheat. If so we have identified a spe-
cific and broad-ranging change in diet and the agrarian economy as human
response to climate change. A shift originating in the changing environmen-
tal conditions of the late Roman Empire will have shaped the spectrum of
cereals sown in medieval Europe, including the much-noted consequences
of ryes propensity to ergotism.60 And that late Roman climate change will
have continued to shape the German diet down to the twentieth century.61
Other case studies have already been adumbrated. They now need ex-
panding in the light of the new climate data. In the British Isles and on
the Continent, opinions continue to develop about the culinary or climate
implication of the late Roman proliferation of corn (grain) drying ovens.62
But what about the impact of cooling on clothing, domestic architecture and
living arrangements?63 Rooms may have gotten smaller to facilitate heating,
57 Ryes deep root system mostly explains this resistance; see Peltonen-Sainio et al. 2009, at
7778; for details, [University of Florida, the Food and Agricultural Organization of the United
Nations, and the National Museum of Natural History of the Smithsonian Institution], Eco-
port, Secale cereale, Description: Physiology, http://ecoport.org/ep?Plant=1929&entityType
=PL&entityDisplayCategory=full.
58 Behre 1992, 141, and 143 for Roman examples.
59 Although ancient farmers practice is of course only very imperfectly recorded, exper-
imenting with different types of crops is in fact explicitly documented in the Geoponica in
the light of astrological influences on crops and soil: Geoponica 2, 15, ed. Beckh 55.113; trans.
Dalby 2011, 82.
60 See Haller 1993 and Carmichael 1993.
61 Wheat supplanted rye as the dominant bread cereal in the Federal Republic of Germany
only in the 1960s: Krber-Grohne 1994, 43.

62 Compare Dark 2000, 8384 and Haas 2006, 256258, with further references.
63 For clothing, see Zabehlicky 1994, 465466 and below, n. 67.
whether by braziers which, because they were portable, are nearly invisi-
ble archaeologically, or by more expensive built in heating systems.64 The
famous Roman hypocaustsfloors and sometimes walls heated by hot air
from nearby furnacesoffer the advantage of being fairly well known and
archaeologically hard to miss. Conventional wisdom is that central heating
spread in the later Empire.65
It is certainly possible that centrally-heated hypocaust dwelling struc-
tures became more common in the Roman Empires northwestern prov-
inces in the third century. But that cannot be considered proven. One dif-
ficulty lies in the often imprecise dating of construction; another in judging
whether the proportion of villas with central heating increases in a given
region; and a third in discerning whether wealthy people began to build vil-
las with centrally heated rooms in response to newly colder temperatures,
to something more cultural such as the spread of a metropolitan fashion,
to increased disposable wealth, or to some combination of all these.66 It has
been argued that the strongest case for a connection between climate cool-
ing and the building of central heating systems may come from pre-existing
dwellings to which heating systems were added, and there are hints that
more resources were devoted to building higher performance heating sys-
tems in the northwestern provinces in the third and fourth centuries.67 If
true, this would accord with palaeoclimatic signals of cooling in central
Europe around 150 and particularly from ca. 200, albeit with some periods of
warming, especially ca. 365.68 For this argument to go beyond an impression
would require a systematic quantitative survey of all villas with or with-
out heating systems in a given region, with careful attention to dating and
also to possible changes in the mode of dwelling room sizes. Such a study
64 See the discussion in Haas 2006, 252256, with further references.

65 E.g., DeLaine 1996, 737.
66 For numerous late Roman heating installations in northern Gaul as well as problems of
dating, see, e.g., Van Ossel 1992, 128130.

67 Remodeling with heating systems: Haas 2006, 255; Zabehlicky 1994 argued that a vol-
canic eruption of 186ad caused a cooling of the Roman Empire that lasted into the third
century, and detects a cultural response in the form of warmer uniforms for Roman soldiers
(that, as he observes, at least in part antedate the eruption), and in an increase of the pro-
portion of villas with central heating. The latter argument is made provisionally and based
on a selection of 18 British and central European villas, chosen because he reckoned them
securely dated between the early second and late third centuries (466); naturally this can be
no more than suggestive. A few more possible indications from Ephesus are added in Vet-
ters and Zabehlicky 2001. For a suggestive overview of Roman built-in heating technology,
including experimental archaeology, see Schiebold 2010.
68 McCormick, Bntgen, et al. 2012.
could illuminate the capacity of a particular socioeconomic group to insu-

late themselves from the effects of cooling climate, and perhaps shed light
on differing patterns of climate change and human response in different
regions of the vast empire. A medieval analogue may turn out to be the
spread of fireplaces or stoves with chimneys in castles and lordly residences
as Europes medieval warming came to an end.69
The built sources of heat are but one aspect of the archaeological detec-
tion of human responses to climate change. How could cold peasants shiver
less in their hearth-heated houses in an age after the Roman braziers but
before the later medieval chimneys, fireplaces and stoves? Insulation is
key. Archaeologists and historians have lately made much historical hay, as
it were, from the early medieval replacement of Roman fired tiles by the
organic roofs of thatched straw. This has been depicted as prima facie evi-
dence of decline in physical comfort, and plausibly interpreted as signaling
an economy so deeply in reverse that it could no longer afford or organize
to bake clay roof tiles for its homes.70
But could changing climate have helped drive this change also? Anyone
who has lived, for instance, in a nineteenth-century Belgian workers cottage
with a kerosene space heater in the age before retrofitting with insulation
knows that its tile roofs and brick walls did not allow the heat from a liter of
kerosene to last long beyond its burning. Modern thatchers claim for their
roofs extreme thermal efficiency comparable to several inches of modern
fiberglass insulation. English Heritage confirms: Thatch has a much greater
insulating value than any other traditional roof covering.71 With respect to
durability, the same organization notes that some of the 50,000 thatched
buildings in modern Britain retain thatch over 600 years old. But that is
clearly very exceptional, when modern estimates for a thatch roofs lifespan
run between 10 and 50 years, depending on the type of straw and many other
considerations.72
69 Although the beginning of their diffusion, ca. 1000, may seem rather early in view
of the conventional wisdom on the medieval warm phase and the development of later
medieval cooling: Meyer 1995. For more details about the development of medieval fireplaces,
chimneys and stoves, see Sirot 2011. Again, rigorous quantification is needed to clarify the
evidentiary potential of medieval chimneys, open fireplaces and stoves.
70 Ward-Perkins 2005, 9496; cf. 110, where he recognizes the superior thermal quality of
thatch.
71 Ogley et al. 2010, 3; and 78. The report is available at http://www.english-heritage.org
.uk/content/publications/docs/eehb-insulating-thatched-roofs.pdf.
72 Quote: Ogley et al., 3. English Heritage 2000, 9, on the longevity spans of English
thatched roofs. This report is available at http://www.english-heritage.org.uk/content/

publications/docs/thatchandthatching.pdf.
So did the spread of early medieval thatchdetectible archaeologically

by the absence of roof tiles, the lighter roof structures it implies and, occa-
sionally, from the thatch itselfresult from economic collapse, or did it
represent something more complex? In the latter case, it may have reflected
changing economic structures and new climate conditions, that is, the likely
broader cooling phases presently detected from ca. 200 to 365, in the early
fifth century, and throughout the sixth century down to about 650; the old-
fashioned tiles would simply not have been worth the presumed extra cost
when they did such a poor job of countering the new climate conditions.
It may only be a happy coincidence in light of the spread of rye, but that
cereals stalks are supposed to make good thatch.73
To test this hypothesis will require examining roofing patterns in similar
dwellings within different regions of northwest Europe as climate science
clarifies the timing, duration, and seasonal and spatial reach of cooling. One
could continue, but the basic point should be clear. Ongoing and future
research in archaeology and climate science will furnish more and more
data that will allow us to investigate in detail whether and how change in the
natural environment forced changes on the built and farming environment,
and that will be the first step in a much longer and exciting archaeological
investigation into the interaction of climate, human health, economies and
culture.
This first reconnoitering of a large and complex theme provides a few pro-
visional conclusions. The era of consilient research is at hand. It draws on
the advances of the natural sciences alongside the humanities and social
sciences to address crucial questions of historical change. The Roman and
early medieval world offers archaeological and historical knowledge that
illuminates most remarkably those human societies and their changes. Now
the rapidly accumulating data of climate science is adding a new dimen-
sion to what is knowable about the environmental conditions of antiquity
and the Middle Ages. This new data can be brought to bear on traditional
questions, as we saw in the case of dating Ausonius Mosella, even as they
expand the range and breadth of investigations we can devise to deepen our
understanding of how Roman and post-Roman civilization interacted with
its natural environment.
Secondly, there are signals of rapid shifts toward favorable or unfavorable
climate conditions in different areas of the Roman Empire. These changes
73 Paillet 2005, 53.

may be correlated and even, to some degree, causative of archaeologically

documentable phenomena such as the spread of rye cultivation, built heat-
ing systems in dwellings, and thatched roofs. Some changes toward less
favorable climate conditions also coincided with periods of crisis in the
Roman Empire. Of course co-occurrence is not cause, even if it is a necessary
precondition for establishing causality. Today the evidence is insufficient to
claim that climate anomalies determined the course of empires and civiliza-
tion. Nevertheless climate change certainly was capable of playing a role in
the complex unfolding of human history by changing the background envi-
ronmental conditions of food production, animal and human health, and
in the seasonal patterns of daily life. Establishing the details will prove deli-
cate. But even short-term sudden climate anomalies such as the tremendous
volcanic winter of 536 must have had serious economic and human conse-
quences, including mental ones, although that event only weakened and did
not destroy the Roman Empire of Justinian. The task before us is to iden-
tify and date rigorously such moments, factors and actors in the growing
record of ancient and medieval environmental history. Only then can we
begin to understand their interaction and implications for our broader task
of explaining the lives of the women and men who have preceded us. In
broader terms, we will need to devise innovative ways of measuring climate
change to test and analyze its possible correlation with bigger developments
of political, social, and economic change. But for that measure to be truly
useful, we will also have to find novel ways of measuring political and eco-
nomic change in ancient and medieval civilization, and of correlating those
two measures.74 There is much to be done.
74 For recent efforts to create instability indexes for ancient Rome and the Byzantine
Empire, see Turchin and Scheidel 2009, with the (online) Supporting Information at http://
www.pnas.org/content/106/41/17276/suppl/DCSupplemental; and Preiser-Kapeller 2010, esp.
2934, on his instability index and related approaches to Byzantium, with further refer-
ences. The working paper is available at http://oeaw.academia.edu/JohannesPreiserKapeller/
Papers/506625/Complex_historical_dynamics_of_crisis_the_case_of_Byzantium. Cf. e.g., the
various indexes devised and deployed by Morris 2010b, 150169, etc.
MEGADROUGHTS, ENSO, AND THE INVASION OF
LATE-ROMAN EUROPE BY THE HUNS AND AVARS
Edward R. Cook
Introduction
The degree to which climate played a role in the invasion of late-Roman

Europe by the Huns in the 4th5th centuries ad and the Avars in the 6th cen-
tury ad has long been debated, e.g. from Huntington (1907) to McCormick,
Bntgen, et al. (2012). The importance of climate in shaping this critical
period of European history has been investigated in some detail using
millennia-long, precisely dated, annual tree-ring chronologies from central
Europe (e.g. Bntgen et al. 2011), but these superb records of past environ-
mental change provide no direct insights into what climate was like around
the same time in the home territories of the Huns and Avars in central Asia.
This limitation is difficult to overcome because of the paucity of millennia-
long tree-ring records directly from the central Asian steppes and western
China where these nomadic barbarian tribes are thought to have migrated
from. However, it is still possible to make some potentially useful infer-
ences and speculations about what climate may have been like based on
the limited tree-ring data available. This will be done through the presenta-
tion of a long tree-ring record from north-central China (Qinghai-Tibetan
Plateau) that spans this interesting cultural period. In so doing, we will
argue that prolonged periods of drought may have contributed to the migra-
tion of the Huns and Avars westward into Europe during late-Roman times.
In addition, we will investigate the long-range teleconnection between the
tropical Pacific Ocean El Nio-Southern Oscillation (ENSO) system and cen-
tral Asian climate as a possible contributor to the development of droughts
there.
Megadroughts in North-Central China
Presently, only two annual tree-ring chronologies from north-central China

are sufficiently long to tell us what climate may have been like in the home-
lands of the Huns and Avars at around the times of their westward
90 edward r. cook
Fig. 1. The Dulan-Wulan annual tree-ring chronology from north-central China and
and the occurrence of severe droughts during the times of the Hun-Avar migrations
into late-Roman Europe. Three multi-decadal droughts, among the worst of the
past 2,000 years, are indicated in the 4th, 5th, and 6th centuries ad at around
known times of invasion by these nomadic peoples from central Asia. The lower
plot illustrates this more clearly.
migrations. The first record is the millennia-long juniper (Juniperus przewal-

skii) tree-ring record from the Dulan-Wulan region in northeastern Qinhai
Province, which extends back over 2,000 years (Zhang et al. 2003; Sheppard
et al. 2004). It has been used to reconstruct annual precipitation (Sheppard
et al. 2004) and thus provides a useful reflection of local hydroclimatic vari-
ability there. This record, expressed as standard normal deviates, is shown
in Fig. 1 over its entire length of record and only for the ad 200700 time
period containing the Hun and Avar migrations for easier assessment. A sec-
ond multi-millennial tree-ring chronology produced by Shao et al. (2010) is
located about 100km west of Dulan-Wulan and is likewise an expression of
moisture variability and change there. It basically supports the story shown
in Fig. 1 during the Hun-Avar period and, therefore, will not be described
further here.
megadroughts, enso, and the invasion by the huns and avars 91
The Dulan-Wulan record (Fig. 1) reveals three periods of intense multi-

decadal drought during the Hun-Avar migration periods in the mid-4th, 5th,
and 6th centuries, with the first megadrought being the worst such event of
the past 2,000 years. The 4th century megadrought centered around ad 360
occurred at about the same time as the first migration of the Huns west-
ward into Roman Europe (McCormick, Bntgen, et al. 2012; Heather 1998).
It is conceivable that this period of intense aridity spurred the nomadic
Huns to seek better living conditions westward of their home territory to
as far as the eastern Roman Empire, with invasion and conquest a natural
part of this migratory process. The second major drought in the mid-5th
century occurred at around the time of the second Hunnic invasion of
Roman Europe in ad447451 (McCormick, Bntgen, et al. 2012; Heather
2000), once again suggesting that drought may have played a role in incit-
ing that invasion too. However, this period of conquest happened mainly in
the early stages of this megadrought, which suggests that it may not have
been the primary inciting mechanism. Interestingly, the 4th and 5th cen-
tury megadroughts are also separated by about 50 years of mostly above
average wetness. This pluvial period is likely to have produced better liv-
ing conditions for the Huns in their central Asian homelands, thus allowing
them to build up their capacity for the invasion and conquest in Roman
Europe. The final drought period indicated in Fig. 1 is centered on ad 550
at about the time of the Avars invasion of late-Roman eastern Europe
(McCormick, Bntgen, et al. 2012; Whitby 2000). While not as extreme as
the previous two megadroughts, this period of dryness may again have
incited the nomadic Avars to migrate westward in search of better condi-
tions and plundered wealth. And like the wet period preceding the inva-
sion of Roman Europe by the Huns, there is a similar period of above aver-
age wetness that preceded the invasion of eastern Europe by the Avars.
This again may have allowed the Avars to build up their capacity for inva-
sion.
In summary, the evidence presented here suggests that climate may have
played a role in the invasions of late-Roman Europe by the Huns and Avars.
This evidence is in the form of three multi-decadal droughts indicated in
the Dulan-Wulan tree-ring chronology at around the times of the Hun-Avar
migrations. Above average moisture supply may have also contributed to
the latter two invasions by improving the capacity for migration and inva-
sion by these nomadic pastoralists. It must be emphasized, however, that the
Dulan-Wulan record is from north-central China and is therefore unlikely
to fully represent what climate was like in the homelands of the Huns
and Avars on the central Asia steppes. However, severe droughts can cover
92 edward r. cook
very large areas (Andreadis et al. 2005). Thus, we suggest that these three
multi-decadal droughts could easily have extended into the steppes of cen-
tral Asia as well. Assuming this to be true, how might these persistent depar-
tures in moisture availability have happened? In the next section we will
present evidence for the long-range teleconnection between precipitation
over central Asia and China and the El Nio-Southern Oscillation (ENSO)
in the equatorial Pacific as a plausible mechanism for how the Hun-Avar
megadroughts occurred. This will be done through the use of two millennia-
long ENSO sensitive tree-ring chronologies from New Mexico in the USA and
the North Island of New Zealand.
ENSO as a Hypothesized Cause

of the Hun-Avar Megadroughts
The modern link between ENSO and drought in central Asia. The El Nio-
Southern Oscillation (ENSO) is the most important mode of internal climate
variability on earth (Rasmusson and Wallace 1983; Cane 2005). Its global
influence on climate was first described by Walker and Bliss (1932), and more
recently for precipitation and drought by Ropelewski and Halpert (1987),
Dai and Wigley (2000), Diaz et al. (2001), and Vicente-Serrano et al. (2011).
Physical mechanisms why precipitation patterns occur as they do under the
influence of ENSO can be found in Seager et al. (2005), and the robustness
of ENSO as an important tool for forecasting climate up to two years ahead
can be found in Chen et al. (2004).
So what is the modern day influence of ENSO on precipitation and
drought over central Asia? Walker and Bliss (1932) were unable to report on
anything there because of insufficient preciptation data over large areas of
central Asia at the time. Indeed, it is only since 1951 that one can readily
claim sufficient spatial coverage of precipitation stations over central Asia
for rigorously testing for the influence of ENSO on precipitation and drought
there (cf. Cook et al. 2010). Therefore, we will only report on correlations
between ENSO and preciptation over Europe and Asia since 1951. To this
end, the measure of ENSO variability we will use is the winter (DJF) season
Nio 3.4 sea surface temperature (SST) index from the eastern equatorial
Pacific, a region and season that are highly representative of ENSO (Tren-
berth 1997). The precipitation data set used is the 0.5 gridded GPCC global
land-surface monthly precipitation data set (Rudolf and Schneider 2005).
All correlations and the resulting maps shown were produced using KNMI
Climate Explorer (http://climexp.knmi.nl/).
Figure 2 shows the corrrelations between DJF Nio 3.4 SSTs and March
June GPCC precipitation data over Europe and Asia. They were calculated
using the original data online and after the data were first-differenced to
eliminate the influence of trend on the results. Only those correlations sig-
nificant at the 90% level of significance (p<0.10, 2-tailed) are shown. Positive
correlations indicate that warm SSTs (El Nios) produce wetter conditions
and cold SSTs (La Nias) produce drier conditions over central Asia and
China under modern day conditions. The MarchJune precipitation season
produced the strongest spatial pattern of significant correlations with DJF
Nio 3.4 SSTs overall. This is highly significant because moisture supply dur-
ing the spring months could be critical to the quality of livestock grazing and
the overall pastoralist lifestyle of the Huns and Avars. Correlations based
on both the original and first-differenced data both show a strong patterns
between ENSO and precipitation over north-central China and central Asia.
These results indicate that ENSO could have played a role in producing the
Hun-Avar megadroughts in the 4th6th centuries ad. The question now is:
what evidence do we have to more directly implicate ENSO that far back in
the past? Are there any proxy records of ENSO that extend back over the past
2,000 years? Next we will show that such records do exist and the appear to
lend further support for the ENSO/Hun-Avar megadrought hypothesis.
Paleo-estimates of ENSO variability during the Hun-Avar period. Two millen-

nia-long tree-ring chronologies from ENSO-sensitive regions in the North-
ern and Southern Hemispheres will be used here. The first is a Douglas
fir (Pseudotsuga menziesii) chronology from the El Malpais lava field in
northwestern New Mexico USA. This Northern Hemisphere record was pro-
duced by Henri Grissino-Mayer and the data are lodged in the International
Tree-Ring Data Bank (ITRDB; http://www.ncdc.noaa.gov/paleo/metadata/
noaa-tree-8517.html). Grissino-Mayer (1996) produced an annual (water-
year) precipitation reconstruction from this record in an area of the Amer-
ican Southwest known to be strongly influenced by ENSO (Ropelewski and
Halpert 1987). The Southern Hemisphere tree-ring chronology used here is
a Kauri (Agathis australis) record from the North Island of New Zealand.
These data were kindly provided for this study by Gretel Boswijk, Anthony
Fowler, and Jonathan Palmer, the primary developers of this chronology.
Subsets of the data used here can be found in the ITRDB (http://www.ncdc
.noaa.gov/paleo/treering.html), but the long multi-millennial record used
here is not yet publically available. Boswijk et al. (2006) describe the devel-
opment of this long record based on a combination of cross-dated tree rings
from living trees, archeological timbers, and sub-fossil wood. Fowler et al.
94 edward r. cook
Fig 2. Correlations between DecemberFebruary Nio 3.4 sea surface temperatures

(a measure of ENSO variability) and MarchJune total precipitation from 1951
to 2003. Significant (p<0.10, 2-tailed) correlations are indicated over central Asia
and China in approximately the regions where the Huns and Avars would have
come from. This illustrates the modern sensitivity of precipitation there to ENSO
variability, with warm eastern equatorial Pacific sea surface temperatures (El Nios)
producing wetter conditions and cold sea surface temperatures (La Nias) in the
same region producing drier conditions.
Fig. 3. Two 2,000 year long annual tree-ring chronologies from ENSO sensitive regions in the
Northern and Southern Hemispheres: Douglas fir from northwest New Mexico and Kauri from the
North Island of New Zealand.
(2000 as well as Fowler 2008 and Fowler et al. 2008) provide the case for the
sensitivity of Kauri to ENSO. The exact cause for this sensitivity is not well
understood, but as we will show here it appears to be highly robust.
Figure 3 shows these chronologies in standard normal deviate form. Each
of these records is 2,000 years long and is thought to express the effects of
ENSO variability and forcing on climate over the regions where the trees are
growing. In order to test this hypothesis, we used KNMI Climate Explorer to
produce maps of correlations between these tree-ring records and Hadley
Centre HadISST1 gridded global SSTs (Rayner et al. 2003) for the winter
(DJF) season from 1871 to 2003. This includes the exact region in the eastern
equatorial Pacific where the Nio 3.4 ENSO index is estimated from the SSTs.
Figure 4 shows the results of these correlation analyses, with the locations
of the chronologies indicated by red stars. In each case, the correlations
are shown (p<0.10, 2-tailed) based on the original data, after the data were
detrended, and after the data were first-differenced. This was done to test
the statistical robustness of the putative ENSO signal in each chronology.
A meaningful ENSO signal should result in a robust pattern of signficant
correlations between tree rings and SSTs in the eastern equatorial Pacific
and this pattern should be relatively insensitive to the way the data were
96 edward r. cook
Fig. 4. Correlations between the Douglas fir and Kauri tree-ring chronologies and Hadley Centre
global sea surface temperatures (HadISST1) for the winter season: 18712003. The locations of the
tree-ring chronologies are indicated by red stars. Only regions with correlations significant at
the 90% level (p<0.10, 2-tailed) are shown. A meaningful ENSO signal should result in a robust
pattern of signficant correlations between tree rings and SSTs in the eastern equatorial Pacific,
with additional correlations found in the Indian Ocean sector, and this pattern should be relatively
insensitive to the way the data were treated. In each case, a robust ENSO signal is indicated
regardless of how the data were treated for presence of trend.
Fig. 5. The average (A) and difference (B) of the Douglas fir and Kauri tree-ring chronologies. If a
true ENSO signal of the same kind exists in both, as suggested by Fig. 4, then the average should
preserve that ENSO signal as well and perhaps even strengthen it. The difference between the
chronologies is a test for an interhemispheric gradient in ENSO variability.
treated. This is exactly what was found. In every case, a robust ENSO signal
over the eastern equatorial Pacific is indicated in both the Douglas fir and
Kauri chronologies located on opposite sides of the equator. While some-
what expected for Douglas fir in New Mexico based on known ENSO/climate
teleconnections (Ropelewski and Halpert 1987), the Kauri result is equiva-
lently strong, which validates the claims of Fowler et al. (2000; Fowler 2008
and Fowler et al. 2008).
The robust ENSO signal in each tree-ring chronology suggests that an
even better result might be found in the average of these two records.
Testing the average is also a stringent additional examination of ENSO signal
robustness because if a true ENSO signal is present in each, it should be
preserved or even strengthened in the average. Figure 5a shows the average
of the two tree-ring chronologies and Fig. 5b shows the difference. The latter
is included here to test for a possible interhemispheric gradient in ENSO
variability as well. These two series were used in another correlation analysis
with SSTs of the kind done previously for the individual chronologies, and
their results are shown in Fig. 6. The average series has an ENSO signal
in it that is at least as strong as any of the individual chronologies. This
98 edward r. cook
Fig. 6. Correlations between DecemberFebruary average SSTs and the average and difference of
the Douglass fir and Kauri annual tree-ring chronologies, each with an identified ENSO signal. The
locations of the chronologies used in the average and difference are indicated by red stars. The aver-
age has an ENSO signal that is at least as strong as any of the individual chronologies. This suggests
that this interhemispheric signal is real and not a statistical artefact. The lack of any meaningful
correlations in the chronology differences indicates no gradient in the interhemispheric ENSO sig-
nal.
Fig. 7.Comparisons of correlation patterns of MarchJune precipitation with actual

and tree-ring ENSO indices. The top figure shows the correlations with the actual
DJF Nio 3.4 ENSO index. The bottom figure shows the correlations with the tree-
ring ENSO index based on the average of the Douglas fir and Kauri chronologies.
While somewhat weaker, the pattern of significant correlations (p<0.10) in central
Asia based on the tree-ring index is essentially the same as that based on the actual
Nio 3.4 index.
100 edward r. cook
Fig. 8. The Douglas firKauri average interhemispheric ENSO index with drought
inducing La Nia periods indicated around the times of the Hun-Avar migrations
into late-Roman Europe. The lower plot illustrates this more clearly. The timing of
these drought inducing periods is not one-to-one with the indicated megadroughts
in the Dulan-Wulan record from north-central China (Fig. 1). However, the general
timing of these drought inducing La Nias overlaps sufficiently well to suggest that
ENSO did in fact contribute to the megadroughts that may have incited the Huns
and Avars to migrate west and invade late-Roman Europe.
suggests that there is a true interhemispheric ENSO signal in these series

that might tell us what ENSO variability was like during the Hun-Avar
megadroughts. In contrast, the lack of any meaningful correlations in the
chronology differences indicates that there is no interhemispheric gradient
in the strength of the ENSO signal.
Given the strong ENSO signal in the average tree-ring series, the logical
next question to answer is how well this tree-ring ENSO index produces
the pattern of corrrelations found in Fig. 2 between the actual DJF Nio
3.4 ENSO index and MarchJune precipitation over central Asia and parts
of northern China. Figure 7 answers this question in a generally affirma-
tive way. The pattern of positive correlations in central Asia based on the
tree-ring ENSO index (Fig. 7b) matches those based on the Nio 3.4 index
(Fig. 7a) reasonably well. The strength of the tree-ring ENSO index correla-
tion pattern is weaker overall, but the geographic location of those corre-
lations matches the Nio 3.4 pattern in central Asia quite well. This results
means that we can use the multi-millennial tree-ring ENSO index as a rea-
sonably accurate surrogate for ENSO variability during the Hun-Avar migra-
tion periods and use it to infer what the MarchJune moisture conditions in
central Asia were like during those times.
So what does this 2,000 year long proxy of ENSO tell us about the causes
of the 4th, 5th, and 6th century megadroughts in central Asia and China?
Figure 8 indicates that unusual periods of drought inducing La Nia-like
conditions occurred around the times of the Hun-Avar migrations. The tim-
ing of these drought inducing periods is not one-to-one with the indicated
megadroughts in the Dulan-Wulan record from north-central China (Fig. 1).
This is not unexpected given that the Dulan-Wulan record itself is proba-
bly a locally biased expression of drought variability over central Asia and
the tree-ring ENSO index is not perfect either. However, the general timing
of these drought inducing La Nia-like conditions overlaps sufficiently well
with the megadroughts at Dulan-Wulan to suggest that ENSO did in fact con-
tribute to their development. Consequently, megadroughts caused in part
by ENSO are viable hypotheses for explaining what might have incited the
Huns and Avars to migrate west and invade late-Roman Europe.
Concluding Remarks
A hypothesis has been presented here for a climate mechanism that could
have incited the Huns and Avars to migrate west and invade late-Roman
Europe in the 4th, 5th, and 6th centuries ad. This hypothesis uses a millen-
nia-long tree-ring chronology from Dulan-Wulan in north-central China to
show that a series of megadroughts probably occurred around those times in
the homelands of the Huns and Avars. The cause of these megadroughts is
then shown to be plausibly related to long-range ENSO forcing of climate,
which influences MarchJune precipitation amounts in central Asia and
northern China in the modern era. Two multi-millenial ENSO-sensitive tree-
ring chronologies from New Mexico and New Zealand support this hypoth-
esis by indicating that persistent drought-inducing La Nia-like conditions
occurred around the time of the megadroughts and Hun-Avar migrations.
To verify these hypotheses, more multi-millennial paleo records of hydrocli-
matic variability in central Asia are needed.
102 edward r. cook
Acknowledgements
This research presented here was stimulated by an invitation to give a

talk at the Dumbarton Oaks Workshop of Climate Change under the late
Roman Empire, April 2425, 2009, organized by Zachary Smith (aka Mike
McCormick). I thank Henri Grissino-Mayer for his remarkable El Malpais
Douglas fir tree-ring data and Jonathan Palmer, Gretel Boswijk, and Anthony
Fowler for generously allowing me to use their amazing Kauri data, much
of it unpublished. Lamont-Doherty Earth Observatory Contribution Num-
ber 7700.
THE ROMAN WORLD AND CLIMATE:
CONTEXT, RELEVANCE OF CLIMATE
CHANGE, AND SOME ISSUES*
Sturt W. Manning
Introduction
This essay seeks to develop what is now a burgeoning discussion on the

role and relevance of climate for the history of the Roman world (espe-
cially McCormick, Bntgen et al. 2012 and literature cited; Bntgen et al.
2011).1 The essay begins by reviewing some of the background issues, ques-
tions of focus (timescales and resolution), contradictions, and problems
which are all inherent when investigating climate context in the Mediter-
ranean region in the later Holocene (last few thousand years). It then con-
siders the main forcings on climate over this recent and relatively short
timescalenamely, the sun and volcanic eruptions (Shindell et al. 2003;
Wanner et al. 2008, 18021804; Wanner et al. 2011, Figure 5)and reviews
what records of changing solar activity tell us for the Roman period, and
the parameters these records suggest for climate and climate change in the
Roman period. To address some specifics, I then review some of the high-
resolution climate archives available and relevant to the Roman world (from
tree-rings and from speleothems), before integrating some of these observa-
tions with some of the other published data for climate and environmental
* I thank William Harris for inviting me to the American Academy in Rome, and for orga-
nizing a very enjoyable event including even a lunar eclipse, and for his editing skills. I also
thank Michael McCormick, both for an invitation to an earlier workshop at the Dumbarton
Oaks which led to focus on climate and the Roman world, and for his comments on a draft
of this text. I thank Jrg Luterbacher, Eleni Xoplaki, Charlotte Pearson and Catherine Kearns
for comments and reviews of drafts at various stages, and thank them very much for making
many good suggestions and improvements. The remaining errors and mistakes are by the
author.
1 In this essay the term Roman world refers approximately to the time period from the
3rd century bc to the end of the 7th century adso around about a millennium. In terms
of space, I mainly refer primarily to the Mediterranean area including the Levant. At various
times central and northern Europeincorporated into the High Roman Empireare also
mentioned, but usually as contrasts with the general Mediterranean zone.
104 sturt w. manning
change in the Roman Mediterranean. Some patterns and implications are

noted; some contradictions and problems are highlighted. At present we
may usefully, but only generally, begin to describe the climate context of
the Roman worldmore data from a wider spatial network, and especially
good chronological control on these data, are essential to overcome current
contradictions and uncertainties.
Some Issues
Climate is complex, and depending on the scale of resolutionwhether in

terms of geographical area or temporal precision (and longer-term trends v.
short-term data)and place of observation, different and even apparently
contradictory observations may be made. Climate forcings may have varying
effects on human history, depending on a societys environmental setting,
social, economic and political structures, and those of its neighbours. As
Trigger (2003, 279) notes, there is no basis for theories that attribute the rise
of civilization to the influence of a single type of environment or climatic
event. The correlation or path from climate to human history, if present,
is instead a complex, multifaceted, multiscalar interaction contingent on
social context and human agency (e.g. Rosen and Rosen 2001; Fisher et al.
2009; Wossink 2009). Rare instances of major stress and challengetrigger
pointsare sometimes easier to diagnose than stability or opportunity.
Within the generally rather mild and stable Holocene era (an interglacial
period) of the last ca. 12,000 years, some key periods of significant or rapid
climate change (linked primarily to ariditywater being the key to life in
temperate to hot loci) are widely noted as relevant (at least) to SW Asia and
the Mediterranean region in prehistory (e.g. episodes ca. 6200bc/8200bp,2
3200bc/5200bp, 2200bc/4200bp, and 1200bc/3200 bp), and these seem to
translate via various processes into historical change-point periods of rel-
evance to prehistory (e.g. Weiss 2000; Weiss and Bradley 2001; deMenocal
2001; Chew 2002; Staubwasser and Weiss 2006; Weninger et al. 2009).
But even these few now well-known episodes, which have become almost
standard issues to mention in the archaeological literature, are often not
entirely without problems if examined critically. Looking to an earlier period
2 bc = Before Christ or Before Common Era (bce); bp = Before Present (which in radio-
carbon contexts refers to dates before ad 1950; but in some other palaeoclimate contexts it
refers to dates before ad 2000, or even other dates stated by authors); ad = Anno Domini or
Common Era (ce).
the roman world and climate 105
before, around, and following the development of agriculture (23,000 to

8,000 years Cal bp), where climate has long been held to be some sort of
prime-mover, the study of Maher et al. (2011) offers an excellent example
of this: they find that there is in fact a lack of close correlation between
rapid climate change episodes and changes in the archaeological record.
Turning to the later Holocene, the review and discussion of Finn et al. (2011,
31633164, 3166) with regard to the 2200bc event is another good example:
they highlight that this event is only clear in a few records, and not really
present in others, and thus the events critical status may be over-played
(see also Wossink 2009). Yet the same study points to another fundamental
issue with respect to trying to investigate past climate regimes. The data
they review are of very different types, qualities, and dating resolutions.
Some of the differing if not contradictory findings which they note may
indicate regional variations, but many reflect data derived from different
types of archives formed by various (often unique) processes, with a range
of complications (some natural, some likely anthropogenic, some a mixture
of both in recent millennia). The end result is considerable noise, and either
almost nothing historically useful is revealed or one is left with very general
conclusions which all but overlook any shorter, even century-scale events,
such as:
46001400 yrs bp: Drier conditions mainly dominate the climate picture of the
eastern Mediterranean, but there are periods of increased moisture creating
more benign climate situations at times, e.g. 16001200 yrs bp.
(Finn et al. 2011, 3169)
And one can note studies cited in the Finn et al. (2011, 3164) review which
suggest contrary findings for some of this period. Further, in terms of the
topic of this essay, concerning Rome and Climate, the 3200-year interval
of 46001400 yrs bp, considered as a vast whole, entirely fails to consider
the shorter-term but potentially key changes, such as those in the 3rd7th
centuries ad argued to be of considerable relevance by the recent study of
Bntgen et al. (2011).
If there can be problems agreeing on instances of regionally-relevant
major negative or stress episodes, then it is almost inevitable that it will
be even harder to identify instances of potential climate opportunity for
particular regions within the overall fairly benign Holocene era (where,
with the exception of the 8200bp eventRohling and Plike 2005, there
are no really dramatic rapid climate change episodes in contrast to the dra-
matic interglacial/glacial changes to be observed over the previous 100,000+
years). It is more likely we shall be able to identify episodes of climate
stress (e.g. drought or substantial reduction/increase in temperature)for

example when a tree all but dies because of water deficit or coldwhereas
the response to improved conditions is never quite so clearly represented
(Hughes 2002, 103). Instances of positive conditions may well have only
regional, versus hemispheric/global, impacts, and they may be expressed
as general changes, or even a stable period, and not an easily spotted short-
term spike (Caseldine and Turney 2010, 90). It is also important to remember
that climate impacts are as relevant to given regions, and the same climate
forcing may be both negative and also positive in different areas and under
different circumstances.
Despite the additional element of massive anthropogenic inputs in the
last couple of centuries (making this now the Anthropocene era on Earth
after Crutzen and Stoermer 20003), the differential impacts, perceptions and
challenges of modern global warming around the Earthand the ensu-
ing debates and uncertaintiesoffer a good example of the multiple con-
temporary scenarios which can apply at different scales and in different
regions, and the complications involved in any attempt to write a single
large regional, hemispheric, let alone global, narrative (Watson et al. 1998;
McCarthy et al. 2001; Parmesan and Yohe 2003; IPCC 2007; Cline 2007).
In pre-modern times significant global warming and cooling events are
another clear example of differential impacts according to scale and area,
with the Medieval Climate Anomaly (MCAsee Diaz et al. 2011) and the
Little Ice Age (LIAgenerally, see Grove 1988; Fagan 2000) offering our best
attested evidence. For example, at a large area and multi-century scale, the
warming MCA sees a sustained period of a positive North Atlantic Oscil-
lation (NAO, or Arctic Oscillation = AO4), and drier conditions in the west
Mediterranean but wetter conditions in the eastern Mediterranean, and the
cooling LIA sees the opposite with a sustained negative NAO phase with
3 Ruddiman (2003), however, argued that anthropogenic climate forcing started with the
origins of agriculture. This hypothesis and the ensuing debate are discussed in a set of papers
in a special issue of The Holocene (Ruddiman et al. 2011). Certini and Scalenghe (2011) make an
attractive case for a date for the start of the Anthropocene about 2000 years ago (mid Roman
period) when anthropogenic soils can first be recognized widely as a significant feature.
4 The AO (Arctic Oscillation) encompasses the NAO (North Atlantic Oscillation), and
they are closely related. The AO is regarded as the more fundamental and wide-ranging cli-
mate structure by those who proposed the term (Thompson and Wallace 1998), whereas they
regard the NAO as regional, and an historical accident deriving from the greater information
available regarding the north Atlantic. I employ the term NAO primarily as it has greater cur-
rency in much of the literature cited, Mediterranean climate relates to the Atlantic, and it is
almost indistinguishable from the AO and vice versa (Deser 2000).
wetter conditions in the west Mediterranean and drier conditions in the east
Mediterranean (Roberts et al. 2012; Kaniewski et al. 2011). But, within this
macro pattern, there are then important regional and shorter time period
subtleties (see also further discussion below). For example, the study of
Shindell et al. (2001) nicely shows how the effect of a major cooling episode
(solar driven) like the Maunder Minimum (one of the two grand solar min-
ima of the LIA) plays out in quite different ways. Northern Europe cooled
quite significantly, but the central and east Mediterranean and SW Asia were
either stable or even slightly warmer, and especially over the winter period
which is in turn key to crop growth in this region (see Shindell et al. 2001,
Figures 1 right and 3). And despite the generally drier LIA multi-century
period in the east Mediterranean, the specific 50-year period ad 17001750
after the Maunder Minimum episode sees sustained increased precipitation
across much of the Mediterranean including many areas of the east Mediter-
ranean except part of central Anatolia (Luterbacher et al. 2006; Nicault et al.
2008). Such changes or non-changes are only evident according to the region
and scale of observation and analysis. As the Shindell et al. (2001) paper
highlights, despite the temperature drops in Europe, the average global tem-
perature change through the Maunder Minimum period was in fact very
smallthus the global or hemispheric average may have little bearing on
observations in a specific region, and vice versa. Altogether, as the critique
of Caseldine and Turney (2010) notes, we need attention to chronology and
resolution, appropriate and often locally-based data, and new ways of inte-
grating palaeoclimate data with historical and archaeological sequences
and stories.
In much of the Mediterranean, where lack of water is the key risk for
agriculture and so for human food supply (see e.g. Garnsey 1988; Issar and
Brown 1998; Mithen and Black 2011), precipitation response is the more vital
statistic, and I concentrate mostly on water availability in this essay. For the
recent period Hurrell (1995; 1996; Hurrell and van Loon 1997) demonstrated
how the state of the North Atlantic Oscillation (NAO) relates to European-
Mediterranean climate. During the period of instrumental records, the NAO
has strongly influenced inter-annual precipitation variations in the western
Mediterranean, while some eastern parts of the basin have shown an anti-
phase relationship in precipitation and atmospheric pressure (see also Trigo
et al. 2000; Xoplaki 2002, Figure 6.1; Hughes et al. 2001, esp. p. 71; Roberts et
al. 2012). In a recent study, Roberts et al. (2012) explored how the NAO and
other atmospheric circulation modes operated over the longer timescales
of the MCA and LIA using high-resolution palaeolimnological evidence
from opposite ends of the Mediterranean basin, supplemented by other
palaeoclimate data. Iberian lakes show lower water levels and higher salin-
ities during the 11th to 13th centuries ad synchronous with the MCA, and
generally more humid conditions during the LIA (15th19th centuries ad)
(Roberts et al. 2012). This pattern is clearly evident in tree-ring records from
Morocco (Esper et al. 2007) and from marine cores in the western Mediter-
ranean Sea. In the eastern Mediterranean, palaeoclimatic records (Turkey,
Greece and the Levant: see Roberts et al. 2012 and references therein) indi-
cate generally drier hydro-climatic conditions during the LIA and a wet-
ter phase during the MCA (Roberts et al. 2012). This implies on longer
timescales that a bipolar climate see-saw has operated in the Mediter-
ranean for the last 1100 years. However, while western Mediterranean aridity
appears consistent with a persistent positive NAO state during the MCA,
the pattern is less clear in the eastern Mediterranean (Roberts et al. 2012).
Results indicate that the long timescale LIA/MCA hydroclimatic pattern in
the Mediterranean was determined by a combination of different climate
modes along with major physical geographical controls, and not by NAO
forcing alone, or, alternatively, that the character of the NAO and its tele-
connections have been non-stationary (Roberts et al. 2012, 3031).
But perhaps the most striking observation to come from the Roberts et
al. (2012) paper is that there appears to be a key difference according to the
timeframe of study. They observe (2012, 3031) that on shorter timescales
(annual to decadal) there is in fact no opposition between the west Mediter-
ranean (Iberia) and the east Mediterranean (Anatolia), and that studies have
usually found a general correlation of a negative winter NAO (and especially
a negative AOsee Unal et al. 2012, 402403) with increased precipitation
in the Aegean region and western and some other areas of Anatolia (but
less so eastern and southeastern Anatolia and on the Black Sea coast) and
other parts of the east Mediterranean, and the reverse (e.g. Unal et al. 2012
and references cited therein; Trke and Erlat 2003; Xoplaki 2002, Figure 6.1;
Hughes et al. 2001, esp. p. 71). It is only when the timeframe of observation
and analysis moves to century and greater scale that the opposite pattern
of general aridity then links with the sustained overall negative NAO period
of the LIA in the east Mediterranean (contrary short-term associations of
opposite character), and the generally positive NAO period of the MCA links
with overall increased precipitation in the east Mediterranean. Hence there
are two different patterns depending on the timeframe of study.
We may plausibly assume a fairly similar regime for the later Holocene,
roughly the last 6000 years, and in contrast to the earlier Holocene when evi-
dence indicates a wetter (and so different) climate regime for the Mediter-
ranean (Robinson et al. 2006; Brooks 2006; Wanner et al. 2008; Black et al.
2011). However, the appreciation of a differential relationship between the

NAO index and precipitation in the Aegean-Anatolia-east Mediterranean
region according to the timescale of observation/analysis, creates two pos-
sible scenarios and so may affect some previous studies which have simply
assumed a (single) stable linkage between negative NAO and precipitation
for the east Mediterranean, such as that by Lamy et al. (2006), who found (or
assumed) a correlation between drier conditions in the east Mediterranean
and a positive NAO, and between wetter conditions in the east Mediter-
ranean and a negative NAO (or AO). Over the decadal scale, yes, but at the
longer century scale then perhaps no following the findings of Roberts et
al. (2012), or Kaniewski et al. (2011). And between the opposite shorter-scale
and longer-term NAO associations, a variety of other forcings may also be
relevant (Roberts et al. 2012, 3031). In particular, we may note again that,
despite a generally cooler, drier LIA and overall more negative NAO across
this period (Trouet et al. 2009), during the half century (ad 17001750) which
followed the Maunder grand solar minimum (ca. 16451715)the major
cooling episode of the past five centuriesLuterbacher et al. (2006) and
Nicault et al. (2008) observe a significant switch to wetter conditions in the
Mediterranean generally (except parts of central-east Anatolia), and Nicault
et al. (2008, 240) find this half-century to form the longest stable wet period
observed in their 500-year record. And, generalizing from the Lamy et al.
study (2006, esp. Figure 8 and pp. 810), there is a plausible association of
wetter conditions in the periods around major solar minima.
Figures 1a and 1b illustrate some of the patterns, complications, and con-
tradictions from a hemispheric scale to a regional scale (west versus east
Mediterranean). The warmer v. cooler MCA v. LIA sees a general shift in the
Palmer Drought Series Index values from drier to wetter in Morocco (south-
west Mediterranean), but the precipitation record from the east Mediter-
ranean (Touchan et al. 2005) offers often an anti-correlation versus the
Moroccan record (see also Figure 15 below).
Comparison of the reconstructed winter NAO by Trouet et al. (2009)
and Cook et al. (2002) versus some decade or longer periods of more neg-
ative (drier) PDSI or more positive (wetter) PDSI observed in the analy-
sis of Mediterranean tree-ring series ad15002000 by Nicault et al. (2008)
for Italy, Greece and the Levant in Figure 1b further highlights differences
and patterns. The longest period when all these regions (and much of the
Mediterranean) enjoyed positive PDSI values and increased water availabil-
ity, ad17001750, correlates neither with a noticeable positive or negative
NAO; indeed, it is if anything a relatively stable period (in both the Trouet et
al. 2009 and Cook et al. 2002 reconstructions), and, as noted above, it follows
Fig. 1a. Comparisons of general northern hemisphere temperature covering the past
millennium (MCA and the LIA and recent warming period) and the often differing
(almost opposite) precipitation records from the west and east Mediterranean. Top:
Temperature reconstructions for the extra-tropical Northern Hemisphere for the
last 1000 years from a multi-proxy synthesis by Christiansen and Ljungqvist (2011)
and for central Greenland from the GISP2 ice-core based on stable isotope analy-
sis and ice accumulation data (Alley 2000; Cuffey and Clow 1997). Bottom: Palmer
Drought Series Index (PDSI) reconstruction from Morocco from tree-ring data by
Esper et al. (2007)more negative values = drier, positive = wetter; and recon-
structed precipitation in the eastern Mediterranean from tree-ring data by Touchan
et al. (2005). Data are shown with 10 and 40 point (pt) (= years in this case) FFT (Fast
Fourier Transform) smoothing (which removes the high frequency noise).5
5 Data for top from: ftp://ftp.ncdc.noaa.gov/pub/data/paleo/contributions_by_author/
christiansen2011/christiansen2011.txt and ftp://ftp.ncdc.noaa.gov/pub/data/paleo/icecore/

greenland/summit/gisp2/isotopes/gisp2_temp_accum_alley2000.txt. Data for bottom from:
Fig. 1b. A comparison of periods noted in the analyses of Nicault et al. (2008) of
decadal or longer intervals of wetter (more negative PDSI) and drier (more positive
PDSI) for Italy, Greece and the Levant versus reconstructed winter NAO indices
from (at bottom) Cook et al. (2002) and (middle) Trouet et al. (2009). The annual
Cook et al. (2002) reconstructed data (light grey) are shown with a 5 year running
average (bold black line). The Trouet et al. (2009) data as published have had a
30-year smoothing (spline) applied to them.6
the Maunder Minimum solar activity episode. The other common wetter
period, ad18701900, is associated with a more positive NAO in the Trouet
al. (2009) record (and the Cook et al. 2002 record includes a major positive
NAO episode in the early ad1880s also). At other times there are marked
regional differences. Wetter periods in Greece or Italy can be the opposite
http://www.ncdc.noaa.gov/paleo/metadata/noaa-recon-8712.html and http://www.ncdc

.noaa.gov/paleo/metadata/noaa-recon-6381.html. Note: all webpages cited in this paper were
accessed in JanuaryFebruary 2012. Sources of data are only cited once even if employed sev-
eral times.
6 Cook et al. (2002) data from: ftp://ftp.ncdc.noaa.gov/pub/data/paleo/treering/
reconstructions/nao_cook2002.txt and the Trouet et al. (2009) data from: ftp://ftp.ncdc.noaa

.gov/pub/data/paleo/treering/reconstructions/nao-trouet2009.txt.
in the Levant, and the reverse. Hence some observed apparent contrasts
in precipitation histories for the Roman period between the western and
central Mediterranean and for the northern coasts of the Mediterranean,
versus the Levant, which we will encounter in later discussions, become
plausible regional phenomena.
8th Century bc Greek Renaissance:

Sun, Climate Change and Patterns
Such complexities and contradictions make the attempt to link climate

change closely with the ancient historical and prehistoric records even more
challenging. There will always be exceptions, and different interpretations
focusing on different elements will very easily reach diverse conclusions.
Eighth century bc Greece offers an interesting but salutary case. In big pic-
ture terms, there is a major solar minimum dated ca. 765bc (Usoskin et
al. 2007)7arguably the largest of the past 3000 yearsand this is plausi-
bly linked to climate change observed widely around the Earth in the 9th
to 8th centuries bc (e.g. van Geel et al. 1996; 1998; 1999; 2000; 2004; van
Geel and Renssen 1998; Speranza et al. 2002; Blaauw et al. 2004; Hall et al.
2004; Chambers et al. 2007; Swindles et al. 2007; Plunkett and Swindles 2008;
Arnaud-Fassetta et al. 2010, 108; Wang et al. 2012). Often cited as the 2800
years bp event (or, in Europe, as e.g. the Iron Age Cold Period, or the Ini-
tial/First Phase of the Subatlantica cooler time marked, for example, by
glacier advances in the Alps: Wanner et al. 2008; Holzhauser et al. 2005; Haas
et al. 1998; Denton and Karln 1973), this forcing can in fact be observed over
a period of time. This is because in some locations there will be what Plun-
kett and Swindles (2008) describe as time-transgressive climate responses
over the following several decades to about a century.
With regard to the Mediterranean, our discussions above indicate in gen-
eral terms that such major solar minima and cooling episodes (e.g. Bond et
al. 2001) seem to link with high storm activity (negative NAO) in the north
Atlantic and high storminess and precipitation in the northwest Mediter-
ranean (Sabatier et al. 2012) and increased precipitation and wetter condi-
7 This is the date for the peak of the episode; Usoskin et al. (2007) estimate the duration at
90 years. The Usoskin et al. (2007) dates are derived from the radiocarbon record. The recent
Beryllium-10 analysis of Steinhilber et al. (2009) places the peak minimum around 817bc and
a sustained period of very low solar activity around 830740bc (see further discussion on
radiocarbon and Beryllium-10 and solar proxy records in the text below).
tions in the Black Sea and northern Red Sea and so wider east Mediterranean
region (Lamy et al. 2006).8 Although, looking at the 2nd millennium ad,
prolonged periods of such conditions (i.e. the LIA) in general lead to drier
conditions in the eastern Mediterranean (Roberts et al. 2012), and, especially
we might predict, in the cooling period leading to a solar minimum (and,
for example, Crete appears notably cool and dry in the mid-16th to mid-17th
centuries ad heading towards the Maunder Minimum: Grove 2001), we can
also observe that on a shorter-term basis the opposite may be the case, and
especially in the period following the solar minimum (as warming occurs).
The ca. 50-year period (ad17001750) following such a major solar minimum
(the Maunder Minimum case) saw increased precipitation and thus better
growing conditions in the Mediterranean (Luterbacher et al. 2006; Nicault et
al. 2008). And, on the short-term in the instrumental period, there is some
correlation of a negative NAO with increased precipitation in the Aegean
region (Xoplaki 2002, Figure 6.1; Hughes et al. 2001; Unal et al. 2012).9
Thus, in macro-terms, such a major solar minimum should/could see
increased aridity in the period leading into the minimum (and, for exam-
ple, Kaniewski et al. 2011 find an increased aridity trend about 800750bc
for coastal Syria), but should/could bring in the following period gener-
ally moister conditions in much of the Mediterranean (so longer and more
reliable growing seasons), as observed for the period following the Maun-
der Minimum (Luterbacher et al. 2006; Nicault et al. 2008 and references
therein), and as a variety of palaeoenvironmental records from the west
(especially Esper et al. 2007; Roberts et al. 2012 and references therein), cen-
tral, and east Mediterranean suggest for the period after about 750bc (e.g.
8 Magny et al. (2003) suggest an approximate correlation of the Bond et al. (2001) cooler
episodes (and reduced solar activity) with drier phases in Lake Siles in southern Spain. But
the chronology seems both critical to a correct interpretation, and less than precise. For the
2800 bp event, in particular, the dry phase at Lake Siles is in fact shown as ending just before
2800 bp (see Magny et al. 2003, Figure 3). Thus we might assume that the period from after
2800 bp was not especially dry. Magny et al. (2003) indicate that a multi-century wet phase in
the central European lakes begins following 2800 bp.
9 The exceptions to this generalisation, of course, could be plant species and contexts in
generally fairly arid areas which are susceptible to cold. For example, Ferrio et al. (2006, 1264)
suggest that cooler temperatures may lead Pinus halepensis trees in Iberia to increase their
reliance on summer precipitation, and, if this was poor, they would seem to record water
stress, and so arid conditions, even if there was in fact generous winter precipitation and
good conditions in agricultural areas and for crops which were not especially intolerant of
cold. Ferrio et al. (2006) suggest this may explain why their study of 13C from archaeological
charcoals indicates an arid period ca. 900300 bc whereas other evidence including for Spain
(as reviewed in this paragraph) generally indicates otherwise: a cooler and wetter period.
Neuman 1985; Heim et al. 1997; Lemcke and Sturm 1997; Issar 2003, 24;
Drysdale et al. 2006; Lamy et al. 2006; Martn-Puertas et al. 2009; 2008;
Martn-Chivelet et al. 2011; Kaniewski et al. 2011; Sabatier et al. 2012; Roberts
et al. 2012).10 Vita-Finzi (2008) has further argued that the Mediterranean
should receive an increase in small, non-erosive, rainfall (during winter and
spring) associated with such a period of major solar cooling, and this too
should be beneficial for agriculture.11 In sum, all these conditions should
provide a more positive regional context for agriculture (and especially
contrasted to the indications of aridity in the preceding period from the 12th
to the 8th centuries bc in the eastern Mediterranean: e.g. Kaniewski et al.
2010; 2011; Drake 2012), and so be good for human populations in this region
(as for example argued by van Geel et al. 2004 with regard to the Scythian
culture), and especially for those in the lower elevation and more arid-risk
regions, such as Athens and central Greece (Manning 2010, 4144).
And so, there seems a potential climate context of relevance for the
indications of dramatic population growth and concomitant demographic,
social, economic, political and material culture change (labelled the struc-
tural revolution by Snodgrass 1980, 1584, or the Greek Renaissance by
Coldstream 1977)including the phenomenon of Greek colonizationin
the Greek world. This starts in the period ca. 750700bc (e.g. Snodgrass 1977;
1980; Hgg 1983; Tandy 1997, 2043; Morris 1998). Morris (1998, 75 and n. 118)
further notes that [i]n the eighth and seventh century, the whole Mediter-
ranean basis experienced substantial population growth, and observes that
[e]xplaining what happened in eighth-century central Greece will require
both a general account of the sudden increase in the pace of life across the
whole Mediterranean and a series of particularized accounts of what made
each region unique. From the previous paragraph, we might suggest that
climate change forced by a grand solar minimum offers part of the general
account across the whole Mediterranean.
However, other scholars postulate severe drought at the very same time
(Camp 1979), and/or instead analyse the ancient sources to suggest the
evidence is more concordant with the view that colonization, and the appar-
10 Some (but not all) records from the southern Levant indicate a relatively arid climate
at this time (Rambeau and Black 2011, 99), and this may reflect the differences between the
conditions for increased precipitation in the western Mediterranean and across the northern
Mediterranean to the Aegean, versus in the southern Levant (see Black 2011).
11 See also Devillers and Lecuyer (2008), who in their Cypriot Little Ice Age case study find
evidence of overall, or net, agriculturally positive change with increased water availability
and moderate-scale floods, and a decrease in highly erosive flash floods.
ent evidence of population growth, is a response to climatic disaster (e.g.

drought) and other historical factors (Cawkwell 1992), rather than to marked
population growth or favourable conditions. Finn et al. (2011, 3168) con-
clude that the palaeoclimate data they surveyed do not in fact show this
event in the east Mediterranean, and, entirely contrary to the hypothesis in
the previous paragraphs, a recent paper by Drake (2012) claims that the Iron
Age recovery in Greece and the rise of Athens do not appear to be climate
related. This is on the basis of Drakes assessment of the limited and rather
coarse information he employed, to the effect that both precipitation and
sea surface temperatures (SST) continue to be low.12 The point, however, for
this discussion is not who is right or wrongrather how the same case can
be and is differently interpreted. There are no easy generalizations in this
field; we also observe the importance of chronology if specific correlations
are to be made (also Maher et al. 2011; Rohling et al. 2002, 590592).
Aims and Limitations
There are a number of works which address aspects of Holocene climate in

considerable detail: e.g. Roberts (1998); Issar and Brown (1998); Issar (2003);
Roberts et al. (2004; 2008; 2012; Roberts, Brayshaw, et al. 2011); Thompson
12 Some of Drakes data plots, such as for paleo-rainfall in his Figure 2, are coarse res-
olution to the point of not being useful, whereas several much higher resolution records
indicate the opposite of his assessment: see Figure 17 below; and, in fact, 3 of his 4 SST plots
in his own Figure 3 indicate a rise in SSTcontrary his assessmentby or from the earlier
1st millennium bcthis includes the data from Emeis et al. 2000, Figure 4.c from both the
Levantine Sea and the Ionian Sea. The study of Rohling et al. (2002, Figure 1d), in particular,
clearly shows the temperature trend in the Aegean Sea (from % warm species of foraminifera:
Rohling et al. 2002, 589) as in fact rising through the early 1st millennium bc (from a cool
period late in the 2nd millennium bc)and, given that the dates for this record appear to be
around 300 years too old by the mid second millennium bc (see Rohling et al. 2002, 590591),
the period of increasing temperature shown as ca. 33002800 Cal bp should in fact likely be
dated as somewhat more recent, and so the rising temperature period will continue at least
through the 8th century bc and perhaps even a little further. As noted in the text above, the
potential for some warming in the specific east Mediterranean region during a wider general
global cooling episode is entirely plausible based on the findings of the Shindell et al. (2001)
study. Elsewhere, a fall in SST around the time of the grand solar minimum in the 8th cen-
tury bc makes general sense (and, for example, the Sargasso Sea record indicates cooling in
the early 1st millennium bc from a warm peak in the 10th century bc, until a sharp warming
trend in 7th century bc, and then generally stable fairly warm conditions through to the 2nd
century ad: Keigwin 1996). The available evidence is consistent with moister and more agri-
culturally favourable conditions in the Aegean region from the mid 8th century bc onwards
following the 9th8th century bc global cooling episode.
et al. (2002; 2006); Mayewski et al. (2004); Staubwasser and Weiss (2006);
Robinson et al. (2006); Rosen (2007); Wanner et al. (2008); Luterbacher et
al. (2006; 2012); Finn et al. (2011); Rambeau and Black (2011). The overall
trends in climate-environment in the Mediterranean through the Holocene
are approximately clear (Roberts et al. 2012). There is something of an east-
west division and see-saw relationship. The eastern Mediterranean saw a
change to less precipitation around 6000 years ago (mid-Holocenein line
with a widely represented trend: Brooks 2006; Wanner et al. 2008, 1795; Ram-
beau and Black 2011), whereas the west saw less dramatic changes gener-
ally, but maximum increased precipitation occurred in the mid-Holocene
60003000 years ago, before a shift to increased aridity leading to present
day levels.
For the Roman period, in particular, a recent paper by McCormick, Bnt-
gen et al. (2012) reviews a number of the sources (scientific and histori-
cal) of data for climate, and climate change, relevant to the Roman world
100bc to ad800. See that paper for a first effort at a respectably compre-
hensive review. I have been much more selective here. My aim is to con-
sider some of the general climate parameters for the wider historical tra-
jectory of the Roman world. I focus on those sources of information which
are indisputably relevant to climate, and not those likely to be affected by
anthropogenic activities (so natural archives which can be linked to cli-
mate forcing, versus secondary records such as pollen sequences, etc., which
can result from both or are a mixture of natural or human activities), since
Roberts, Brayshaw, et al. (2011) conclude that by the mid-1st millennium
bc human-induced landscape changes and activities were a significant ele-
ment of the Mediterranean environment (and Certini and Scalenghe 2011
argue this case from about 1ad for several areas of the Earth). These impacts
range from agricultural activities, to deforestation, to industrial activities.13
13 The issue of human impacts on the natural landscape (e.g. Goudie 2006; Redman 1999)
is fraught and complex in the Mediterranean. Whereas some claims of dramatic impacts
can be critiqued (e.g. Horden and Purcell 2000, 318, 324326), and Grove and Rackham 2001
argue a general thesis that human activity has at most merely abetted climate change (but
cf. Butzer 2003; 2005), there is also undeniably a major human element involved (we can
side-step the issue of prime causality) in the environmental record of the Mediterranean
region from several thousand years ago (e.g. Bottema et al. 1990; Butzer 2005; Eastwood et
al. 1998; Leng et al. 2010; Luterbacher et al. 2012). On trees and timber in the ancient world
generally, see Meiggs (1982), who did not view the ancient world as responsible for massive
deforestation (in contrast to the 19th and 20th centuries ad); the recent studies of Harris
(2011) and Harris (this volume) argue for a mixed responsibility. Even the general assumption
of modern deforestation needs care, since in some areas recent rural depopulation and
I have also not considered historical sources in this essay; appropriate use of
such data requires its own study and methodological approaches.14
How the Roman world interacted with, and was impacted by, climate and
possible climate change, is an obvious subject for interest and speculation.
Climate does not cause or create history in some reductionist paradigm; but
it does provide a context for human society, lives, interactions and decisions,
and potentially provides opportunities or challenges which may influence
both small scale and macro scale human history. Indeed, recent attention
and reconsideration of the dynamic and recursive role landscape and its
perceptions and projections play in producing and maintaining human
social, political and economic organization (e.g. Smith 2003; Falconer and
Redman 2009), only further foreground the relevance of climate and its part
in the formation of the overall social landscape and environment context
for any archaeological or historical study. Hence climate is of considerable
relevanceand yet it has if anything been rather overlooked in scholarship
on the Mediterranean until quite recently (as noted by e.g. Broodbank 2011,
31). More particularly, in light of fairly sweeping and general claims made in
recent papers such as Bntgen et al. (2011)where climate change is, more
or less, suggested to be a primary driver in several key historical changes
such as the decline of the Roman world (and previously see papers such as
Perry and Hsu 2000 for even grander scale linkages of climate and human
development)I wish to assess some aspects of what we know and what we
do not know.
changes in agricultural strategies have led to rapid re-forestation (Debussche et al. 1999).
Nonetheless, there clearly were impacts on the natural landscapewhether through human
silviculture (e.g. Conedera et al. 2004) or the harvesting of natural forests, as evident from
the requirements of the pyrotechnical industries of the Bronze Age through Roman periods
such as metallurgy, and, in the Classical era, the production of bricks and lime-cement, apart
from needs for building and so on (Wertime 1983; Horden and Purcell 2000, 334336; Meiggs
1982). The Roman period sees the first evidence of large-scale pollution of the environment
and atmosphere from industrial-type anthropogenic activities (versus instances of burning
of landscapes), in particular, there is evidence of widespread lead and other metals (mercury,
copper) pollution from metallurgical activities (e.g. Mighall et al. 2009; De Vleeschouwer et
al. 2007; Renberg et al. 2000; Shotyk et al. 1998; Martinez-Cortizas et al. 1999; 1997; Hong,
Candelone, Patterson and Boutron 1996; 1994).
14 There is enormous potential in the extant textual record. For some discussions of such
material, see e.g. Bonneau 1971; Hassan 1981, 2007; Neumann 1985; Garnsey 1988; Sallares 1991;
Stathakopoulos 2000; 2004; Teleles 2004; Kondrashov et al. 2005; Little 2006; Lehoux 2007;
McCormick et al. 2007; McCormick, Bntgen et al. et al. (2012). In particular, see the Digital
Atlas of Roman and Medieval Civilization (McCormick et al. 2010 at http://darmc.harvard
.edu/).
The key issues are the types of data, the scale, accuracy and precision of
resolution, and the approaches taken. At present we can still begin by say-
ing that we need (many) more high-resolution datasets and more regional
studies to be able to advance understanding of human-climate relationships
in specific terms (versus very general associations) for the Mediterranean
region.15 The recent review of a range of climate data available from geologi-
cal sources for Italy by Giraudi et al. (2011) is a good example both of how far
we have comenamely there are quite a lot of databut also of these prob-
lems, since the data reviewed offer some general long-term trends, and some
specific insights (like picking up major multi-regional events such as the
event at 4200 years bp), but no easy synthesis into a coherent high-resolution
story for Holocene, let alone Roman, Italy. Chronological resolution and/or
control is a, or the, key missing element in many cases. The excellent, thor-
ough, but, ultimately, rather uninformative synthesis of Finn et al. (2011) on
the last 6000 years of climate in the east Mediterranean is a similar case. Of
over 80 data sources reviewed in that paper, only 3 approach high-resolution
(that is around less than 10 (or even less than 20) calendar years resolution),
and of these the Oman speleothem data do not relate to Mediterranean cli-
mate,16 leaving just two directly relevant high (15 years: Jones et al. 2006)
or near-high (240 years: Kaniewski et al. 2007) resolution records from
Turkey. In many other cases, the underlying chronologies are much less pre-
cise (with precisions of decades to several centuries), such that records are
coarse and lacking the necessary detail to allow correlations with history
and archaeology, or it is not even really certain exactly where key features are
placed, with possible movement of decades to centuries potentially allow-
ing entirely different syntheses and scenarios. Much better chronological
resolution and control are necessary.
However, this situation raises another fundamental topic: what type of
resolution and what type of climate data are ancient historians and archae-
ologists seeking to uncover and address? Their goals are not necessarily the
same as those of the palaeoclimate community. While sometimes as his-
torians we may seek the specific yeara climate record linked to a key
famine, for examplemore often we actually want the trend on a more
15 The statement of Rosen (2007, 153) remains largely true: we still do not have fine
enough resolution in the dates of climatic data to be able to clearly mesh climatic fluctuations
with precise historical events in the Roman-Byzantine period.
16 Instead they primarily reflect the movements of the Intertropical Convergence Zone,
ITCZ, and the Indian Summer Monsoon (Fleitmann et al. 2003; 2007).
decadal to century-level basis (but with precision on the time period this
trend applies to). We accept and assume year to year variability whatever
the climate regime (as observed in the recent instrumental era17 and our own
lives). What we want to know is whether the trend over a decade to several
decades to several generations (so to 100 years or so) is generally positive in
terms of key resources (e.g. agriculture), or the reverse, or stable; this sort of
trend can then act as an underlying forcing parameter which could shape
sets of human decisions and actions. This does not have to be as sensational
as a mega-drought (e.g. Cook et al. 2004; 2010)though this is just the sort
of palaeoclimatic forcing data historians and archaeologists keenly want to
know about for their regions of interest (e.g. Buckley et al. 2010). In palaeo-
climate terms historians are more focused on low-frequency variations (the
trends over multi-year and longer periods) and not the high-frequency vari-
ations year-by-year.
Yet this has not always been the aim of palaeoclimate research. Tree-ring
studies, despite offering high-resolution data, are an especial case in point.
As Finn et al. (2011, 3164, 3168) note, most tree-ring studies so far in the
Mediterranean region have employed statistical methods which in the effort
to remove noise also remove trends in favour of year to year variation (e.g.
Hughes 2002, 102104). Thus good tree-ring studies can identify specific pos-
itive or negative years in terms of tree-ring growth, and, where the sources
are available, these can sometimes be linked with historical records of, e.g.,
drought or high precipitation (e.g. Kse et al. 2011), but, as Finn et al. (2011,
3164, 3168) lament, many of the available tree-ring studies from the Mediter-
ranean do not in fact pick up such obvious trend events as the MCA or the
LIA Clearly this is a serious short-coming. Of course scientists in the field
are well aware of this issue, and considerable effort has gone into developing
methods (such as Regional Curve Standardization, or RCS) to try to extract
the low frequency signals from tree-ring series, and to overcome such limi-
tations of these data (e.g. Esper et al. 2002; 2003; 2007; Briffa 2000; Briffa et
al. 2001; Frank et al. 2007; Hughes et al. 2011). In the western Mediterranean
17 The period for which we have instrumental data on climate is short. I refer loosely to the
last couple of centuries. The longest continuous temperature record, from central England,
dates from ad 1659 (monthly data) and ad 1772 (daily data) (Manley 1974; Parker et al. 1992):
data and further bibliography available from http://www.metoffice.gov.uk/hadobs/hadcet/.
Despite some sporadic instrumental records from the 17th century onwards, national mete-
orological services were only established widely in Europe and the Mediterranean in the
mid-19th century (Luterbacher et al. 2006; 2012), and there are many areas where meteoro-
logical records only begin in the second half of the 20th century.
Esper et al. (2007) employed both a large data set focused on long-lived trees,
and a range of approaches, and were able to recognize trends indicative
of the MCA and the LIA. Similar work is now needed elsewhere in the
Mediterranean. The tree-ring study of Bntgen et al. (2011), though outside
the Mediterranean, is also relevant to the Roman world. We can also hope
for more high-resolution data sets from speleothems in coming years; in the
right circumstances these can offer good correlation with instrumental data
comparable to tree-rings (see Jex et al. 2010 and Gktrk et al. 2011 for such
very high resolution studies in the recent period from Turkey).18
As one focuses down in scale to the region or site, there is the important
added complication of the human subjects, and their role in the environ-
ment under study, especially for the later Holocene. Rosen (2007, 150171)
nicely highlights some of the issues and problems (and the potential for
circularity in argument) in her review of the interpretations of evidence of
increased and widespread human activity in the Roman-Byzantine period
in arid areas like the Negev. She points out that climate is only one factor
in historical explanation (see also Horden and Purcell 2000, 298341).19 We
cannot write climate history on the basis of the record of human activities
alone, or vice versa.
What the Sun Was Doing in the Roman Period
For the Roman period, ca. 300bc to ad800, the main general (background)
forcing factors on climate which operate at this scale are the Sun and vol-
canic eruptions. Volcanic activity relevant to the northern hemisphere
seems to have been notably low for much of the Roman period, and cer-
tainly for the ca. 570 years from ca. 35bc to ad 535: see Figures 19a and 19b
below. This is an important absence of evidence and is noted further below.
Here I therefore discuss the Sun.
18 There is however an important methodological issue of relevance here. There is a major
difference between a record which is demonstrably annual and absolute (for example a
robust securely crossdated tree-ring chronology) and a record which is only approximate
(however high resolution). As Betancourt et al. (2002) argued, high-resolution speleothem
records are not absolute or annual resolution unless they can achieve the robustness levels
of tree-ring chronologies, and should be described instead as near annual or subdecadal.
19 Barker (2002) offers another relevant critique, reviewing two cases (Tripolitanian pre-
desert in Libya and the Wadi Faynan in Jordan) where marginal arid environments (then as
now) were transformed by systems of floodwater farming through the centuries of Roman
imperialism. The archaeological evidence thus reflects primarily economic and political tra-
jectories (each very different), rather than climate.
A linkage between variations in solar activity and the Earths climate

became apparent with the examination of records of sunspot activity. Schol-
ars such as Eddy (1976; 1977) observed an association between times of few
or no sunspots (and so low solar activity) and cooler temperatures and cli-
mate anomalies on Earth, and the reverse, and Magny (1993)20 argued that
solar variations were associated with (and were a main cause of) varia-
tions in European precipitation. The key example where there are historical
data is the Maunder Minimum (about ad16451715). There was little or no
observed sunspot activitythe open solar flux is calculated to have been
a mere 11% of that observed during the very recent grand solar maximum
(Lockwood and Owens 2011)and this period is associated with some of
the colder temperatures of the last four centuries on Earth (Bradley and
Jones 1993). Hence for some time it has been agreed that solar forcing was
the predominant factor in the observed cooler 17th and 18th century ad
temperatures (e.g. Lean and Rind 1998)although this may be especially,
or particularly, in the case of Europe, where most of the historic Maunder
Minimum-related data come from, based on analysis of both historic and
recent data (Lockwood, Harrison et al. 2010; Luterbacher et al. 2001). In the
pre-instrumental era there is more debate over the exact mechanisms, and
relations are complicated, but again there is now generally agreed to be a
clear correlation of the observed palaeoclimate record with changing solar
activity (e.g. Lockwood 2006; Bond et al. 2001; Neff et al. 2001). Recent work
has also started to offer plausible models to explain how changes in solar
activity may force climate on Eartha previously debated point (e.g. Shin-
dell et al. 2001; Gray et al. 2010; Ineson et al. 2011; Lockwood, Bell, et al. 2010;
Hegerl et al. 2011).
Without considering the many complexities of this topic (see e.g. Gray
et al. 2010 for a review), we can regard the changing activity of the Sun as
a key climate parameterwith low (high) solar activity associated over the
Holocene with low (high) temperaturesand general associations between
solar activity, climate, and human historical trajectories have been noted by
many scholars (e.g. Perry and Hsu 2000). Some cosmogenic isotopes14C
(radiocarbon) and 10Be (Beryllium-10)provide approximate proxy records
of changing solar activity for the period before historical or instrumental
data (i.e. for before ad1600) (Vonmoos et al. 2006).21 The 14C record from
20 Recently, with further literature review, see Magny et al. (2010).

21 For discussion of the production of these cosmogenic isotopes, see Masarik and Beer
(1999).
known-age tree-rings offers an absolutely dated and highly resolved record

of changing atmospheric 14C levels through the Holocene for the Northern
Hemisphere: Figure 2. Although affected by other factors (e.g. changes in
the oceans, or changes in the geomagnetic field), this record of changing 14C
in recent millennia largely reflects changing solar activity (e.g. Stuiver and
Quay 1980; Stuiver and Braziunas 1988; 1993; 1998; Stuiver et al. 1991). Given
various assumptions about the carbon cycle, it is in turn possible to model
14
C production as (largely) a contemporary proxy of changing solar activity
(e.g. Usoskin and Kromer 2005; Siegenthaler et al. 1980). Figures 34 show
how the changes in the 14C record, and the modeled production record of
14
C, broadly reflect the activity of the Sun as reflected in observed sun-spot
numbers (SSN) over the last 350 years. A model transforming such 14C pro-
duction estimates into a total solar irradiance (TSI) reconstruction has been
published recently by Vieira et al. (2011): see Figure 6a.
10
Be records from (principally) polar ice-cores offer the other main long-
term solar proxy record. Over recent centuries these have annual resolution,
and there are only very small calendar errors to beyond the Roman period.
As an example, Figure 5 shows two 10Be records for the recent centuries ver-
sus both observed sun spot numbers and the 14C record. Beer et al. (1990)
demonstrated that the annual resolution Dye-3 record closely recorded solar
activity and especially the 11-year solar cycle. A number of studies have
demonstrated the good comparison of 10Be and 14C records (e.g. Beer et al.
1988; Bard et al. 1997; Beer 2000; Beer et al. 2006; Muscheler et al. 2007). The
advantage of 10Be is that it is deposited at the Earths surface more quickly
and directly after production in the atmosphere (typically 12 years) com-
pared to 14C (typically a 1520 year lag is expected comparing 10Be to 14C:
Stuiver and Braziunas 1993; Bard et al. 1997), and it is not damped by the car-
bon cycle or complicated by changes in it (in contrast with 14C). However, the
problems with 10Be are twofold. First, the 10Be records contain considerable
noise reflecting transport (and associated mixing) and then deposition pro-
cesses (local weather conditions: precipitation, wind speed), and so it can
be difficult to disentangle climate-driven versus solar-driven changes (e.g.
Field et al. 2006). Such issues have been reviewed by Heikkil et al. (2011),
who nonetheless find that 10Be offers a useful record when treated appropri-
ately. Second, the current precision and accuracy of ice-core dating becomes
less certain backwards over time (Baillie 2010), and various anomalies are
clearly evident at decadal and greater levels over some periods, such as in
the second millennium bc, indicating dating or other issues in various key
records (e.g. Southon 2002; 2004; Muscheler 2009see further in footnote
24). A recent record of modelled total solar irradiance (TSI) from 10Be for
Fig. 2. A. The standard radiocarbon (14C) calibration curve for the period 3000bc
to ad1950 from known age trees (IntCal04: Reimer et al. 2004the tree-ring based
14C data are the same as in the more recent IntCal09: Reimer et al. 2009). INSET:
The inset shows a section (1000bc to 500bc) of the standard 14C calibration curve
(IntCal04) in more detail (1 error band)note the characteristic wiggly shape.
B. The 14C record per mille () from Athis is the relative 14C content decay
corrected and normalized (for the definition of this, see Mook and van der Plicht
1999): data from Reimer et al. (2004). C. The residual 14C record per mille () after
a 1000-year moving average is removedthis record thus highlights the relative
changes in the 14C record around the underlying longer-term trend: data from
Reimer et al. (2004). Note: the stated errors on lines A, B and C are not shown for
reasons of clarity and scale.22
22 Radiocarbon data files are available from: http://www.radiocarbon.org/IntCal09.htm
and http://www.radiocarbon.org/IntCal04.htm.
Fig. 3. Bottom: observed sun-spot numbers (SSN) per year. The marked declines
reflect two known cooler climate episodes: the Maunder Minimum and the (much
smaller) Dalton Minimum. The recurrent spikes in SSN reflect the well-known
approximately 11-year cycles in solar activity (and a longer 22-year cycle). The SSN
data from historical records of sun spot numbers come from Hoyt and Schatten
(1998a; 1998b). Top: The annual 14C record per mille () ad16001900 (Stuiver
et al. 1998) (grey) and an 11-year moving average of this record (black). Middle: The
14C record per mille () from IntCal09 and IntCal04 (Reimer et al. 2009; 2004)
and two models of 14C production: (a) from the MarMod09 model in Reimer et al.
(2009); and (b) from the iterative model in Usoskin and Kromer (2005). No errors
shown for lines in Top or Middle plots.23
23 The SSN data from historical records of sun spot numbers come from Hoyt and
Schatten (1998a; 1998b) and are available from http://www.ngdc.noaa.gov/stp/SOLAR/

ftpsunspotnumber.html. The annual 14C record per mille () ad16001900 (Stuiver et al.
1998) is available from http://depts.washington.edu/qil/datasets/. The iterative 14C produc-
tion model data (from Usoskin and Kromer 2005) were kindly provided by Bernd Kromer.
The Suess Effectfirst noted by Suess (1955; Revelle and Suess 1957)refers to the post-
Industrial Revolution anthropogenic situation where large amounts of fossil-fuel derived CO2
have been emitted into the atmosphere diluting the natural atmospheric 14CO2 concentra-
tion (and hence have noticeably modified what would have been the natural atmospheric
radiocarbon record from about ad 1900 onwards).
Fig. 4. Top: A. The residual annual 14C with 2 point (pt) FFT smoothing from the
data shown in Figure 3 (top in grey) from Stuiver et al. (1998) calculated minus
a 22pt FFT smoothing to emphasise the change around the longer-term trend.
B. The residual annual production of 14C (iterative methodsee Figure 3) with
2pt smoothing. No errors shown. Bottom: Observed sun spot numbers (SSN) from
Figure 3. Observe that the 14C derived residual records A and B closely reflect the
peaks and troughs of the SSN recordindicating that they are largely a solar activity
proxy.
the past 9312 years from Steinhilber et al. (2009) is shown in Figures 6a and
6b, and compared against (i) the Vieira et al. (2011) TSI reconstruction from
the 14C record in Figure 6a, and (ii) modeled 14C production in Figure 6b. A
good general comparison is evident in each case: the Steinhilber et al. (2009)
and Vieira et al. (2011) TSI records are generally similar in trend and timing
(same variability range to quote Vieira et al. 2011). The absolute levels of
reconstructed TSI are not relevant to our discussions in this essayrather
the trends and the solar activity extrema, and these are very similar for the
last 3000 years (and so the Roman period): see Figures 6a and 6b.24
To consider what the Sun was doing during the Roman period, we can
therefore look more carefully at both the 14C record, a derived reconstruction
of Sun Spot Numbers (Solanki et al. 2004), and the 10Be record, in the period
300bc to ad800: see Figures 7 and 8.
24 Looking at Figure 6b in detail, the 10Be series (from ice cores) and the 14C production
models (from known age trees and the carbon cycle model) correlate in terms of calendar
placements relatively closely over the ad period and to the ca. 360bc minimum, and again
in the 4th millennium bc, but there is something of a mismatch in the intervening period.
There are apparently errors in the ice-core timescale especially in the period of the mid 1st
millennium bc through 2nd millennium bc, and the ice-core chronology has been incorrectly
placed as too old by around a couple of decades or a little more. The 14C record on known
age tree-rings is regarded as likely the more solid record for the period of time shown
in Figure 6b because the relevant dendrochronology and 14C have been replicated more
than once in the course of different groups building the radiocarbon calibration sequences
(e.g. Reimer et al. 2009; 2004; Muscheler 2009, 282284), and also in various larger-scale
tree-ring wiggle-match exercises (e.g. Kromer et al. 2001; Manning et al. 2010). In contrast,
the ice-core records are less replicated and review of the literature indicates clear grounds
for possible errors in the period indicated above, and indeed small possible errors even by
the Roman period (e.g. Southon 2002; 2004; Baillie 2008; 2010). The issue of an offset of the
10Be ice-core record from the GRIP core versus the 14C timescale for the 2nd millennium
bc has already been noted by Muscheler (2009). While minor in terms of assessing the
general history of solar activity, this issue potentially has substantive implications for the
exact dates for major volcanic eruptions in the ice-core records in the period from the mid
1st millennium ad through the 2nd millennia bc where a dating problem is identified: in
particular, it suggests that it is likely that the ice-core dates are too old by a couple to a few
decades. To take the best known case: the ice-core signal argued to perhaps represent the
Santorini (Thera) eruptionfollowing the arguments of Vinther et al. (2008); Muscheler
(2009, 276277)which is dated from the ice-core records ca. 1642bc, might instead be
considered to perhaps date ca. 20 years later (Muscheler 2009). This revision could lead to
a consistent scenario where the 14C evidence for the dating of the Santorini eruption (from
both short-lived sample material from the volcanic destruction level on the island and from
the last extant growth increment of an olive branch buried by the eruption on the island:
Manning et al. 2006; Manning and Kromer 2011; Friedrich et al. 2006), as well as the ice-core
record of a major eruption which (following Vinther et al. 2008; Muscheler 2009, 276277)
could be Santorini, all point to a date in the late 17th century bc for this event (Manning and
Kromer 2012).
Fig. 5. High resolution 10Be data from Greenland for the most recent four centuries.
The NGRIP record (Berggren et al. 2009) is annual and it is shown as a 3-year
average.25 The Dye-3 data (Beer et al. 1990; 1998) are also annual and are shown
smoothed on a 3pt basis (compare approximately Beer et al. 1990, Figure 1).
25 The Berggren et al. (2009) data are available from ftp://ftp.ncdc.noaa.gov/pub/data/
paleo/icecore/greenland/summit/ngrip/ngrip-10be.txt.
Fig. 6a. Comparison of the Total Solar Irradiance (TSI) reconstructions of Vieira et
al. (2011) from the 14C record versus Steinhilber et al. (2009) from the 10Be record. For
the definition of dTSI see Steinhilber et al. (2009).26
26 Data for the Vieira et al. (2011) TSI reconstruction available from https://www.mps
.mpg.de/projects/sun-climate/data/tsi_hol.txt; data for the Steinhilber et al. (2009) dTSI re-

construction available from ftp://ftp.ncdc.noaa.gov/pub/data/paleo/climate_forcing/solar_
variability/steinhilber2009tsi.txt. The offset of 1365.6 watts/m2 to plot and compare the Vieira
et al. (2011) TSI follows Vieira et al. (2011, Figures 7 and 8).
Fig. 6b. Top: Total Solar Irradiance (TSI) reconstruction from ice-core 10Be records
from Steinhilber et al. (2009)see Figure 6a. Bottom: 14C production from Mar-
mod09 (Reimer et al. 2009)note the 14C production record has been inverted (con-
trast with Figures 3 and 5). The two curves thus appear very similar, but increased
14C production in fact correlates with decreased TSI. The grand solar minima and
maxima are taken from Usoskin et al. (2007note dates for major minima or max-
ima episodes from the 10Be record of Steinhilber et al. 2009 are similar and some
comparisons are noted in my text and notes below).
Fig. 7. Top: detail of the Total Solar Irradiance (dTSI) reconstruction from ice-core
10Be records from Steinhilber et al. (2009)see Figures 6a and 6bfor the period
300bc to ad800. Bottom: two 14C production models: (a) from Marmod09 (Reimer
et al. 2009) and (b) the iterative model from Usoskin and Kromer (2005). Note that
the 14C production record has been inverted (contrast with Figures 3 and 5). The
curves at top and bottom thus appear similar in shape, but increased 14C production
correlates with decreased dTSI. The grey panel indicates where the variation of the
dTSI values are within one standard deviation of the average value between 300bc
to ad800. Note the relative lack of signal, and stability, in the 2nd century bc to the
2nd century ad.
Fig. 8. Top: the two 14C production models in Figure 7now not invertedfor the
period 300bc to ad800. Bottom: the Sun Spot Number (SSN) reconstruction from
Solanki et al. (2004) for 300bc to ad800. Larger SSN are correlated with a more
active sun. Observe that decreased 14C production corresponds to increased Sun
Spot Numbers, and the reverse. The two grey panels indicate where variation of
the Usoskin and Kromer (2005) iterative method 14C production values and the
reconstructed SSN values of Solanki et al. (2004) are within one standard deviation
of the respective average values between 300bc to ad800. Note the relative lack of
signal, and stability, in the 2nd century bc to the 2nd century ad.
If we consider Figures 2 and 6a and 6b for the general context (looking at

the last 6000 years), we see that in the Mediterranean climate context the
grand solar minima ca. 2860bc and 765bc each equal a clear 14C produc-
tion peak or TSI/dTSI minimum (see Usoskin et al. 2007 for dates of grand
solar minima and maxima: the respective dates for the peak minima from
the analysis of the 10Be record of Steinhilber et al. 2009 are slightly earlier at
ca. 2920bc and 817bcbut, as evident in Figures 6a, 6b and 7 above, the
major minima and maxima episodes are closely aligned and approximately
of the same dates in both the 14C and 10Be derived records). Each also roughly
demarcates the beginning of widely represented cultural floruits, and each
in addition follows what appear to have been negative or challenging cli-
mate change episodes: (i) the sharp global downturn ca. 5200 years ago
observed in several records (e.g. Thompson et al. 2006, 10542; Staubwasser
and Weiss 2006, 379380); and (ii) the changes at the end of the 2nd millen-
nium bc to an episode of aridity (e.g. Kaniewski et al. 2010; Roberts, East-
wood, et al. 2011, 153). In the case of the 3rd millennium bc, the grand solar
minimum episode in the 29th century bc marks the beginning of the main
high-point Early Bronze cultural periods in the east Mediterranean, such as
the Early Bronze Age II international spirit of the Aegean world (Renfrew
1972; Broodbank 2000) and the time of the formation of early states and the
contemporary major early world system of the wider Levant and Near East
(van de Mieroop 2004; Sherratt and Sherratt 1991). In the case of the ear-
lier 1st millennium bc, as noted above, the grand solar minimum in the 8th
century bc marks the beginning of the Greek Renaissance and related cul-
tural developments, or the Italian Advanced Iron Age Orientalising period =
(Transalpine early Iron Age) = Hallstatt C horizon (especially adopting the
new higher radiocarbon-based chronology: e.g. Nijboer et al. 1999/2000; van
der Plicht et al. 2009), and a general take-off of human activity across the
Mediterranean region. The total ca. 1470 year period between (and with-
out) grand solar minima (2860bc to 1390bc), notwithstanding four grand
solar maxima, is notable as a largely extended interval of relatively stable
solar conditions compared to more marked variations in the 4th millen-
nium bc or the early 1st millennium bc or from the mid 3rd century ad to
the present. The overall Roman period (or Greco-Roman period) occupies
the one other long (1045 years) relatively stable period lying between two
well-spaced grand solar minima (between ca. 360 bc and ca. ad 685).27
27 The dates for the major (extreme) solar activity minima from Steinhilber et al. (2009)
in these cases are about 372368bc and about ad 657672, with the two episodes thus closely
and similarly placed from both the 14C and 10Be derived records.
It is thus tempting, looking at the overall solar proxy record summa-

rized in Figures 6a and 6b, to see these two long relatively stable intervals
each between two well-spaced apart grand solar minima as providing espe-
cially favourable, or at least benign, conditions, and context, for the devel-
opment of complex civilizations in the Mediterranean region. The first cov-
ers the Early Bronze Age takeoff through the main palatial periods of the
civilizations of the eastern Mediterranean in the Late Bronze Age (there
was a negative blip around, and for a period, after ca. 2200bcWeiss et
al. 1993; Dalfes et al. 1997; Staubwasser and Weiss 2006, 380383; cf. Finn
et al. 2011)but the large-scale picture is of the main Bronze Age civiliza-
tion phase of the Mediterranean region covering some 1500 years in all (e.g.
Mathers and Stoddard 1994; Van de Mieroop 2004). The second and arguably
even more generally stable millennial-scale period, from the 4th century bc
to the 7th century ad, covers the whole of the Roman period. Figure 9 sum-
marises the approximate record of solar activity within this Roman millen-
nium.
In detaillooking at Figures 69 and with the dates and approximate
phase numbers after 300bc and before ad700 below from Figure 9we see
a grand solar minimum ca. 360bc (a time of change), then a major active
solar peak around or just before 270bc followed by (1) a less active sun (from
this peak) to around 200bc, but then, from about 200 bc to around ad 135 (2),
there was an unusually stable period of over 300 years (so likely favourable).
Variability then increases (with a sharp downturn ca. ad 200260, and then a
reversal) from about ad135305 (35). Solar activity around ad 300 is as high
as it has been for some centuries: a warm 4th century ad follows from this
peak with fairly stable conditions ca. 305370ad (6). From around ad 370 to
ca. ad690 there is then a long period with an overall trend down to a very
inactive sun (a grand solar minimum about ad685or ca. ad 657672 for
the extreme minima in the 10Be based record of Steinhilber et al. 2009), and
we might expect this to be a period of change. This process occurs in stages:
decline in solar activity ca. ad370 to ca. ad435 (7), period of reversal and
likely amelioration ca. ad435515 (8), and then a period of marked decline
in solar activity ca. ad515 to ad690 (9), with a reversal or plateau about
ad540630. The impact of these changes, especially the main change ca.
ad515690, will have varied according to context, but the previous era of
relative stability was gone. West-East and North-South differences could be
anticipated in a period of greater climate variability.
While clearly there is some risk of gross simplification and over-generali-
zation, the solar activity record seems to reflect a suitable overall context for
the Greco-Roman period. Such a general Roman Climate Optimum (wet,
Fig. 9. The main trends of the solar proxy records in Figures 7 and 8 for the period
300bc to ad700 are summarized and compared from both the various 14C based data
(from the two 14C production models and the reconstructed SSN datarounded to
nearest ca. 5 years and averaged) and the 10Be data in terms of (i) less active sun
(grey), (ii) more active sun (white) and a long relatively stable period ca. 200bc to
ad135 and a shorter relatively stable period ca. ad305370 (both cross-hatched).
The intervals are only approximate, and there is a 510 year error, given the
variations in the records, for the 14C summary. This interpretation is the authors
subjective assessment. The 9 phases identified on the right are taken from the
14C-based model data (see Figure 8); generally the 14C and 10Be records are quite
similar as can be observed in Figure 7, but I have favoured the 14C record as the record
to cite because it is a more robust general signal (from multiple measurements from
different loci in the northern hemisphere all of absolute date), and it avoids some
short-term noise.
humid), or Roman Warm Period (RWP), has been recognized in various

palaeoclimatic records, and is usually placed from the mid-later 1st millen-
nium bc through to the earlier 1st millennium ad to at most about ad 400
(e.g. Lamb 1995, 156159; Hass 1996; Martinez-Cortizas et al. 1999; Desprat
et al. 2003; McDermott 2004; Ji et al. 2005; Jiang et al. 2005; Seidenkrantz
et al. 2007; Garca et al. 2007; Martn-Puertas et al. 2009; Patterson et al.
2010). Among other conspicuous observations which indicate a warm inter-
val: the period ca. 400bc to ad400 is a notable long period of major glacier
recession in the Alps (Holzhauser et al. 2005). In finer detail, the likely peri-
ods of benign or positive solar activity correlate approximately with, first,
the development and floruit of the Roman world and empire in the 2nd
century bc to the 2nd century ad, then, second, the recovery of the 4th cen-
tury ad, and, third, the re-emergence of the Late Roman-Byzantine world
in the mid-5th to early 6th centuries ad (P. Brown 1971; Cameron 1993).
The likely less stable, and less positive to negative periods, like the 3rd cen-
tury ad (and the so-called crisis of the 3rd century: Watson 1999, 120), and
the mid-6th to 7th centuries ad, correlate broadly with periods of decline or
change and re-orientation.
Some Other Records of Climate and

Their Relevance to the Roman World
There are myriad possible resources one might cite by way of records or
proxies indicative of global, hemispheric, regional or local climate and envi-
ronment for all, or parts of, the past few millennia. A number of these
other potential sources of information, of greater and lesser resolution,
are surveyed by McCormick, Bntgen et al. (2012) focused on the Roman
world and region. Papers by Luterbacher et al. (2004; 2006; 2012; Xoplaki
et al. 2005; Guiot et al. 2010) provide comprehensive reviews of the current
sources of data from both climate science, and from the mining of historical
records, for the study of Mediterranean climate in the past 500 to 2000 years.
To date, a number of researchers have producedoften differingmulti-
proxy reconstructions especially of global or hemispheric temperature cov-
ering the last 1000 to 2000 years, and thus including the later two-thirds
of the Roman period (see Jones et al. 2009). Future work will see further
and more sophisticated statistical approaches applied to palaeoclimate data
beyond the models employed so far aimed at richer and more robust out-
comes (Tingley et al. 2012; Werner et al. 2013). The integration of advanced
climate science is therefore close to becoming unavoidably relevant to the
study of the Roman world; indeed, a data-rich synthesis of an impressive net-

work of tree-ring data from central Europe has recently claimed to demon-
strate this relevance in dramatic fashion (Bntgen et al. 2011).
Tree-Rings
Tree-rings are the best available high-resolution (annual to subannual) cli-
mate archive for most parts of the world, especially for studying a spe-
cific region (e.g. Hughes et al. 2011; Hughes 2002). The problem for the
Mediterranean region (south of the Alps)contrast for example central
Europe (Bntgen et al. 2011; Nicolussi et al. 2009)is that we presently
lack long absolute tree-ring chronologies which cover the time period of
the Roman world.28 No current tree-ring chronology runs back continuously
from the present beyond the mid to late 1st millennium ad, and no sig-
nificant dendroclimate series extends beyond the late 1st millennium ad
(where available, most Mediterranean dendroclimate analyses reveal hydro-
climate information and not temperature). For the Mediterranean in the
Roman period, we thus rely on extrapolation from areas elsewhere which
have longer sequences and dendroclimate analysessuch as the large cen-
tral Europe dataset analysed in Bntgen et al. (2011).29
It is worth noting first, however, that some potential long-lived trees and
sources of wood may exist in the Mediterranean region and may offer future
promise. For example, Cremaschi et al. (2006) report some long ring count,
but not cross-matched (i.e. not dendro-dated), tree-ring samples of Cupres-
sus dupreziana from the central Sahara region, which seem on the basis of a
few radiocarbon dates to cover parts of the Roman period. These authors
interpret briefly alternating periods of wide and narrow rings to perhaps
indicate recurrent rainy periods in the 16th century bc to 5th century ad,
28 A number of areas of the circum-Mediterranean world have produced tree-ring records
covering variously multi-century periods to the whole of the last millennium, but so far they
do not reach back to the period of the Roman world (e.g. Touchan et al. 1999; 2003; 2005; 2007;
2010; Touchan, Anchukaitis, et al. 2008; Touchan, Meko, et al. 2008; Akkemik and Aras 2005;
Akkemik et al. 2005; Griggs et al. 2007; Esper et al. 2007; Nicault et al. 2008; Kse et al. 2011;
Seim et al. 2010; Serre 1978). A recent exception is a tentative oak dendrochronology from the
Aegean region approximately placed as reaching from the present to ca. ad398 (Pearson et
al. 2012)see text below.
29 Further away from the Mediterranean world there are other long tree-ring series (to
give just two examples from Eurasia, see e.g. Grudd et al. 2002; Sheppard et al. 2004), but only
the central European datasets (i.e. those collected and analysed in Bntgen et al. 2011) offer
a relatively proximate source for the Mediterranean region and for the European part of the
Roman world during the Roman period.
and then a shift to prevailing narrower rings to indicate drier conditions

from about ad450 onwards. Another Cypress species, Cupressus semper-
virens, from the upper timberline region of west Crete, may be another very
long-lived but problematic resource. Rackham and Moody (1996, 189190)
note some examples with over 1000 rings (years) age and speculate that they
might provide a record of good and bad years back to Roman times. To
gain a hint of possible promise, the ring count on one living stem cored in
the White Mountains (Lvka Ori) in 2007 by a team from the Malcolm and
Carolyn Wiener Laboratory for Aegean and Near Eastern Dendrochronol-
ogy was 926 (ANG-13), and the approximate test 14C wiggle-match dating
of a sample from a previously cut branch of another tree (ANG-7) demon-
strates the likely millennial scale age of the parent tree: Figure 10.30 But the
cross-dating of these trees is a considerable challenge due to irregular and
eccentric growth, missing rings, and, often, scarring or other injury-induced
issues (human and animal).
In terms of climate, the approximately-dated record shown in Figure 10
would indicate a general and rough correlation of increased growth during
warmer periods in the northern hemisphere (and the reverse), when com-
pared to the composite extra-tropical northern hemisphere temperature
record of Christiansen and Ljungqvist (2011). This pattern, which mimics
that of northern and central Europe, and not the arid Mediterranean (so
positive growth conditions ca. ad15001650 when in general this is a dry
and unfavourable period in much of the Mediterranean: Nicault et al. 2008),
highlights the potential differences between controls on growth at the upper
timber line in some areas of the Mediterranean, versus those operating at
lower elevations where humans mainly live and farm.
Another exciting recent development is the oak dendrochronology
(Quercus sp.) for the 1st millennium ad that is starting to take shape from
timbers from recent archaeological work in Istanbul (Pearson et al. 2012;
Manning et al. 2012). Linking this material to the existing oak chronologies
for north Turkey and north Greece (Griggs et al. 2007; 2009) and south-
east Europe and further north offers the future prospect of a 2+ millennia
oak dendrochronology from this region. The exact sources of these oaks are
not yet known, and they may have been imported from local forests, from
S.E. Europe, or from the Black Sea region. For example, at present, there is a
good 213-year oak series running from ca. ad398610, comprising 85 series
30 See also http://dendro.cornell.edu/reports/report2007.pdf pp. 34 and http://dendro
.cornell.edu/reports/report2008.pdf pp. 12.

Fig. 10. Top: Extra-tropical Northern Hemisphere temperature record from Chris-
tiansen and Ljungqvist 2011. A. 50-year smoothed curve; B. A 10pt FFT smoothed
curve. Middle: IntCal09 radiocarbon calibration curve (Reimer et al. 2009) with the
approximate test wiggle-match best placement of a time series of seven 14C dates
on wood section ANG-7B according the ring counts between the sampled rings for
this wood section (not crossdated) using the D_Sequence function of OxCal 4.17
(Bronk Ramsey et al. 2001). Vertical error bars show the 1 14C date errors; horizon-
tal error bars indicate the 95.4% probability range on the wiggle-match found. The
wiggle-match indicates a relatively good overall fit to the specific shape of the cali-
bration curve, indicating that, despite likely issues of some missing or indistinct or
eccentric tree-rings, the ring count recorded cannot be more than a few decades
incorrect for the period of time represented by the branch section. The most recent
date lies off (above) the calibration curve (and so appears as an outlier), but it in fact
appears to be a good miss for the upwards wiggle in the calibration curve peaking
at ad1605 and especially if one allows for a few potential missing rings. With this
date included the series narrowly fails a Chi-square test at the 5% level (although
the OxCal Acomb value is 33.2 > An 26.7); with it excluded the series passes and has
a good OxCal Acomb value of 80.5 > An 28.9 (quoted values with curve resolution
set at 5).31 Bottom: Ring widths of ANG-7B in 100ths of a mm.
31 The 14C data, SDs, and the ring count gaps (= years) between the mid-points in order
(older to more recent, according to position in the dendro sample) are: ETH-35577, 71614bp;
(samples) with a mean sample series length of 93.33 years, but only a much
shorter interval offers a robust dendrochronology for potential climate anal-
ysis (see Figure 11). The controlling constraints on growth are not known for
these trees; for example, a series of better growth years in the ca. ad 480s
to early 490s correlate both with a period of increased solar activity (Fig-
ures 78) and decreased precipitation in central Europe as reconstructed
by Bntgen et al. (2011: Figure 4)see Figure 12 belowand generally there
is some correlation between increased growth in the Istanbul record and
decreased precipitation in central Europebut nothing strong. One fea-
ture that does not appear represented is the ad536541 event(s); although
widely represented in cold-sensitive trees in northern latitudes (Larsen et al.
2008; Baillie 2008), and in historical records from Constantinople and else-
where (Gunn 2000), this event or events does not appear to have noticeably
affected the growing conditions (negatively or positively) of the oaks shown
in Figure 11 (see also Pearson et al. 2012). The Istanbul record is of course very
short and needs to be secure and much longer to be of potential climate use.
Thus, for the present, we continue to lack a direct Mediterranean (to south-
east Europe and Black Sea) Roman-era tree-ring record, but one may exist
relatively soon.
But let us return to the most important recent tree-ring based record and
analysis for the Roman world which has been provided by Bntgen et al.
(2011), based on the integration of a very large number of tree-ring series
from central and northern Europe to provide both precipitation and tem-
perature reconstructions for the last 2500 years. Figure 12 shows the April
MayJune (AMJ) precipitation reconstruction from that paper; it was based
on 7284 oak (Quercus spp.) tree-ring series and the correlation of the recent
part of these records against instrumental data (e.g. ad 19011980 in Bntgen
et al. 2011, Figure 3), and then an extrapolation backwards from this. Fig-
ure 13 shows the JuneJulyAugust temperature reconstruction from Bnt-
gen et al. (2011) based on 1089 Stone Pine (Pinus cembra) and 457 Larch
(Larix decidua) series and their correlation against instrumental data for
ad18642003, and then again an extrapolation backwards from this.
Gap 60.5 years; ETH-35578, 616 13bp; Gap 41 years; ETH-35579, 61814bp; Gap 57 years;
ETH-35580, 466 13bp; Gap 43 years; ETH-35581, 373 13bp; Gap 56 years; ETH-35582, 30714
bp; Gap 55 years; ETH-35583, 393 14 bp. I thank Lukas Wacker, ETH, for the 14C data.
Fig. 11. Top: the mean tree-ring widths (mm) record for the tree-ring series from
Istanbul (Pearson et al. 2012; Manning et al. 2012) for the interval where the Ex-
pressed Population Signal (EPS) is above 0.85 (typically deemed a satisfactory level:
Cook and Kairiukstis 1990) (data from ARSTAN: Cook and Holmes 1999); a 20pt
(year) FFT smoothed curve is also shown. Calendar placement after Pearson et al.
(2012) and as found to be compatible with 14C data and historical information (Man-
ning et al. 2012). The residual curve showing the differences of the mean raw series
versus the 20pt smoothed curve is then shown underneath. Bottom: the recon-
structed temperature and precipitation records from central European oak time-
series shown in Bntgen et al. (2011, Figure 4) are shown for the time interval covered
by the Istanbul tree-ring data at the top of the figure. The curved lines through these
data are 60pt FFT smoothed curves. The number of samples in the Istanbul chronol-
ogy per year is shown at the very bottom of the figure.32
32 Bntgen et al. (2011) data from ftp://ftp.ncdc.noaa.gov/pub/data/paleo/treering/
reconstructions/europe/buentgen2011europe.txt.
Fig. 12. Reconstructed precipitation (AprilMayJune = AMJ) in mm with respect

to the instrumental ad19012000 period record from the Bntgen et al. (2011) study
for central Europe and for Germany (dotted linefrom the previous study of
Bntgen et al. 2010). Two lines show 10pt and 60pt FFT smoothed curves through
the Bntgen et al. (2011) data.
Fig. 13. Reconstructed summer (JuneJulyAugust = JJA) temperature anomalies

with respect to the instrumental ad19012000 period record from the Bntgen et al.
(2011) study for central Europe and for Switzerland (dotted linefrom the previous
study of Bntgen et al. 2006). Two lines show 10pt and 60pt FFT smoothed curves
through the Bntgen et al. (2011) data.
In each case we see that relatively stable and favourable climate condi-
tions changed around the mid 3rd century ad to a period of 350 years or so
marked by much greater variability and much less favourable conditions.33
There was a change towards cooler temperatures, with a notably cooler
interval in the 6th century ad, and two periods of major drops in precipita-
tion (3rd century and 6th7th centuries ad). As Bntgen et al. (2011) argue,
relatively wet and warm summers occurred during the periods of Roman
and medieval prosperity. The period of increased climate variability and the
change to drier and cooler conditions ca. ad250600 broadly coincides with
the demise of the western Roman Empire and the turmoil and changes of
what is called the Migration Period (Halsall 2007).34
We can further check on the reliability of, and especially the relevance
for the wider mid-latitudes of the Northern Hemisphere of the Bntgen et
al. (2011) reconstructions, by comparing their last 1000 years of temperature
record with the temperature reconstructions recently compiled from 40 dif-
ferent proxies from a range of source data (from ice-cores, tree-rings, sea sed-
iments, documentary sources, varved sediments, lake sediments, pollen and
a speleothem) by Christiansen and Lungqvist (2011) for the extra-tropical
Northern Hemisphere: see Figure 14. The two reconstructions show con-
siderable similarity. At the annual scale the correlation is 0.39, but when
considered as, e.g., 10pt smoothed curves as in Figure 14, the correlation rises
to a strong 0.61. This indicates that the general annual to decadal scale of
temperature variation indicated in the Bntgen et al. (2011) curve is fairly
robust for the earlier 1500 years.
If we compare the precipitation record from central and northern Europe
of Bntgen et al. (2011) with some precipitation records from the east-
ern Mediterranean (Griggs et al. 2007; Touchan et al. 2005), we see some
potentially interesting patterns: see Figure 15. There are more or less oppo-
site trends in precipitation in the period before ca. ad 1500 (so before the
LIA), but then there is a noticeable switch to similar trends in direction in
33 This finding is consonant with previous tree-ring work by Briffa (2000), which indicates
a favourable climate for northern Sweden through to ca. ad200.

34 The precipitation reconstruction speaks of a deterioration (drier conditions) ca.
ad 220340, then a reversion ca. 340460 ad, before an especially drier period ca. ad460680.
The temperature record indicates cooler weather ca. ad325430, a partial reversion ca.
ad 430530, and then an unusually cool period (similar to the worst of the LIA) ca. ad530
650. These patterns correlate reasonably (given some delays in climate processes) with the
solar activity record for the period (compare with Figures 69 above).
Fig. 14. Comparison of the temperature reconstruction for the extra-tropical North-
ern Hemisphere for the last 1000 years by Christiansen and Lungqvist (2011) with
the temperature anomalies reconstruction over the same period by Bntgen et al.
(2011; 2006).
precipitation from ca. ad1500 to ca. ad1840i.e. for the main LIA period.
This change broadly correlates with the change to less negative Palmer
Drought Series Index values in the 15th century ad and through the LIA
period in the Esper et al. (2007) record from Morocco. After the LIA, so
from ca. ad1840 and since, there are again opposite trends in direction
between the Bntgen et al. (2011) data and the Griggs et al. (2007) data (only
partly so with the Touchan et al. 2005 data). All three precipitation records
attest generally increasing precipitation in the post-Maunder Minimum
18th century ad. This might suggest that during Little Ice Age-like times
and/or periods of major solar minima the Bntgen et al. (2011) precipitation
record can indicate increasing or decreasing trends in precipitation in at
least some parts of the Mediterranean, but not necessarily at other times,
and, especially perhaps, it does not always give guidance for parts of the
eastern Mediterranean. This complicates simple transfer of the Bntgen et
Fig. 15. A comparison of the Bntgen et al. (2011) precipitation record from Central
and Northern Europe versus two east Mediterranean precipitation reconstructions
(Griggs et al. 2007 and Touchan et al. 2005) and the Palmer Drought Series Index
reconstruction by Esper et al. (2007) from Morocco. All data shown after 30pt
FFT smoothing applied. The three precipitation records (lower three lines) are
separated and scaled on a relative scale on the Y axis to make them easier to see
and compare.35
al. (2011) record to the Roman Mediterranean (whereas it is clearly directly

relevant to Roman Europe in and north of the Alps).36
35 Griggs et al. (2007) data available from http://www.ncdc.noaa.gov/paleo/metadata/
noaa-recon-6379.html.
36 This issue is very noticeable for the period around ad400700. Bntgen et al. (2011)
reconstruct a period of reducing precipitation with a major low in the 6th century adyet
Finn et al. (2011, 3168) conclude that a variety of east Mediterranean evidence appears to
indicate rather wetter conditions at this time (see also p. 3164, Fig. 6A)although one of the
high resolution records they discuss (Bereket) in fact indicates the reverse (p. 3165)high-
lighting (again) the challenges of comparing different data types and high-resolution versus
non-high-resolution data.
Speleothems
The other important source of high-resolution terrestrial climate data avail-
able from the Roman region, or from close to it, comes from the analysis
of speleothems and similar sources, primarily via the creation of 18O and
13C records. By high-resolution I mean annual to at worst decadal resolution
data; I contrast non-high-resolution sources such as century-scale informa-
tion from pollen records and related sources (such as the pollen-based data
available from Lake Van in Wick et al. 2003). For a review and discussion
of how climate proxy data are extracted from speleothems, see McDermott
(2004).37 In brief, changes in the 13C of speleothems can often be related
to variations in vegetation and soil microbial activity above the cave, and
these in turn relate to the temperature and precipitation in the region of
the cave. Typically in temperate regions, higher temperatures and increased
precipitation are expected to lead to lower 13C valuesalthough there are
numerous complications (for a discussion of such complications and the
Sofular Cave casefrom coastal northern Turkeysee Gktrk et al. 2011,
24362437). Speleothem 18O records can often be even more challenging
(Lachniet 2009). In the right conditions, the recovered 18O reflects the 18O
of precipitation in the cave catchment, and can also indicate temperature,
but there are numerous complications in many cases.
To date, only a relatively few high-resolution speleothem records have
been published which are directly relevant to the Roman world and period.38
Gktrk et al. (2011, Figure 9) provide an overview of several of the main
current records from the Mediterranean region, with only the Sofular Cave
record offering a long high-resolution Holocene dataset.39 A new record
from the Kocain Cave in southern Turkey covering the ad period is men-
tioned in Luterbacher et al. (2011; 2012), and this may be a useful additional
resource when published.
37 A useful brief on-line review can also be found at www.ncdc.noaa.gov/paleo/reports/
trieste2008/speleothems.pdf.
38 Finn et al. (2011) include the impressive high resolution records from Oman (Fleit-
mann et al. 2003; 2007). However, as Fleitmann et al. argue, the precipitation indicated by
this 18O record reflects processes primarily relating to the mean latitudinal position of the
ITCZ and the dynamics of the Indian summer monsoon. The record is thus not relevant to
the Mediterranean in any specific way. With reference to the Roman period, this speleothem
record is also rather crucially missing the temporal period between 2700 and 1312 years ago
i.e. the entire Roman world time range.
39 I do not consider the recent Soreq record ca. 2.2ka to 0.9ka published in Orland et al.
(2009) as suitable in this category, as the underlying dating precision does not in fact appear
to be entirely high resolution and the temporal coverage is not regular: see esp. Orland et al.
(2009, Table 1 and Figure 6 and comments p. 33).
However, as noted, the interpretation of speleothem records is compli-

cated and may reflect the processes of record formation, and the local envi-
ronment, more strongly than regional or wider climate. In the case of the
Sofular Cave, Fleitmann et al. (2009) demonstrate that the main source of
moisture relevant to this record is the Black Sea. Thus, as Gktrk et al. (2011,
2442) conclude: Consequently, the Sofular 18O profile show[s] no signs of
an isotopic pattern that could be ascribed to Mediterranean-derived mois-
ture. Another interesting record which seems of limited or no value for the
Roman period is the Jeita Cave speleothem from Lebanon (Verheyden et al.
2008).40
Figure 16 shows some of the other better 18O data available relevant to the
Mediterranean region for the Roman period, from two speleothem sources
(Soreq Cave, Israel = A and Bucca della Renella, in the Apuan Alps, Italy = B),
along with another 18O record (C) from analysis of planktonic foraminifera
from a core from the southeast Mediterranean just off the coast of Israel.
The Soreq Cave data (A) in Figure 16 (from Bar-Matthews et al. 2003) are
uninformative for the Roman period beyond being noticeably stable (see
Discussion below). The Bucca della Renella data are argued by Drysdale et
al. (2006) to pick up the 4200bp/2200bc drier event, and also seem to be
showing drier conditions for the 1st through 5th centuries ad with a peak
in the earlier 5th century ad. In between, it seems moister in the early 1st
millennium bc and relatively stable through to the earlier Roman period.
The Schilman et al. (2001) record is compared with the Soreq Cave data
and argued by Schilman et al. (2002; also Rosen 2007, 90) to show humid
(wetter) periods peaking around 1200bc, 700ad and 1300ad and arid peaks
at around 100bc, ad1100 and ad1700 (Schilman et al 2002; Rosen 2007, 90; see
also Schilman et al. 2001, 165, 168169). The age-depth model of the marine
record of Schilman et al. (2001) is not, however, entirely solid.41
40 For the period ca. 1000 bc to ad 900 there is a general slightly drier or fairly stable trend,
but no real signal. According to Verheyden et al. (2008, 380):

Between 3.0 and 1.1 ka, soil activity progressively decreased, as indicated by increasing
13C values. 18O values present more variability and it is therefore less clear if the 13C increase
is due to a progressive dryer climate (in which a 18O increase would be expected) or if it is to
be ascribed to a decrease in soil activity linked with increasing agriculture and/or grazing. [my
italics].
41 Its matching versus Soreq could be adjusted in at least one place; for example, there
is something of an excursion of the 14C data v. model in the early 1st millennium bc (depth
300350 cm) if one studies Schilman et al. (2001, Figure 3).
Fig. 16. Three 18O plots from Mediterranean locations. They are plotted so wetter
is to the top and drier is to the bottom. A: Soreq Cave from Bar-Matthews et al.
(2003); B: Bucca della Renella from Drysdale et al. (2006); and C. the record from
planktonic foraminifera (Globigerinoides ruber) from the southeast Mediterranean
off Israel (Schilman et al. 2001).42
Figure 17 shows some of the available 13C records from speleothems from
the Mediterranean relevant to the Roman period. Martn-Chivelet et al.
(2011) model their impressive high resolution record from North Spain as
a temperature proxy over the last four millennia. They find the Iron Age
Cold Period cooling event around and following ca. 800 bc, then warmer
conditions ca. 500bc to ad300 with a peak ca. 150 bc to ad 250 (the Roman
Warm Period). There is then a trend to cooler conditions ca. ad 350600
with a cool minimum around ad500, before, clearly indicated, the sub-
42 Data available from: ftp://ftp.ncdc.noaa.gov/pub/data/paleo/speleothem/europe/
italy/renella2006.txt; ftp://ftp.ncdc.noaa.gov/pub/data/paleo/speleothem/israel/soreq_
peqiin_2003.txt; and from Schilman et al. (2001, Table 1).
sequent warm conditions of the MCA, and, after this, the cooler period
of the LIA (see Martn-Chivelet et al. 2011, 910 for summary and refer-
ences to other literature). The Bucca della Renella record (Drysdale et al.
2006) indicates moister conditions in the first half of the 1st millennium
bc and then drier conditions from ca. 500bc to ad 300, with, just as the
record ends, a suggestion of wetter conditions from the 4th century ad. The
high-resolution Sofular Cave record (Fleitmann et al. 2009) shows a dip
to wetter conditions in the 9th century bc, and then a fairly stable record
through to the mid 1st millennium bc. There is a gradual trend to drier
conditions overall, but largely a very stable record through to just after
ad100, when a slight change to a wetter trend occurs through to around
ad310. The stability of the main Greco-Roman period is a feature (as again
during the period of the MCA, especially 10th13th centuries ad). From
ad310 there is a clear shift to drier conditions to about ad 430, and then
a clear shift from about ad430 through about ad 640 to wetter conditions
(peaking around ad600640), before then shifting back to a drier trend
from the mid 7th century ad.43 The Late Roman shifts are noticeable as
major events in the relatively placid Sofular Cave Late Holocene record (see
the Inset to Figure 17 for a detailed view). The Soreq Cave record in Bar-
Matthews et al. (2003) is not very high resolution, but indicates a fairly
stable Greco-Roman period, and a slight up/down change in the 57th
centuries ad matches the Sofular Cave record. Whereas the Soreq Cave
record in Schilman et al. (2002) also shows a humid/moister trend in the
1st7th centuries ad, and Bar-Matthews and Ayalon (2004; Rosen 2007, Fig-
ure 5.3) derive an increasing rainfall record for the 2nd7th centuries ad
from the Soreq Cave records, we may note that the analysis in Orland et
al. (2009), employing a different methodology, argues for the exact oppo-
site, with decreasing rainfall ad100700 (with steep drops around ad 100 and
ad400).
43 There is an additional stalagmite 13C effective moisture record from the Kocain Cave
in southern Turkey shown by Luterbacher et al. (2011, Figure 1B), from work by Gktrk et al.
However, it would appear that the graph is shown reversed (or the arrow for precipitation
should point down) in Luterbacher et al. (2011, Figure 1B). On this assumption, this record
seems to show stable conditions to after ad 100, then a shift to a drier trend to ad300, then
wetter to about ad 400, a short drier shift in the early 5th century ad, and then a generally
wetter period from the mid-5th century to late 6th century ad, before a major drier interval
ca. ad 600800. (Note: the corrected version of the Kocain record showing the above may be
found in Luterbacher et al. 2012: Figure 2.7.)
Fig. 17. Four 13C records from Mediterranean region speleothems from Martn-
Chivelet et al. (2011); Drysdale et al. (2006); Fleitmann et al. (2009) and Bar-
Matthews et al. (2003). The Inset shows a detail ad250750 of the Sofular Cave 13C
record. Wetter (or cooler for North Spain dataset) is downwards; drier (or hotter for
North Spain dataset) is upwards.44
If we compare some proxy temperature records from two speleothems

with available tree-ring records from central Europe (see Figure 18a), we
see some general patterns, but also shorter-term variations between loca-
tions and data sources. All the records indicate a generally stable main
Roman period through to around the earlier 3rd century ad, and then
changes to cooler temperatures in the 35th centuries ad but with some
reversal in either the 4th or 5th centuries ad (varying by record). The two
44 For the North Spain dataset of Martn-Chivelet et al. (2011), see ftp://ftp.ncdc.noaa.gov/
pub/data/paleo/speleothem/europe/spain/nspain2011d13ct.txt; and for the Sofular Cave

dataset of Fleitmann et al. (2009), see ftp://ftp.ncdc.noaa.gov/pub/data/paleo/speleothem/
asia/turkey/sofular2009.txt.
speleothem-based records and the two tree-ring-based records usually cor-

relate fairly closelyallowing for some minor lagswith the solar activity
proxy offered by the 14C production model (Marmod09 from Reimer et al.
2009).45 The ca. 360bc major solar minimum seems to be picked up in the
available north Spain and Bntgen et al. (2011) records; the major solar mini-
mum ca. ad685 is in the Spannagel and north Spain records but seems only a
small deviation in the Bntgen et al. (2011) record; the ca. ad 1040 solar mini-
mum seems to be represented in all records as is the ad 1305 solar minimum;
the ad1470 solar minimum is in all except the north Spain record; and the
ad1680 solar minimum is in all records. The solar maximum 445 bc is in the
available north Spain and Bntgen et al. (2011) records, and the sustained
low 14C production (= active sun) period from the mid 8th to mid 13th cen-
turies ad (the MCA) is clearly visible in all records, as is the subsequent LIA
(dates for major solar minima/maxima from Usoskin et al. 2007see also
Figure 6b above). There is thus some encouraging consistency in general
findings. Perhaps the key feature we can note across several records is the
relative stability of these favourably warm conditions in the main Roman
period compared with later times.
The least coherent narrative comes from the two alpine temperature
reconstructions in Figure 18a: comparing the Bntgen et al. (2011) temper-
ature record from central (alpine) Europe with the Spannagel speleothem
also from the central Alps (for the specific comparison of just these two
datasets, see Figure 18b). Since the two reconstructions are both for temper-
ature, and come from the same area, one might expect them to be relatively
similar. But, although these two records do offer some general similarities
in trends (so general direction and shape of the curves)give or take some
flexibility on about a decadal scalefor some periods like the MCA (e.g.
ad8001300), or again around ad15001700, at several other times there is
no correlation, or even an opposite one (e.g. ad 200400, around ad 1400,
early 19th century ad). In the Roman period the two records indicate sim-
ilar trends to warming conditions in the 5th century ad, before significant
cooling in much of the 6th century ad, and then warming conditions from
the late 6th or start of the 7th century adat other times there is less agree-
ment in trends.
45 Note that the Marmod09 curve is inverted in Figure 18a so that production peaks (= less
active sun and so likely cooler conditions) point downwards.

Fig. 18a. Comparison of two temperature proxy records derived from speleothems
(a record from northern Spain from Martn-Chivelet et al. (2011) and a record from
Spannagel Cave in Austria from Mangini et al. 2005) with the proxy temperature
records for central Europe and for Switzerland from tree rings produced by Bntgen
et al. (2011; 2006). Top: the Marmod09 14C production model (Reimer et al. 2009).46
46 For the Spannagel Cave data of Mangini et al. (2005), see ftp://ftp.ncdc.noaa.gov/pub/
data/paleo/speleothem/europe/austria/spannagel2005.txt.
Fig. 18b. Detail from Figure 18a of the summer temperature reconstructions from
(top) the central Alps speleothem from Spannagel Cave (Mangini et al. 2005) with
(bottom) the central European temperature reconstruction based on tree-rings in
Bntgen et al. (2011) (see also Figure 13).
Discussion
No dramatic or major global climate events are currently posited for the
main Roman periodsay from the 3rd century bc to the 5th century ad.
Contrast for example the well-known and much discussed climate event
82008000 years ago (e.g. Cheng et al. 2009; Rohling and Plike 2005; Alley
and gstsdttir 2005), or any other of the major or rapid Holocene climate
change events noted in the literature (see e.g. Mayewski et al. 2004; Bond
et al. 2001; Weiss 2000; Staubwasser and Weiss 2006; Thompson et al. 2002;
2006; Bar-Matthews and Ayalon 2011; Chew 2002; Weninger et al. 2009; etc.).
The main Roman period lies after the pair of grand solar minima of
ca. 765bc and 360bc and the Ice Rafted Debris (IRD) event 2 of ca. 700bc
(Bond et al. 1997; 2001),47 and before the (next) grand solar minimum ca.
47 Ice Rafted Debris (IRD) events refer to periods when there seems to be an increase in
ad685 and the IRD 1 event around ad500 (Bond et al. 1997; 2001) and the
less favourable later 1st millennium ad time period (McCormick et al. 2007).
It appears as a relative sustained era of stability, and can be contrasted with
the scale of changes evident in the 2nd millennium ad during the MCA or
the subsequent LIA (Bradley et al. 1993; Le Roy Ladurie 1971).
The main Roman period is also conspicuous for a relative dearth of large
climatically effective volcanic eruptions; indeed, it was an extended period
of relative volcanic calm. There was no period of enhanced major volcanic
activity or sets of eruptions of the sort which can be argued to change
climatecontrast the arguments for just this type of scenario for the start of
the LIA (Miller et al. 2012); nor were there several volcanic eruptions across
a time period which might be argued to cause short-term climate forcings
of the sort suggested for the 8th10th century ad period by McCormick et
al. (2007). To illustrate the situation, Figure 19a shows the volcanic erup-
tion trace records from two Greenland ice-cores.48 There was a large erup-
tion around 50bc (likely not of high northern latitude49), and then, for the
next several centuries through to ad500, there was a low level of major vol-
canic activity compared to periods either before or afterwards. The Bristle-
cone Pine (BCP) tree-ring record offers another reasonable volcanic activity
proxy record for the northern hemisphere (Salzer and Hughes 2007). Fig-
ure 19b shows the number of events per century between the 10th century bc
and the 10th century ad where there was a ring-width minima year that cor-
responds with a volcanic signal in an ice-core (allowing the ice-core signals
to be dated as 5 years of ring-width minima) from Salzer and Hughes (2007,
Table 2). Certain centuries stand out as volcanically active to a significant
the discharges of icebergs which travel some distance into the sub-polar north Atlanticthe
assumption is that this increased iceberg survivability occurs because of relatively cold water
temperatures. This phenomenon was studied in particular by Gerard Bond, who further
found that the source of the icebergs could be determined from petrologic tracers, and that
the pattern of such tracers being found at some distance from source (indicating a notable
ice-rafting event) corresponded with cooler climate episodes every few thousand years (Bond
and Lotti 1995; Bond et al. 1997). Bond et al. (2001) further showed that the pattern of
increased discharge events (cooler episodes) corresponded with a long-term pattern (roughly
a 1500-year cycle) of solar activity through the Holocene.
48 One or two major volcanic events in the 530s ad to ca. ad541 should be added to the
records shown in Figure 19a (see Larsen et al. 2008; Baillie 2008). It must also be remembered
that the GISP2 ice-core was unluckily, but crucially, missing several meters of core around
this time, which is probably why there is no record in this core of a major 6th century ad
volcanic eruption (as first highlighted by Baillie 1994, 216).
49 Clausen et al. (1997, 26721). Some potential candidate volcanic eruptions are listed
by the Global Volcanism Program on their website at http://www.volcano.si.edu/world/

largeeruptions.cfm.
Fig. 19a. Two records of past major volcanism relevant to the Northern Hemisphere
from two Greenland ice-cores (GISP2see e.g. Zielinksi et al. 1996; and GRIPsee
e.g. Clausen et al. 1997). The timescales of the two ice-core records agree fairly well
(and with independent markers) from the present back through the 1st millennium
bc (so to before the Roman period)whereas there are issues before this point (see
review and discussion of Southon 2004).50
degree, for example the 5th century bc and the 6th, and especially the 7th,
and 10th centuries ad. Salzer and Hughes (2007, Table 4) note some packages
of years with notable growth decreases and volcanic signals in the ice-core
records which are indicated by the grey diamonds at the top of Figure 19b:
425419bc, 282280bc, 4236bc, ad536547, ad 687698 and ad 899903.
The period from ca. 35bc to ad535 (= 570 consecutive years; and, excluding
the mid-1st century bc event, the period 279bc to ad 535 = 814 consecutive
years) is conspicuous for the lack of evidence for major volcanic activity
50 Data with further references from: ftp://ftp.ncdc.noaa.gov/pub/data/paleo/icecore/
greenland/summit/gisp2/chem/volcano.txt and ftp://ftp.ncdc.noaa.gov/pub/data/paleo/

icecore/greenland/summit/grip/chem/gripacid.txt.
Fig. 19b. Bottom: number of ring-width growth minima in the Bristlecone pine (BCP)
record of Salzer and Hughes (2007) associated within 5 years of a volcanic signal in
an ice-core record (from Salzer and Hughes 2007, Table 2). Top: Packages of years
where there are notable decreases in Bristlecone pine growth and ice-core volcanic
eruption signals (from Salzer and Hughes 2007, Table 4).
relevant to the mid latitudes of the northern hemisphere. There were of

course volcanic eruptionsthese happen regularlyand some during the
Roman period had major impacts on human societies in different areas of
the world (such as the eruption of Popocatpetl in central Mexico about
2000 years bp: Plunket and Uruuela 2006; or, of course, Vesuvius in Italy in
ad79: Sigurdsson and Carey 2002). But there is a notable lack of evidence
for major climatically effective (thus climate forcing) volcanic eruptions
relevant to the area of the Roman world over most of the Roman period until
the 6th century ad.
In sum, the Roman period seems to be a relative era of stability com-

pared to the subsequent noisy 1500 years, and many of the periods that
went before. It is also positiverelatively warmas well as stable, and
might be considered something of an optimum for much of the Europe-
Mediterranean region. Not surprisingly, the Greco-Roman period as a whole
has indeed been regarded for some time as something of a climate opti-
mum (e.g. Perry and Hsu 2000; Denton and Karln 1973). This era of stabil-
ity was noted by Bar-Matthews et al. (1998), who observed from the Israeli
speleothem record that the period from the end of the 2nd millennium bc
to the late 1st millennium ad had the most stable 18O and 13C values in
the past 6500 yearsand hence stable rainfall (noted also by Rosen 2007,
91, 168 and her Fig. 5.3 on p. 82). Other indications are generally consis-
tent with favourable conditions. For example, Ferrio et al. (2006) report a
humid phase ca. 300bc to 300ad in northern Iberia from archaeological
charcoal analysis, and, based on a the lake varve study, Martn-Puertas et
al. (2009; 2008) report similar findings of humid/moist conditions broadly
600bc to ad400 (the Iberian-Roman Humid Period). There is something of
a drier (more arid) dip in the middle ca. 190bc to ad 150 which might cor-
respond with the reconstructed temperature dips around this time in the
central European data of Bntgen et al. (2011), centered ca. 40 bc, and in
the north Spain data of Martn-Chivelet et al. (2011)see Figure 18a. Oth-
erwise, Bntgen et al. (2011) find generally positive precipitation and tem-
perature conditions for central and northern Europe ca. 300 bc to ad 200.
Higher levels of fluvial activity (increased flood intensities) are also noted
for the Roman High Empire period (roughly 1st century bc through 2nd
century ad) in studies of French rivers (Arnaud-Fassetta et al. 2010, 101, Fig-
ures 9, 12), and similar findings are reported from several western to cen-
tral Mediterranean areas (Luterbacher et al. 2012). McCormick, Bntgen et
al. (2012) highlight that Nile flood data indicate good conditions in gen-
eral for Egypt especially from 30bc to ad164the height of the Roman
worldand then a deterioration; and Dead Sea levels suggest a wet era in
the century or two either side of the bc/ad transition (Bookman et al. 2004;
Migowski et al. 2006), as does isotope analysis of wood from 1st century ad
Masada by Yakir et al. (1994; Issar and Yakir 1997but cf. Lev-Yadun et al.
2010).
If we look more widely at the northern hemisphere, the central Green-
land temperature proxy available from the GISP2 ice-core (Alley 2000), and,
for the last 2000 years, the multi-proxy based Arctic summer temperature
reconstruction of Kaufman et al. (2009), reveal that the Greco-Roman world
enjoyed relatively warm temperatures not very dissimilar to the very late
2nd millennium ad, and potentially as warm as, or warmer, than the MCA
at least for high northern latitudes: see Figure 20.51 The Sea Surface Tem-
perature proxy of Keigwin (1996, Figure 4) from the northern Sargasso Sea
provides a comparison from the lower mid latitudes of the Atlantic region,
and indicates broadly a similar pattern.52
Unfortunately, there is never complete consistency in our data, or at least
there is something of difference between the west and east Mediterranean.
Despite a fairly general picture of a warm and/or moist regime in the earlier
Roman period in many areas, some of the available eastern Mediterranean
records, especially, indicate gradually increasing dryness into and through
the earlier Roman period. For example, the Lake Van oxygen isotope record
(Wick et al. 2003, Figure 4) indicates increasing aridity from the end of the
3rd millennium bc through to a peak ca. 110bc (2100 bp from ad 1990 in the
Wick et al. 2003 plot),53 as does the Sofular Cave 13C record less dramati-
cally (see Figure 17 above). Schilman et al. (2002), using Soreq Cave data and
analysis of foraminifera from an east Mediterranean core, also identify an
arid peak ca. 100bc (and then a change to more humid conditions). These
records are consistent with a warm period, but with regionally more nega-
tive outcomes.
It has been suggested by many scholars that climate change was a factor,
even a significant factor, in the decline and re-organisation of the later
Roman world (e.g. Issar and Zohar 2007; Bntgen et al. 2011; McCormick,
Bntgen et al. 2012). The review of the Figures and other literature cited to
this point has indeed found evidence of some substantive changes, but they
are not all consistent. The available temperature proxies indicate a general
decline in temperatures from the 3rd to the 7th centuries ad, but with some
showing a short reversal variously in parts of the 4th to early 6th centuries
51 Humlum et al. (2011, Figure 8) achieve generally similar model results from hindcasting.
The period from the 8th century bc through to the 1st century bc saw a rising temperature
trend with a peak period (the Roman Warm Period) in the 4th1st centuries bc. The Green-
land temperature record (Figure 20) then shows the change to the general major cooling
trend in the 5th8th centuries ad (and especially the 7th8th centuries ad), the subsequent
warmer period of the MCA at the end of the 1st millennium ad and the start of the 2nd mil-
lennium ad, before the LIA era which only ended with the warming of the past 150 years.
52 The period ca. 300 bc to ad 200 sees fairly stable and relatively warm conditions, before
cooling ca. ad 200500 (and then the warming trend of the MCA, and subsequent cooling of
the LIA).
53 The earlier analysis of Lake Van data by Lemcke and Sturm (1997, Figure 5) also indi-
cates relatively dry conditions in the last couple of centuries of the 1st millennium bc, before
a sharp change to more humid conditions in the 1st century ad.
Fig. 20. Two temperature reconstructions for high northern latitudes of the north-
ern hemisphere. The plot in grey is the central Greenland temperature proxy record
derived from the GISP2 ice-core based on stable isotope analysis, and ice accumula-
tion data (Alley 2000; Cuffey and Clow 1997). The plot in black is the reconstructed
Arctic summer temperature proxy derived from a multi-proxy suite of datasets in
Kaufman et al. (2009).54
ad.55 The general solar forcing record offers some correlation. It indicates a
quite active sun around ad300 and thus likely a fairly warm 4th century ad
as this activity declined, then a cooler period ca. ad 370435 as solar activity
reduces significantly, before a recovery ca. ad435515, and then a return to
significant cooling for most of the 6th and 7th centuries ad: see Figures 69.
However, our information about precipitation in the Roman world from
the first century ad onwards is more difficult to synthesize, since some of
54 GISP2 data (Alley 2000) from ftp://ftp.ncdc.noaa.gov/pub/data/paleo/icecore/
greenland/summit/gisp2/isotopes/gisp2_temp_accum_alley2000.txt. Kaufman et al. (2009)

data from http://www.ncdc.noaa.gov/paleo/pubs/kaufman2009/kaufman2009.html.
55 A range of other evidence reviewed by McCormick, Bntgen et al. (2012), ranging from
glacier advance histories to insect remains, supports some 4th to 5th century ad warming.
it seems contradictory (and especially in the southern Levantsee for one

recent summary Rambeau and Black 2011, 99100which may well reflect
climate reality as different processes lead to precipitation here versus in the
northern Mediterranean and further west: Black 2011). Bar-Matthews and
Ayalon (2004) model increasing or positive precipitation ca. 600200 bc on
the basis of the Soreq Cave data, and then more arid conditions ca. 200 bc to
ad200. As noted above, Schilman et al. (2002) see an arid peak ca. 100 bc and
then increasingly moister (humid) conditions to a peak about ad 700. This
is in line with a humid/moist phase in the first half of the 1st millennium
ad in the Black Sea record of Lamy et al. (2006), and the evidence of the
Lake Van 13C studies (Lemcke and Sturm 1997, Figure 5; Wick et al. 2003,
Figure 4) indicating an arid peak at the end of the 2nd century bc followed by
more humid conditions for the next century or so. But the more recent study
of Orland et al. (2009) on Soreq Cave material literally finds the opposite
to the Schilman et al. (2002) paper, with a major arid trend ad 100700
(whereas Bar-Matthews and Ayalon 2004 had precipitation start to increase,
slowly, from around ad200 and especially in the 4th to 7th centuries ad).
Ehrmann et al. (2007) working on Aegean marine cores also find indications
of likely drier conditions from the mid-2nd century ad peaking around
ad650. The Dead Sea records are halfway between the above two opposites
(following Bookman et al. 2004; Migowski et al. 2006; Rosen 2007, 9394 and
Figure 5.6). There is a wet phase around the end of the bc period and start
of the ad period (so contra the arid peak ca. 100 bc in Schilman et al. 2002),
and then an arid trend (like Orland et al. 2009 and contra Schilman et al.
2002) before a short sharp reversal either in the 4th century ad (Bookman
et al. 2004) or more with the 5th century ad (Migowski et al. 2006)the
opposite of the arid trend in Orland et al. (2009) and especially their sharp
arid trend ca. ad400but consonant with indications of a change to wetter
conditions ca. 450650ad in southwest Turkey reported by Kaniewski et al.
(2007, 2212), or, more coarsely, around the 4th to 8th centuries ad in the
record of Lemcke and Sturm (1997, Figure 5). Migowski et al. (2006, 426)
report indications of arid conditions from about ad 500, and the Dead Sea
water-level reconstructions place a peak level either ca. ad 500 or ca. ad 600
(Migowski et al. 2006, Figure 3 comparing their data with Bookman et al.
2004), before a long-term arid trend (Bar-Matthews and Ayalon 2004 model
precipitation as falling ca. ad600 to 1000). A coherent synthesis is clearly
impossible, as some data are simply contradictorywhether in terms of
dating, or the trends indicated. Whereas the synthesis of Wanner et al.
(2011, Figure 4e) regards the period ad300500 ad as drier in the Levant,
the more widely based synthesis assessment of the southern Levant by
Rambeau and Black (2011, Figure 7.2) favours relatively moist conditions
ca. 200bc to ad150, and then wetter conditions again ca. ad 350550 (thus
contra the Wanner et al. 2011, Figure 4e synthesis), and then drier conditions
ca. ad550750. I use the Rambeau and Black (2011) assessment below in
Figure 21.
It is fair to say that the overall picture from the published literature is
less than clear! Finn et al. (2011, 3168) come down in favour of a generally
moister period ca. ad1600 in the eastern Mediterranean, but their litera-
ture review also notes contrary evidence. It is of course not obvious how
many of these variations and contradictions are real, and how much they
are merely a product of the relatively loose, approximate, or potentially flex-
ible chronologies which underlie several of the non-high-resolution to fairly
low-resolution records. Some differences between regions are also likely, e.g.
Aegean versus southern Levant, based on modern weather systems (Black
2011). Perhaps the main conclusion (adding the records noted in the next
paragraph) is that the relative stability and coherence and positive condi-
tions of the earlier Roman period seems to have ended as one moves beyond
the 1st to 2nd centuries ad; instead, there seems both more variation in
records, as well as some indications of aridity especially in the east.
There are some indications of changing climate regimes in the later
Roman empire in the east. There is evidence for a regional long-lasting
drought in central Turkey ca. 400540ad from the study of a diatom record
from Nar Gl (Nar Crater Lake) by Woodbridge and Roberts (2011), and
generally for arid conditions from the mid 4th to early 6th centuries ad
(Jones et al. 2006). The environmental study in the Bereket Basin, southwest
Turkey, by Kaniewski et al. (2007) found indications of more arid conditions
ca. 40bc to about ad450, but then found moister conditions ca. ad 450650
(as noted above). Sofular Cave shows drier conditions in the mid-4th to
mid-5th centuries ad, and then a shift to wetter conditions in the later 5th
to mid 7th centuries ad (Figure 17). And, looking more widely, likely more
arid/drought conditions peaking in the 5th century ad (1500 cal bp from
ad1950) are reported from Iran (Stevens et al. 2006), and there are indica-
tions of increased aridity in the mid 4th and 5th7th centuries ad across
central Asia based on the dendroclimatic study of Sheppard et al. (2004),
which found long dry/drought periods ad426500, 526575, 626700, and
cite generally consonant indications from other sources (pp. 875, 877), and
from other recent studies in western China (e.g. Qin et al. 2012).56 There
56 Cook (this volume) provides additional analysis and evidence of arid episodes in the
appears something of a contrast (opposite regimes) between Anatolia and

the southern Levant, therefore, in some of the period around the mid 4th to
mid-5th centuries ad, with Anatolia dry and the southern Levant becoming
wetter. Both areas seem to enjoy more favourable conditions (from several
records, but not all) in the mid-5th to mid-6th centuries ad. The southern
Levant then sees a sharp change to more arid conditions, somewhere by dur-
ing the 6th century ad and certainly by about ad 600 (see discussion above,
and see Rosen 2007, 150171; Rambeau and Black 2011, 100). Anatolia seems
to enjoy more favourable wetter conditions for longer, through the mid 7th
century ad, before a shift to more arid conditions.
The period around ad400600 represents a time of wider climate reorga-
nization. This is the period of cooling in the north Atlantic (and IRD event
1) (Bond et al. 2001; 1997), cold temperatures around Iceland (Patterson et
al. 2010), glacier advance in the Alps (Holzhauser et al. 2005), and lower sea
surface salinity in the northeast Caribbean (Nyberg et al. 2010), as occurs
again around the start of the LIA. In the northwest Mediterranean this is a
stormier period (Sabatier et al. 2012, Figure 9c), and such stormier periods
broadly correspond to the north Atlantic IRD and associated episodes over
the Holocene. Increased fluvial (flood) activity occurred in several French
rivers in the late antique period, ca. ad450700 (Arnaud-Fassetta et al. 2010,
Figures 9, 12), and ca. ad500700 more widely across the Mediterranean
(Luterbacher et al. 2012). The differential climate histories of Anatolia and
the southern Levant in the later Roman period may thus represent a version
of the Mediterranean see-saw at work. In the MCA Spanish lakes generally
indicate aridity but then wetter conditions during the LIA (Corella et al. 2011,
364365), whereas the opposite is reported from central Anatolia from the
Nar Crater Lake (Roberts et al. 2012). Such patterns may also occur in the first
millennium ad. The Zoar Lake record indicates drier conditions ca. 190bc
to ad150, matching the later Roman Warm Period (Martn-Puertas et al.
2009),57 before a shift to a more humid period ca. ad 150350 (end of record
reported) which perhaps matches some of the stormier period in the north-
west Mediterranean seabed record of Sabatier et al. (2012). But this same
period (and especially the 4th to mid-5th centuries ad) sees evidence of arid-
ity in central Anatolia (and in contrast moister conditions in the southern
4th, 5th and 6th centuries ad in central Asia, and suggests a possible long-range association
with the El Nino-Southern Oscillation (ENSO) in the equatorial Pacific. See also discussion
in McCormick, Bntgen et al. (2012).
57 In view of parallel changes in other records, Martn-Puertas et al. (2009, 117118) see
their lowered lake levels at this time as mainly climate driven, but accept some possible
anthropogenic element from human pressures during the Roman period.
Levant).58 As precipitation drops in the later 5th6th centuries ad in central

Europe (Figure 12) it appears to get wetter in parts of Anatolia (above).
I conclude with an attempt to synthesize the data reviewed into a sum-
mary by Roman time periods.59 Refer to Figure 21, especially, unless other-
wise noted.
200bc to ad200: Broadly stable solar activity, see Figures 68 (drop and then
recovery in 1st century bcespecially in 10Be recordand an up/down in
2nd century ad); generally relatively warm but with a dip in the 1st cen-
tury bc (also in central Greenland ice-core record: see Figure 20 above) and
glaciers in recession; relatively moist in the western empire and the south-
ern Levant (Central Europe trees, French rivers, Southern Levant, general
Iberian charcoal 13C and in southern Spain the Zoar Lake overview
with, Zoar Lake, slightly less moist/drier period ca. 190bc to ad 150), but
drier in Anatolia except in 1st century bc (when there seems a linkage with
the marked temperature drop). Thus most areas positive including Anato-
lia around the 1st century bcwhich then returns to drying trend in the
ad period, while the western empire regions and the southern Levant are
largely positive through the 2nd century ad.
ad200 to 300: Warm generally (solar activity starts relatively high and then
declines: Figures 68), and change (unfavourable) to drier conditions in
central, and some of southern, Europe, and in the eastern empire in general;
Spain in general still humid/moist (see also Figure 18a).
ad300 to 430: Temperature trends seem to vary by region or be inconsistent.

The central European tree data indicates a clear cooler shift. However, the
Spannagel Cave in Austria indicates the opposite with warming through the
4th century ad to a peak about ad400 (see Figure 18a above), and there
is also a rising temperature spike in the last couple of decades of the 4th
58 Such differences can be observed or reconstructed at periods in the more recent
past. For example in the post-Maunder Minimum period ad17001750, Nicault et al. (2008,
Figure 9) reconstruct wet conditions for most of the Mediterranean including the southern
Levant except for central Anatolia. In support, Xoplaki (2002) finds that whereas some regions
of Anatolia show a weak correlation with a positive NAO, the southern Levant (and much of
the Mediterranean) exhibits a negative correlation.
59 The following summary by time periodsalthough varying in some of the details and
data employed and emphasizedbenefits from, and follows, the time horizon approach
employed in McCormick, Bntgen et al. (2012), and is indebted to this larger and more
comprehensive study.
century ad to a peak about ad400 in the north Spain speleothem data (Fig-
ure 18a). Solar activity is relatively high until late in the 4th century ad
(Figures 68). We might wonder if these two rises in reconstructed tempera-
tures perhaps also relate to increased precipitation, since this time interval is
within the last part of the general humid/moist phase recognized in south-
ern Spain by Martn-Puertas et al. (2009; 2008), and as there is increased
precipitation evident at this time in central Europe (cf. Mangini et al. 2005,
746747). Although there is not a clear correlation in the north Spain record
between 13C and precipitation (Martn-Chivelet et al. 2011, 6), there is some-
times something of a general correlation in trend based on visual observa-
tion of Figure 5 in Martn-Chivelet et al. (2011). It is wetter in central Europe
and the southern Levant, but dry in Anatolianote especially the opposite
trends in Figure 21 between central Europe precipitation (D) versus Sofular
Cave 13C (E, where effective moisture = precipitation is upwards for drier
and downwards for wetterthus opposite in direction to D). The overall
scenario seems similar to the general LIA pattern where a change to cooler
conditions (and more negative NAO) means wetter in Europe and west-
ern/northern Mediterranean, but drier in Anatolia. Thus we might surmise
somewhat more positive conditions in the western empire, and, by later 4th
century ad, in the southern Levant, but less positive conditions in Anato-
lia.
ad430 to 500: A little warmer in central Europe but still cool versus typical
range 200bc to ad300, and there is a more active sun through this inter-
val (Figures 68, 13, 18a). It is notably drier in central Europe, but wetter
at Sofular Cave in northwest Anatolia and from ca. ad 450 in the Bereket
Basin in southern Anatoliathese are the biggest and opposite shifts (from
wet to dry and dry to wet respectively) in both central Europe and Anatolia
evident in the timeframe under reviewbut it remains dry at Nar Crater
Lake in central Anatolia. The north Spain speleothem indicates warmer
temperatures (Figure 18a), and this might also indicate somewhat more
humid/moist conditions (but no other clear evidence), which would be con-
sistent with indications from southern France of increased fluvial activity.
There are (still) wetter conditions in the southern Levant.
ad500 to 550: Markedly colder in central Europe (see also Figures 13, 18a)
(also Spannagel Cave record: Figure 18a60), and solar activity declines sharply
60 The Spannagel, Austria, speleothem record indicates falling temperatures for the 6th
from a peak. This is a cold period in the north Atlantic (Figure 20 and text
above), and a time of widespread glacier advance in the Alps. This was also a
very dry period (driest 50-year period in timeframe under review) in central
Europe (see also Figures 12, 18a). An opposite pattern continues with wet-
ter conditions at Solfular Cave and in some of the rest of Anatolia (Bereket
Basin)but not everywhere, as this is the last few decades (to ca. ad 540)
of dry conditions at Nar Crater Lake. The southern Levant remains wet-
ter to about this interval of timeto when exactly is not clear from the
less than precisely dated palaeoclimate data. Historical and archaeological
data indicate drier times in Palestine and parts of the eastern Roman world
(McCormick, Bntgen et al. 2012), so it may be that drier conditions start
even by the earlier 6th century ad in some of this region. There is fluvial
(flood) activity in southern France and the final period of the earlier 1st mil-
lennium stormier phase in the northwest Mediterranean. There is a gradual
apparent temperature increase in the north Spain speleothem record (Fig-
ure 18a)it is not known if this may also correspond to somewhat moister
conditions (suggested by records in previous sentence). The last part of this
interval correlates with the widespread effects of what seems to be one or
two large volcanic eruptions, perhaps ca. ad536 and 541 (Baillie 2008 with
Larson et al. 2008), and a package of poor conditions which seem plausibly
associated across the period ca. ad536550 (Gunn 2000). The Justinianic
Plague breaks out ad541 (Little 2006; Stathakopoulos 2000), and spreads,
and it is as yet unclear whether this links to generally stressed times, or to a
specifically changing climate context.
ad550 to 600: Still cold and fairly dry in central Europe (see also Figures 13,
18a), and the 7th century ad is a time of glacier advance in the Alps and
colder northern hemisphere temperatures (e.g. Figure 20 above; Spannagel
Cave in Figure 18a above). Arid conditions have definitely started in the
southern Levant ending the more positive late Roman period. But Anatolia
(in general) now seems to be enjoying wetter conditions (Sofular Cave,
Nar Crater Lake, Bereket Basin). It is probably wetter in southern France.
Rising temperatures are reconstructed from the north Spain speleothem
(also Figure 18a).
century ad (and again in the earlier 7th through mid-8th centuries ad), but the North Spain
speleothem indicates the opposite, only showing a fall in the 7th century ad: see Figure 18
above. There are thus some differences or contradictions comparing the tree-ring versus
available speleothem derived data.
ad600 to 700: As Figure 21 summarises, temperatures start cold but substan-

tially improve through the century in central Europe (see also Figures 13,
18aalthough Spannagel Cave indicates warming in the earlier part of
interval, it then changes to cooling from mid 7th century ad in contrast to
the central European tree data reconstruction: Figure 18a above). Precipi-
tation remains fairly poor (central Europe) (also Figures 12, 18a). The north
Spain speleothem indicates falling temperatures therebut this may also
link to decreased precipitationand this line might offer something of an
explanatory route for the contrasting records from central European trees
versus the two speleothem records just noted. Wet peak (in the timeframe
under review) in the Sofular Cave record through to around ad 670, and else-
where in Anatolia conditions seem wetter also. Fluvial activity attested in
southern French and other Mediterranean rivers. But dry in the southern
Levant.
ad700 to 800: Temperatures return to average and above average in central

Europe (also Figures 13, 18a); precipitation improves towards average in
central Europe (also Figures 12, 18a); and temperatures rise and perhaps
precipitation improves in the north Spain speleothem (Spannagel Cave is
the exception with its temperature record falling until after the mid 8th
century adit only starts its long MCA climb about ad 760: Figure 18a
above). But the favorable wetter regime seems to end in Anatolia. Sofular
Cave sees a drier trend from about ad670 and a dry phase right through
the 8th century ad, and the wet phases end by ca. ad 700 in the other
available records from Anatolia. Dry in the southern Levant. Thus the 8th
century ad sees a return to more positive conditions in much of the west
of the Roman world, but less favourable conditions in most of the eastern
Roman world.
Conclusions
We are a long way from a detailed climate map through time for the Roman
world, but more and more, higher-quality and diverse data are being pro-
duced at a rapid rate. We may expect major progress within the foresee-
able future. At present, the majority of proxy datasets from the Mediter-
ranean region reflect changes in the hydrological cycle (critically relevant
for relatively arid regions), whereas, with two high-resolution exceptions
(from north Spain and the Austrian Alps: see Figure 18awhich may not
really reflect much of the Mediterranean region), and some fairly coarse-
Fig. 21. A comparison of a selection of records or events discussed in the text for the
period 300bc to ad800. A. Total Solar Irradiance (dTSI) reconstruction from the
GRIP 10Be record from Steinhilber et al. (2009). B. 14C production from Marmod09
(Reimer et al. 2009) inverted. C. Central Europe Temperature Anomalies from Bnt-
gen et al. (2011) 10pt FFT smoothdotted horizontal grey line is average 499bc to
ad2003. D. Central Europe Precipitation Reconstruction from Bntgen et al. (2011)
10pt FFT smoothdashed grey line is average 398bc to ad2008. E. 13C record from
Sofular Cave, NW Turkey from Fleitmann et al. (2009)wetter downwards, drier
upwards. F. Temperature Deviations from North Spain Speleothem from Martn-
Chivelet et al. (2011) 10 year average. IRD Event 1 from Bond et al. (1997; 2001). Alpine
Glacier Recession, Glacier Advance (GA) from Holzhauser et al. (2005). Northwest
Mediterranean storms from Sabatier et al. (2012). Zoar Lake from Martn-Puertas
et al. (2009)note record ends ca. ad350 and that H = Humid; Iberia charcoal 13C
from Ferrio et al. (2006). Rivers and fluvial activity for southern France from Arnaud-
Fassetta et al. (2010). Southern Levant synthesis summary from Rambeau and Black
(2011). Van from Wick et al. (2003); Bereket Basin from Kaniewski et al. (2007) and
Nar Crater Lake from Jones et al. (2006) and Woodbridge and Roberts (2011). Change
to wetter phase in 1st century bc: Wick et al. (2003).
resolution Sea Surface Temperature proxies (e.g. Sangiorni et al. 2003;

Rohling et al. 2002; Cacho et al. 2001; Emeis et al. 2000), we have much
less information with regard to temperatures across most of the greater
Mediterranean region. The scale of local to regional variability is clear from
current often contrasting records. Where possible, creating data networks
(versus isolated individual records) is key to spatial control. Tighter chrono-
logical control is also necessary in a number of cases. Many data are pre-
sented as rather convincing graphs and curves, but in a number of instances
the underlying time placementsusually from radiocarbon dataremain
quite flexible. This situation makes it difficult to compare such lower-reso-
lution records with each other, or to align them with such high-resolution
series which are available. It encourages a potentially false tuning or align-
ing of apparent features, and the general suck-in and smear problem (Casel-
dine and Turney 2010, 90). This need not be the case going forward. With
substantial series of quality accelerator mass spectrometry (AMS) radiocar-
bon dates allied with Bayesian analytical modelling or other approaches,
it is possible to achieve much greater chronological control, and to better
accommodate patterns and variations observed in age-depth time-series
(e.g. Bronk Ramsey 2008; Blaauw 2010), with Blockley et al. (2008) and
Blaauw et al. (2011) offering two examples. Use of such techniques can
improve and allow interrogation even of existing time-series (e.g. Manning
2010, 2732 and Figures 812).
We should also focus on records where there is the potential to make rig-
orous linkages with human responses and history (Diaz and Stahle 2007).
But, while a rigorous timescale is key, it is also important not to let precision
become overly dominant (Caseldine and Turney 2010, 90). Some substantial
climate forcings work themselves out over larger regions, like Europe and
the Mediterranean, during periods of years to decades to even a century
as Plunkett and Swindles (2008) argue with respect to the 28002700bp
eventwhereas other forcings are much more immediate. We need robust
knowledge of what happens when to enable us to see packages of events and
processes independent of a subjective suck-in and smear tendency. While
Diaz et al. (2011, 1495) encourage a move away from descriptive climatology
towards physical climatology, we remain at present a little short of the nec-
essary range of adequate and robust data for the Roman world to get beyond
description.
What may we conclude at present? The solar proxy records give us a
broad outline for climate, within which other forcings (e.g. internal variation
including ocean systems and atmospheric processes, volcanism, land use
changes, etc.) then operate, on various scales. Several records exhibit rea-
sonable linkages to the solar activity record (e.g. Spannagel Cave: Mangini
et al. 2005, 745746, Figure 7, 749).
A range of records indicate that a stable and reasonably positive (warm,
and in a number of areas or cases also mainly moist) climate regime was
in place for the period from about the 2nd century bc through the 2nd
century ad. This unusual situation, reducing some of the typical variabil-
ity, uncertainty and risks of the Mediterranean climate regime for farming
(Garnsey 1988; Gallant 1991), would have been conducive to the growth of
the Roman world. It was also an especially favourable time (warm, moist)
for both agricultural and demographic expansion in central and northern
Europe.
The overall climate context generally deteriorates in the centuries that
follownotwithstanding reverses/recoveries in the mid-4th and 5th cen-
turieseventually to a low (cool and arid) point in the later 7th century ad.
In particular, the stability of the previous several centuries ended; agricul-
tural uncertainty and bad years would have increased. The 2nd to 5th or 6th
centuries ad seem to be relatively arid in several areas of the eastern Roman
empire, and the indications of less favourable climate conditions further
east into central Asia may have been one of the forcings behind the move-
ments of populations that led to invasions/migrations into the late Roman
world. Several records from the east indicate a change to a moister regime
for a period in the 4th to 6th or even 7th centuries ad61which may have
61 The switch to wetter conditions in the mid 5th or mid 6th centuries ad in the Sofu-
lar Cave (see Figures 17, 21), or ca. ad 450650 in the Bereket Basin study (Kaniewski et al.
2007), is also evident in the Kocain Cave speleothem record (see Luterbacher et al. 2011,
Figure 1Bsee with footnote 43 above). However, the relationships with and between the
speleothem records from the Sofular (Figure 17 above) and Kocain Caves in NW and S Turkey,
respectively, vary (noting footnote 43 above). After ca. 1150ad, when both records are rela-
tively stable, Sofular trends wetter to ca. ad 300 whereas Kocain trends drier. Sofular then
shows a drier period in the 4th to mid 5th century ad showing no relationship with the
precipitation jump in the later 4th through mid-5th centuries ad in central Europe, nor
the matching the first half of a wetter period in the southern Levant. Kocain does how-
ever show a precipitation spike in the later 4th century ad, but then has a short drier phase
around the start of the 5th century ad. Both the Sofular and Kocain records show trends
to wetter conditions from the mid-5th through 6th centuries ad as noted at the start of
this footnote (the opposite of the central Europe record). Both then have changes to much
drier conditions: Sofular especially from the mid-7th to mid-8th centuries ad (and gener-
ally to the mid-9th century ad), and Kocain from the late 6th through 7th centuries ad (so
a little earlier). Kocain thus seems to pick up the change to arid conditions in the south-
ern Levant around ca. ad 550750, and Sofular only rather later. Further investigation and
data will be required to judge how much these two speleothems reflect local conditions
or their wider regions. The Black Sea (versus Mediterranean) driven climate regime pri-
marily recorded in the Sofular Cave probably accounts for some divergence at times. The
helped the revival and consolidation of the Eastern Roman Empirebefore

then a long arid and thus generally negative period. In contrast, central and
northern Europe saw less favourable conditions arrive earlier than the east.
Precipitation and temperatures both trend downwards from the mid-5th
through the 6th centuries ad, which would have been less than optimal. This
dichotomy of less favourable west in the mid-5th through 6th centuries, and
more favourable east at around the same time, offers some context for the
divergent trajectories between the eastern and western empires and their
territories in the 5th through 7th centuries ad. The reverse then follows in
the later 7th through 8th centuries ad, as the west recovers and the east
declines.
With increasing data and an interdisciplinary perspective it can be
argued that climate is becoming an important element in the analysis of the
history of the Roman world (McCormick, Bntgen et al. 2012). But we must
nonetheless end by noting the weakness of the present story. We lack good
data from many areas of the Roman world, and we lack good chronological
control in many cases. A wider network of data is needed, and precise and
accurate chronology. We have only an inkling so far of the regional mosaic
underneath macro-level observations and generalizations. What role cli-
mate change played in the history of the Roman Empire will continue to
be debated, but hopefully with more and better data.
most obvious divergence between the two records relevant to the Roman period is in the
period before the common shift to wetter conditions noted at the start of this footnote: thus
the Sofular record (for the 4th to mid-5th centuries ad) partly corresponds to other studies
indicating arid conditions in areas of Turkey in the 4th/5th to 6th centuries ad (Jones et al.
2006; Woodbridge and Roberts 2011; Kaniewski et al. 2007), whereas the Kocain record does
not (except perhaps briefly in the early 5th century).
PART THREE
WOODLANDS
DEFINING AND DETECTING MEDITERRANEAN
DEFORESTATION, 800BCE TO 700 CE
W.V. Harris
The Problem
Did the ancient Mediterranean world experience real deforestation?* Opin-

ion is divided.1 In this paper I shall first attempt to clarify the problem
(definition is essential), describing the destructive forces that were at work
in the period from the rise of the Greek polis to the first phase of the Mus-
lim conquests. I shall then consider the evidence for possible shortages of
fuel-wood and construction wood, and make some comments on the mar-
kets in each of these classes of commodity. Next I shall briefly review the
most pertinent evidence provided by pollen analyses and by alluvial sedi-
mentation. I shall evaluate a recent discussion based on demography. I shall
attempt to demonstrate that the reason why the effects of heavy demand
for wood were not more severe was probably active woodland management
combined with an effective system of distribution. Finally I shall consider
the likely effects of known climate changes.
The substantive problem of deforestation arises here from the fact that
the Greeks, Romans and other peoples of the Mediterranean certainly used
and destroyed great quantities of trees. In the period 800 bce to perhaps
165ce, they cleared huge amounts of arable land. They were always very
heavy consumers of timber: under the Roman Empire especially, the de-
mand was heavy both for fuel wood (including fuel for the production
of metals and glass), and for timber to be used for innumerable kinds of
building and manufacturing. From ships to writing tablets, from spears to
ploughs, almost everything was made of wood. The vast majority of fuel
was wood,2 often in the form of charcoal, and cremation was widespread
until at least the second century ce. Those who do not believe that all this
* I thank Karl Butzer, Paolo Malanima and Robyn Veal for comments on an earlier draft
of this article, and Milena Vasiljevic for useful information.

1 See Harris 2011a, 108.
2 But see Veal in this volume, p. 37.
174 w.v. harris
wood use had severe effects on Mediterranean woodlands have to empha-

size that woodlands, in temperate climates at least, can regenerate them-
selves.3
Both periodization and definition are crucial to what follows. Periodiza-
tion first of all: people began clearing tree-covered land during the Neolithic,
and the larger kind of settlement could have serious consequences. Investi-
gators of a key site in Jordan, for instance, have written of dramatic local
deforestation before 4500bce.4 By the third millennium bce pharaohs were
importing timber to Egypt by sea.5 Bronze Age changes in the environment
have been widely recognized across the whole region. By 800 bce, at all
events, Mediterranean lands, especially in the east, are likely to have been
much less tree-covered than they were at the beginning of the Holocene,
though there is no way of quantifying this change. Population and economic
production then intensified from the eighth century bce till the second or
third century ceand what that did to Mediterranean woodlands is the pri-
mary subject of this paper.
What happened demographically and to production later on, say from
200 to 700ce, is hotly disputed; I shall hypothesize here that by 450 both
population and production had greatly decreased in Mediterranean Europe
but less so in the eastern and North-African parts of the Mediterranean zone.
And there is no natural cut-off point in the seventh century either. Thus
the period discussed in this paper, while it consists of a time of intensified
timber use, was neither preceded nor succeeded by periods of zero demand;
far from it. We must also remember that demand for wood could rise and
fall dramatically in particular regions independently of the trends I have
been describing: Attica, for example, and some other parts of Greece too,
were substantially denuded of tall trees in the era of Athenian naval power
in the fifth and fourth centuries bce, but Attica seems to have recovered to
a notable extent when Athens lost its independence and its fleet.6
3 See Rackham 1996, etc. Much depends, naturally, on the local climate and geology, on
how the land is treated (whether for example it is used for pasturage), and on the species in
question (cedar forest, for instance, takes centuries to regenerate: cf. Rauh et al. 2009, 267).
Conifers in general require reseeding/replanting from scratch.
4 Ain Ghazal, just north of Amman: Fall et al. 2004, 143; the climate was too arid to allow
woodlands to renew themselves effectively.

5 Gardiner 1961, 42, Polzer 2011, 351.
6 Harris 2011a, esp. 123; Harris 2011b.
defining and detecting mediterranean deforestation 175
Definition
But what is deforestation?7 Not simply, in any case, anthropogenic distur-

bance of the local forest ecosystem,8 as when, because of human activity,
one species supplants another.9
Part of the problem of definition is an excessively simple dichotomy
between undisturbed woodland and cleared land. Everyone, or almost
everyone, recognizes that landscape types are numerous in every climate
zone, with differing quantities of open areas. We need at the very least a tri-
partite categorization of ancient Mediterranean landscapes, which should
include a lightly wooded category, such as would sustain the everyday needs
of an agrarian population. But whenever woodland was cleared away and
stayed cleared away, that must count as deforestation, whether the result
was farmland, scrub, or eroded hills, or any combination of these conditions.
It will only fail to count as deforestation if, because of natural causes or man-
aged re-growth, the effect was short-lived.
Normal usage, however, seems to reserve the term deforestation for cases
that produce easily recognizable ecological crisis. And much of the argu-
ment about the Mediterranean environment is about who or what caused
the ruined landscapes that characteristically occur when karstic limestone
hills, stripped of their major vegetation, become seriously eroded. (Not that
these are the only badlands). But some regrettable vagueness remains, and
all that seems possible at the moment, in the context of ancient history, is to
distinguish between deforestation that is more radical and that which is less
so. It may be fair enough to describe the clearing of a certain area for farm-
land as an episode of deforestation,10 but episodes of this kind will only take
us so far.
7 I have to leave aside here the differences that may exist between this term and similar
concepts in other languages such as dboisement, Entwaldung, Abholzung, and disbosca-

mento.
8 Hughes 2011, 49, appears to use this expression as an equivalent of deforestation. He
claims (55) to have evidence of major deforestation in many parts of the Mediterranean
Basin in classical Greek and Roman times, and that this deforestation caused environmental
damage contributing to the disruption of ancient economies. For the latter kind of claim
see above, pp. 23. There is indeed evidence of some deforestation in the Graeco-Roman
Mediterranean, though as it happens the two studies from the 1980s that Hughes principally
relies on do not provide evidence of this even for the localities in question (Planchais
1982; Lamb et al. 1989; see below, p. 184), but the reality is more complicated and less
disastrous than he supposes. There have now been scores of palynological studies within the
Mediterranean region.
9 With the exception, I suppose, that land planted with olive trees or fruit trees can be
considered deforested.
10 Hughes 2011, 50.
176 w.v. harris
Destructive Forces
What special factors tended to destroy woodland in the ancient world?

The most obvious factor is population growth. Let us suppose that the
population of the Mediterranean world rose from 20 million to 50 million
by the second century ad,11 and that each person required on average the
product of from 1 to 2 hectares of arable land;12 that meant at a minimum an
additional 300,000km2 under cultivation. We might think that the amount
actually cleared was larger by a factor of at least two and probably more.
The land area of the Mediterranean Roman Empire, mountains, marshes,
deserts and other uncultivable land included, amounted to approximately
2,700,000km2. So it begins to be quite evident that land-clearance made a
major difference.
But it is a mistake to suppose that all pre-modern cultures have more or
less the same per capita demand for wood.13 No one, admittedly, has so far
succeeded in quantifying the Graeco-Roman demand for wood, or even for
wood-fuel, in any impressively precise way. Consider some of the difficulties.
It was metallurgy that required the truly lavish use of wood fuel in the
Roman world, not the manufacturing of bricks or glass or the heating of
baths, though all these processes required very large amounts. But we have
only a quite vague idea of the overall scale of metal production. It is often
said, for example, that the high Roman Empire is likely to have produced
80- to 85,000 tons of iron a year.14 80,000 tons would have required the fuel
produced by approximately 26,000km2 of coppiced land.15 8085,000 tons
of production is unlikely to be a greatly exaggerated figure (the evidence of
the Greenland ice-cores suggests that the production of metals in Europe
did not reach Roman levels again until the Industrial Revolution),16 and it
may well be too low, but in any case it is quite speculative.
I see no way of estimating in any useful way the amount of wood fuel that
was needed across the whole ancient Mediterranean for metallurgy, brick-
and glass-production and the heating of baths, but it was obviously enor-
mous and continuous. It could of course be argued that the ancients would
11 For the first figure see Morris et al. 2007, 9; I have not forgotten the sceptical comments
of Cawkwell 1992 about population growth in archaic Greece.

12 For 1 hectare per head as the bare minimum needed for food and fodder, see Malanima
this volume, p. 18.

13 Thus I disagree with Kaplan et al. 2009; see further below, p. 187.
14 The figure derives from a guess by Healy 1978, 196.
15 Harris 2011a, 119.
16 See the bibliography in Sallares 2007, 26.
not have been able to continue certain cultural practices, such as heating
water for baths and cremating the dead, if prices had risen precipitously
(and eventually cremation became less popularsee below).
Timber for construction sometimes seems to have severe effects, as we
shall see below, especially during building-booms and when large navies had
to be built. All the more so, because particular species were always preferred
for shipbuilding17 and for any but the simplest on-land building. On the other
hand, the volume of timber needed for fuel must vastly have exceeded the
volume needed for construction.18
Landowners and farmers must often have faced the dilemma of choos-
ing between trees and animals (a slow return versus a faster return). All
ancient Mediterranean people who knew anything about the countryside
the great majority in other wordswere doubtless aware of the fact that
trees and goats do not get along with each other (the classic text is a vivid
fragment of Eupolis comedy The Goats).19 There are many other traces
of this conflict in Greek and Latin literature.20 The trees must often have
lost.
Wood Shortages
As far as fuel wood is concerned, most Mediterranean micro-regions prob-

ably remained self-sufficient at most periods. There were three exceptions
to this pattern, possibly reduceable to two: (1) some areas of intense and
prolonged metal-working, (2) the great metropolitan centres, and (3) Egypt,
where however the root of the problem may have been Alexandria.
Athens, which had long before run short of ship-building timber, seems
still to have produced much of not all of its fuel wood and charcoal in the
late-fifth century and in the fourth,21 but it is known by a chance find to
have imported fuel wood from Torone (on the north coast of the Aegean)
in the third quarter of the fourth century bc,22 just in the period in which
its Laurion silver mines became more active again. Together with other
evidence, all this shows that Attica was probably quite severely deforested in
the mid-fourth century, but several centuries later it seems to have recovered
17 Meiggs 1982, 118, Giachi et al. 2003, Guibal and Pomey 2009.
18 See Veal, this volume, p. 40.
19 Fragment 13 Kassel-Austin (Poetae Comici Graeci V, pp. 308309).
20 See for instance Varro, RR 1.2.1718, with J. Heurgons commentary (1978).
21 See Olson 1991, 414419.
22 See the letter reported in SEG 43 (1993), no. 488, with Harris 2011a, 123 n. 89.
178 w.v. harris
to a considerable extent.23 Simply to cite one other case, it is reasonably

clear that Elba eventually ran short of fuel for processing its iron ore, so
that the processing had to be carried out at least in part on the main-
land.24 But the normal pattern of Greek and Roman production of metal
goods was that ingots of the raw material were transported to centres of
demand.25 That meant that every prosperous community, in every part of the
region, required fuel for the actual manufacturing processes, Tebtunis with
its much-differentiated metal-workers as well as the towns of high-imperial
Italy.
Rome and Alexandria had to draw on areas beyond their immediate hin-
terlands. The construction of the porticus inter lignarios outside the Porta
Trigemina at Rome in 192bce26 occurred at a time when the city of Rome
was growing very rapidly and its location suggests that the lignum in ques-
tion was coming up the Tiber, not down it as we might have expected;27
it would be reasonable to conclude that the lower Tiber valley was no
longer rich in timber (the era of carefully-managed Italian woodlands had
perhaps not yet begun). We can only speculate about where the famous
navicularii lignarii of Ostia found wood fuel to import near the end of
the second century ce (Sardinia perhaps).28 In the fourth century, under
Valens and Valentinian in the 360s, Rome was importing fuel from North
Africa; it may have been the huge baths of Diocletian and Constantine
that necessitated this.29 We shall see later that evidence from Antioch sug-
gests that by the second century ce it was running out of local fuel, and
the same may well have applied to other second-tier cities such as Perga-
mum.
As for Egypt, there seems by Severan times to have been a chronic short-
age,30 to such a degree that fuel wood was sometimes even imported from
23 See Pausanias 1.32.1.

24 Contrast Diodorus Siculus 5.13 with Varro (ap. Servius, Verg.Aen. 10.174) and Strabo 5.223
(cf. Ps.-Aristotle, On Marvellous Things Heard 93). Corretti and Firmati 2011, esp. 229, show
that iron production on the island itself had largely ceased by about 50bce.
25 Harris 2000, 723.
26 Livy 35.41.10.
27 This view is fortified by Tuccis demonstration (2004, 199) that the Pons Sublicius (and
hence in all likelihood the Porta Trigemina) was lower down the river than recent doctrine
has maintained.
28 The inscription: CIL 14.4549. The date: Meiggs 1982, 339.
29 The main texts are C.Th. 13.5.10, 14.5.1. See further Harris forthcoming.
30 See Ulpian in Digest 32.55.5 in Aegypto, ubi harundine pro ligno utuntur , where they
use reeds in place of wood.

Italy;31 the actual deforestation, however, may pre-date the period under
investigation here.
With regard to construction timber (Lat. materia), the story is quite
different. The earliest signs of local shortage concern naval power. By the
late fifth century, and probably earlier, Athenian naval hegemony depended
on importing ship-building timber from Macedon and Thrace; it was out
of the question to rely on Attica itself,32 all the more so because particular
species were required. Here is the well-informed Theophrastus:33
fir, mountain pine and cedar are the standard ship-timbers. Triremes and long
ships [i.e. warships] are made of fir because it is light, while merchant ships
are made of pine because it does not rot. Some people, however, make their
triremes of pine also, because they are short of fir . These woods are used
for the main timbers, but for the triremes keel oak is used . They make the
cutwater and cat-heads, which require special strength, of ash, mulberry or
elm.
Theophrastus shows in fact that there was a serious lack of ship-building
timber in Greece by the late fourth century bce. There is only a small area
[he means in the eastern and central Mediterranean] which produces wood
suitable for ship-building: in Europe, Macedon and certain parts of Thrace
and Italy, in Asia, Cilicia, Sinope, Amisus, and Mounts Olympus and Ida
(but they do not have much); Syria has cedars, which they use for warships.34
The other large naval powers of the high-classical Greek worldSyracuse,
Corcyra, and above all Corinthare likely to have encountered problems
similar to those of Athens.
31 P. Tebt. II.686 (second or third century). It may have been transported by grain ships
returning from Ostia to Alexandria (via the Calabrian coast). Whether we can associate fuel
shortage with a decline in brick production in Italy after Severus Alexander (there was a
recovery under the tetrarchy) or with the large-scale change from cremation to inhumation
is regrettably obscure (Meiggs asserts the one possibility while denying the other [1982, 504
n. 119 and ib. 257]; I incline to the opposite view). However the whole notion that cremation
ceased, or largely ceased, in the Roman Empire as a whole is in need of revision: in some
provinces it certainly did not see for example Pop-Lazic 2002.
32 For the evidence for Attic deforestation to be found in Plato, Critias 111c, see Harris 2011b.
33 Hist. Plant. 5.7.13.
34 Hist. Plant. 4.5.5. This might be taken to imply that Xenophons report (Anabasis 6.4.4)
of ample ship-building timber at Calpe, on the Bithynian coast far to the west of Sinope, was
now out-of-date. Italy for Theophrastus still meant Calabria; he underestimated the resources
of the rest of the peninsula, which he had not apparently visited. He wrote this very early
in a period of about a century during which, because of the number and increasing size
of warships, the long-timber supplies of the eastern Mediterranean were put under greater
strain than ever before or later; cf. Meiggs 1982, 139.
180 w.v. harris
Theophrastus tells a somewhat similar story about long timber for the
builders needs, but there is a significant difference, because here he is
concerned with quality not availability: the main cities of European Greece
had to import the best building timber from Macedon, the Black Sea and
other places, but rustic Euboea produced some building timber even though
it was of poor quality.35 Such evidence as we have for the sources of timber
listed in various fourth-century temple-accounts shows that Arcadia and
even such a centrally located place as Sicyon could still provide some long
timber.36 We might conclude that most but not all of mainland Greece,
Sicily, the Aegean islands and the west coast of Asia Minor had lost most of
their tall trees by about 310bc, leaving them with simply enough coppiced
trees and shrubs to provide fuel. Outside Attica, however, we have no strong
reason to suspect ruined landscapes. Mount Lebanon meanwhile was rich
in timber, cedar and cypress especially, when Antigonus set out to build a
fleet on that coast in 315.37
Ptolemaic Egypt was far from self-sufficient in this respect. Since this had
probably been true of Egypt since the second millennium bc if not earlier,
there is no need to set out all the evidence. Naval power, not to mention
the building of Alexandria itself, depended on timber from overseas pos-
sessions, in particular Cyprus.38 In Egypts Greek period, the governments
concern is well attested from the middle of the third century, and one has
the impression that in Greek and Roman times every tree in Egypt was
under surveillance.39 Regulations promulgated by the king in 118 bce show
not only that it was illegal to fell a tree on private land without permission
but that the occupiers of some legal categories of land had some obligation
(not explained in detail) to plant trees.40 The most interesting fact, however,
may be that a region possessing only minimal quantities of harvestable long
timber could in fact manage reasonably well within the broader Hellenistic
35 Hist. Plant. 5.2.1.

36 Meiggs 1982, 423457, reviewed this evidence in detail, but I do not share his confidence
that a timber-sellers geographical origin reveals where his timber came from (there are
too many Corinthians). The most interesting evidence from our point of view comes from
Delphi, showing among other things that the rebuilders of the temple of Apollo could obtain
seventeen cypresses from Sicyon in 335, but at exceptionally high prices (Meiggs 431).
37 Diodorus Siculus 19.58.
38 Meiggs 1982, 133135.
39 On all this see Cadell 1976, 346, and Kramer 1995, 218222. But note that the government
was interested in trees partly because of their importance for the security of dykes: Drew-Bear
1995, 34, and P. Tebt. III.703 (second half of the third century bce).
40 Select Papyri II.210 = C.Ord.Ptol. 53, lines 200206. (This depends on whether the plant-
ings referred to have to be trees).

and Roman economies, helped by a wheat surplus, the de facto papyrus

monopoly and other advantages. But ambitious ship-building required the
control of territory elsewhere, and that was why M. Antonius gave Cleopatra
the territory around Hamaxia in western Cilicia.41
Long before this time the Romans had come to dominate the Mediter-
ranean Sea, having paid careful attention to the supplies that they needed
for naval warfare.42 Their naval history started from small beginnings in the
late fourth century and reaching a crescendo at the Battle of Myonnesos,
off the Ionian coast, in 190. Fortunately for Rome and for the Mediterranean
woodlands, the interests of imperial power seldom thereafter required the
deployment of a large fighting navy.
The evidence about Italy is a little mixed, but as it stands now it suggests
an intensifying shortage of tall trees in early imperial times (further research
on Campanian and Ostian architectural timber might possibly validate or
invalidate this conclusion). When Dionysius of Halicarnassus, a worshipper
of Rome, lauds the woodlands of Italy, claiming in particular that they pro-
vide ample timber for ship-building, he is demonstrably exaggerating, and
committing something of an anachronism.43 Other evidence suggests that
some areas of Italy became more and more dependant on long-distance
imports.44 A text of Strabo about Pisa points in this direction: most of the
timber in the Monti Pisani is now [in Augustan times] being used up for
construction at Rome and elsewhere.45 This is what was likely to happen
when there was long timber fairly close to a good port. But Strabo also says
that most of Romes building timber comes down the Tiber from Tyrrhe-
nia.46 Somewhat later the situation seems to have changed: some of the
fir and spruce (picea sp.) from the Campanian cities came from far away,
according to one theory from the Austrian Alps.47
41 Since it was suited to the building of fleets, Strabo 14.669. He is somewhat vague about
the extent of the region in question. This was also how she obtained Rough Cilicia (ibid. 670),
presumably.
42 It is symptomatic of this concern that at some early date the state took title to all coastal
woodlands (Cicero, De rep. 2.58he attributes the measure to King Ancus Marcius).
43 Dionysius of Halicarnassus, Roman Antiquities 1.37.4 (taken literally by Nenninger 2001,
200). It was quite untrue, for example, that Italy had mines of all kinds, 1.37.5; cf. Brunt 1971,
128129. And he is plainly inaccurate when he says (20.15) that the Sila suffices for the needs
of Italy.
44 I leave aside the matter of super-luxurious tables made of citrus-wood (Callitris quadri-
valvis) imported from Mauretania, a fashion that led to the deforestation of the Mons Anco-
rarius (Pliny, NH 13.95).
45 5.223, mistranslated by among others Grove and Rackham 2001, 173 n.
46 5.222.
47 Kuniholm 2002, 236237 (dendrochronological evidence). It is a pity that the re-
searchers in question did not check samples from Liguria, since that would be a more
182 w.v. harris
The elder Pliny thinks of hills freshly stripped of trees as a common

phenomenon: springs often arise when woods have been cut down [he
refers to an incident from Greek history] harmful torrents often run
together when the woods which used to hold and absorb the rains have
been stripped from the hills.48 It is quite wrong to cite him as a witness
to the supposed fact that there had been widespread deforestation,49 but
the frequent allusions in Natural History Book 16 to sources of timber such
as Raetia, Histria and Corsica that are just outside the Italian peninsula,50
suggest that certain species at least were not readily to be had nearer at
hand, even though he also mentions the Appennines. Elsewhere he alludes
to logging in the upper reaches of the Tiber, and he mentions the silva Sila.51
The source areas mentioned by Pliny are thus, in the main, well removed
from the densest concentrations of population.52
There may be other signs too of an eventual shortage in the area of the
metropolis: Ulrich has pointed out a difference in construction practices
between the Vesuvian cities in the first century and Ostia in the secondthe
latter used less timber, perhaps, as he suggests, because the supply of long
timber was now under stress.53
Emperors always owned a good deal of woodland,54 but the only ambi-
tious attempt by an emperor to look after the supply of long timber seems
to have been an elaborate initiative undertaken by Hadrian in the province
of Syria. In the northern half of Mount Lebanon, he had markers put up
over vast areas of forest. There were at least 800 such markers (they were
plausible source. Robyn Veal (pers.comm.) shares my doubts about the Alpine source, noting
that Picea grows in Liguria and Tuscany (Pignatti 1982, I, 7475).
48 NH 31.53.
49 Hughes 1983, 437, on the basis of NH 13.65 (together with Livy 9.36), completely misrep-
resented (the Pliny passages refers only to a supposed silvestris regio near Memphis; actually
the whole passage about trees in Egypt, sects. 6065, is an intriguing one).
50 Raetia (sects. 66, 190), Histria (66), Corsica (71, 197). The other places he mentions are
Macedonia, the Pyrenees, some specific zones of Asia Minor and Gaul, the Tyrrhenian coast
of Italy (meaning Liguria?), the Alps and Appennines, Crete, Africa, Syria and the land of the
Vaccaei in Spain.
51 NH 3.53, 74.
52 It is intriguing that a resident of Italy such as Hermas (first- or second-century) seems
to have been familiar with tree-less mountains (see at length Parable 9.1 and 9.1929 in the
Shepherd); he sets the scene in Arcadia simply because it was famously mountainous.
53 Ulrich 2007, 121. An interesting topic which we are not (or not yet) in a position to
clarify, is why certain places, Gades for example, became ship-building centres at certain
periods.
54 But I would not go as far as Thonemann 2011, 280, according to whom throughout
antiquity large and potentially profitable stretches of woodland were normally regarded
as state property.
often numbered),55 reserving four types of trees as imperial property.56 This

definitio silvarum obviously shows that the emperor or some powerful sub-
ordinate regarded these forests as a valuable resource (and one recalls that
Hadrian travelled in this region in 129130). It seems unlikely that he would
have intervened if he had not thought that the Mount Lebanon timber was
at risk; the exact nature of his concern is unknown, but military prepared-
ness, Mediterranean shipping and imperial building plans may all have
come into it.57
Under the high Roman Empire, there is good deal of evidence about
shortages of long timber in certain areas, but nothing on the other hand that
would justify diagnosing widespread deforestation throughout the Mediter-
ranean.
The Palynological Evidence
Pollen deposits may one day settle the issue at hand in a definitive fashion,
and over some thirty plus years studies of this material have proliferated
in many Mediterranean regions58though not in some of the areas where
they would be most interesting, such as Tunisia, Appennine Italy, Sicily, and
Dalmatia. This evidence comes with various caveats, however. When wood-
land begins to be used for grazing rather than being displaced by cultivation,
it may be hard to detect in the pollen record.59 Furthermore, many pollen
studies concern more or less remote areas that are of secondary importance
for the overall problem of deforestation. And it is only recently that it has
become feasible to take proper account of three important variables, the
pollen productivity of different species, the fall speed of different kinds of
pollen, and the prevailing winds at the various sites in question.60 Research
that takes inadequate account of these variables may be next to useless.
Then there is the problem of low-resolution chronology.
However the following results, set out here very briefly, have seemed to
emerge.61
55 See IGLS VIII, 3, edited by J.-F. Breton; at least 187 of these inscriptions have been
recorded. There are now dozens more: A 2006, 1572 f.

56 Probably cedar, fir, and pine, and either cypress or juniper: Mikesell 1969, 20.
57 For other comments on this evidence see Harris 2011a, 130131.
58 It is regrettable that there is no thorough up-to-date bibliography. Di Rita and Magri
2009 is useful from this point of view.

59 Atherden 2000, 6466.
60 See Poska and Pidek 2010.
61 For a more detailed but already somewhat outdated survey see Harris 2011a, 133136.
184 w.v. harris
(1) In a number of places, both the Bronze Age and the later spread of Greek
agriculture sharply reduced arboreal pollen; there is good evidence to this
effect from, for example, bronze-age Miletus and the early Hellenistic Golan
Heights.62 In Acarnania about 800600bce there was a marked change in
vegetation, obviously to be connected with increased human presence.63
Whether any of this amounted to deforestation seems to me to be to some
extent a matter of definition. Unfortunately the areas most likely to have suf-
fered deforestation in classical Greek and Hellenistic times, such as Attica
and the hinterland of Alexandria, have no palynological evidence to offer, as
far as is currently known.
(2) The Roman Empire witnessed the serious depletion of some woodland
resources, and the effective management of others. Here in fact is our cen-
tral problemwas the former pattern widespread? We have already seen
some textual evidence that may make us suspect that it was so. Recent
pollen research has detected this effect at Miletus,64 in coastal Puglia,65
and in north-western Iberia.66 Older reports make similar claims about, for
instance, sites on the coasts of Catalonia67 and Provence.68 Pollen research
62 See Knipping et al. 2008 (Bafa Gl, Miletus) and Neumann et al. 2007 (Birkat Ram,
northern Golan Heights). This is how I would interpret the marked retreat of forest vegeta-
tion (Gerasimidis 2000, 35) at Lailias in the mountains of Macedonia (1420m. a.s.l.) in the
second century bce; if the Romans had anything to do with it, it may have been because they
brought this district into the timber market (the Strymon valley was not far away).
63 Jahns 2005 (Lake Voulkaria; the main feature was a big increase in Phillyrea, a genus of
small flowering evergreen tree).

64 Knipping et al. 2008.
65 Di Rita and Magri 2009. They say that an accelerated decline in arboreal pollen took
place at 2100 bp, but their Fig. 3 seems to date the event about 23002200bp. They do not,
however, apply the term deforestation to anything that occurred in Roman or late-Roman
times.
66 Mighall et al. 2006.
67 Riera-Mora and Esteban-Amat 1994.
68 Planchais 1982. Hughes 2011 relies heavily, though indirectly, on this study as evidence
for Graeco-Roman deforestation. Planchais showed that pollen recovered from a site near
the tang de Mauguio (43 350 N) in Provence demonstrated a sharp contrast between a
millennium-long period (roughly 22701300bp) and what came before: the former showed
lots of beech and oak, whereas the only trees well represented in the latter period were walnut
and olive, and non-arboreal pollen (Typhaceae and Cyperaceae as well as Gramineae) was
abundant. In other words, the local population and/or the Romans massively changed the
pattern of local vegetation (be it noted that there was a huge amount of Roman centuriation
nearby, though not in the immediate vicinity). Another study cited by Hughes, Lamb et al.
1989, discussed a site in the Middle Atlas in northern Morocco; these authors unfortunately
wrote a misleading abstract in which they asserted that anthropogenic forest degradation
dates from about 2250bp, a date, be it noted, when the local economy had nothing to do with
also seems to show, however, that at Sagalassos in Anatolia, at a site deep in

the Libyan desert, and at another high up in the Alps, to name three rel-
atively clear examples, Roman practices did not have a strongly negative
effect, if any at all, on the local tree population;69 but the last two of these
sites can easily be regarded as peripheral.
(3) Late antiquity presents a similarly complex array of evidence. In some

places woodlands grew back: at the site in the Golan Heights already referred
to, olive cultivation collapsed in the first half of the seventh century and
Quercus calliprinos expanded.70 But the same period seems to have been
quite harmful to trees in some places, for example in Spain, in spite of an
economic shrinkage that might have permitted woodland regeneration: at
a site in the north-west of the peninsula, both total arboreal pollen and
the percentage of pollen attributed to Quercus seem to be declining in the
approximate period 550750ce;71 at sites on the Catalan coast, there is con-
siderable evidence of a decrease in arboreal pollen in the period 500 to
700ce, and the reduction in woodland led to erosion (the investigators pro-
posed to associate these changes with an increase in grazing).72 At the site in
Puglia mentioned above, olive trees come to dominate the pollen evidence
more and more in late antiquity, especially after about 500 ce, while most
trees keep their declining trend;73 was this trend owed to Byzantine demand
for the regions olive-oil? It seems obvious, in short, that as far as the pollen
evidence is concerned we cannot yet generalize about the whole Mediter-
ranean world in late antiquity.
If Mediterranean woodlands suffered in the period 500700ce, climate
change may have been partly or wholly responsible (a more arid climate
may have prevailedsee below), but we do not yet know enough to assert
this with any confidence. Another possible explanation might be a decline
in the quality of land-management. The violent disruption of the barbarian
either the Greeks or the Romans; what actually happened at about that date was the decline
of Fraxinus sp. and Quercus canariensis (72); a more severe phase of forest degradation, these
authors says, begins at about 1600 years ago (ibid.), that is to say with Roman power and
influence in this area now on the decline.
69 See respectively Vermoere et al. 2003; Hunt et al. 2001; Moe et al. 2007 (the area in
question is between 1830 and 2304 m. a.s.l.). The same seems to be true at Lake Voulkaria
in Acarnania (above, n. 63).
70 Neumann et al. 2007, 340.
71 Mighall et al. 2006, esp. 209 fig. 3.
72 Riera-Mora and Esteban-Amat 1994, esp. 2021. The key sites were at Ullastret (decrease
in AP values at 1500 +/- 80 bp) and at Bess (not much before 1300 +/- 40bp).
73 Di Rita and Magri 2009, 300.
186 w.v. harris
invasion, especially in the early fifth century, may have made careful, long-
term estate management a rarer phenomenon. (There is admittedly some
dispute about the whole nature of these invasions).
(4) Being close to water-borne transport under the Roman Empire meant
economic opportunity and also ecological trouble (contrast for example two
sites in the Alps, the Lac de Praver near Grenoble,74 with excellent access
to river transport, and Val Febbraro in the province of Sondrio,75 with very
poor access). But this statement has an important corollary: much of the
woodland of the Roman Empire, even in Italy, not to mention the Danube
and other provinces, was beyond the range of affordable water-transport,
and was therefore relatively safe from deforestation. A wooded area well
away from water-borne transport, such as the Troodos Mountains in Cyprus,
was to some extent protected from exploitation.76
Sedimentation and Erosion
Erosion is a very large topic in the geological literature, but here I sim-
ply want to ask whether any of the resulting sedimentation in the Graeco-
Roman Mediterranean can be traced to deforestation within the period we
are examining. This is widely assumed to be the case, but precise chronology
has generally been lacking.
Recent literature has provided some relatively precise sites in Turkey.
Results from the hinterland of Antioch on the Orontes are especially inter-
esting. A recent investigator has shown that in the Jebel al-Aqra region
immediately to the east of Antakya, after millennia of human habitation,
severe erosion set in from about 150ce onwards. In keeping with current
trends, he attributes this partly to climate and partly to land-use practices,77
but in a moment of candour he admits that there is no correlation with any-
thing we know about climate change in the Levant.78 It is reasonably obvious
that for centuries after its foundation in 300bce Antioch managed its need
74 Nakagawa et al. 2000.

75 Moe et al. 2007.
76 Cf. Butzer and Harris 2007, 1951, with Theophrastus, Hist.Plant. 5.8.1.
77 Casana 2008. The 400-year lag between the initial settlement of upland areas and the
first evidence of soil erosion suggest[s] that it may have been the intersection of extreme
precipitation events with particular land use conditions of the Roman and late Roman
periods which worked together to drive soil erosion (429).
78 Casana 437.
for fuel-wood without severely deforesting the nearby hills (the area in ques-
tion is between 7 and 14 kms. from the centre of the city), but eventually
the situation got out of hand; given a population of perhaps 250,000 in the
second century, that is hardly surprisingand the greatest density of set-
tlement in this particular areas, together with the worst erosion, was still to
come (from about 300 to 600ce).79
An earlier study, concentrating on south-western Turkey, detected similar
effects in a more vaguely defined classical period, offering as an example
the area between Burdur and Elmal, that is in ancient terms between Lycia
and Pisidia.80 Sedimentation at Lake Bafa, near to ancient Miletus in western
Turkey and to the mouth of the River Maeander, reached a peak in the period
from the first century bce to the fourth century ce.81
A Demographic Approach
In a previous article I implied that known population growth must have

led to massive woodland clearance in the period from 800 bce onwards,
but also that this approach cannot by itself settle the major question about
ancient Mediterranean deforestation.82 Meanwhile, in a paper entitled The
Prehistoric and Preindustrial Deforestation of Europe, published after mine
went to press, Kaplan, Krumhardt and Zimmermann made a resolute and
more optimistic attempt to come at the deforestation problem from the
angle of demography.83
They estimate the population of ancient Europe and most of the Mediter-
ranean, posit the amount of arable land that would have been necessary to
support these populations (taking account of climate and of soil properties);
that, according to them, was the amount of land cleared and deforested.
Their population estimates for ancient times are mostly reasonable ones
(and we can overlook the incongruity of their claim to know what the popu-
lation of, say, Belgium was in 1000bce; they insist on using modern territorial
units).84 And it is a positive aspect of this study that its authors take account
of soil properties and climate change.
79 Casana 433.
80 Roberts 1990.
81 Knipping et al. 2008. On possibly related changes in the rural economy of Miletus in
the first century bce see Thonemann 2011, 293, and on Miletuss prosperity for much of this
period ib. 334338.
82 Harris 2011a, 116117.
83 Kaplan et al. 2009.
84 And some historical oddities are perhaps inevitable: Maghreb regions of North Africa
188 w.v. harris
The most striking results of this approach, as far as the period 800 bce to
700ce is concerned, are a major reduction in forest coverage on usable land
between 500bce and 1ce both in North Africa and in a zone designated as
central and western Europe,85 with some recovery by 500 ce, especially in
North Africa and the zone designated as eastern regions with consistently
low forest cover.86 Thus
1000bce 500bce 1ce 500ce 1000ce

Central and western Europe 77.2 70.6 50.7 52.4 40.2
North Africa 93.9 75.7 33.3 43.6 27.8
Eastern regions with consistently 48.7 40.4 32.5 41.0 36.5
low forest cover
But there are several serious weaknesses in this approach.87 One is a vastly
oversimplified idea of consumption patterns. Kaplan et al. shunt this prob-
lem aside, announcing that they assume that population change is the pri-
mary driver of change in forest area for the years 1000 bc to 1850.88 Taken
literally, this may be true, but their assumption ignores the enormously vari-
able pressures on both timber for fuel and construction-timber within the
period in questionjust consider the single item of fuel for metallurgy,
which I discussed earlier. The second serious weakness in this paper is that
its authors hugely overestimate the amount of land necessary for subsis-
tence in a pre-modern economy, by a factor of perhaps five with regard to
Mediterranean populations. Their Figure 9,89 though it offers no speculation
about the population density of the Mediterranean world in the period I am
discussing here, implies, in conjunction with the rest of the article, that the
cultivable land in that region could in antiquity have supported between 10
and 30 people per km2. The origin of this estimate is unclear, but it is gen-
erally held, on the basis of extensive research, that one to two hectares per
display a distinct pattern of low clearance during early times with a relatively rapid increase
at around 200 B.C., Kaplan et al. 2009, 3025, referring to their Fig. 8, group 4.
85 North Africa means for them Morocco, Algeria, Tunisia and Libya; central and west-
ern Europe means Czechoslovakia, France, Germany, England-Wales, Ireland, Italy, Poland,
Portugal and Spain.
86 This means Cyprus, Greece, Iraq, Palestine-Jordan, Syria-Lebanon and Turkey-in-Asia.
87 One might quarrel about many details here. For instance, it is not plausible to suppose
that the central and western European zone had notably less forest coverage in 1000ce than
in 1 ce (the culprit here is clearly an indefensible progressive view of history). Ancient Cyprus
was certainly not a region of consistently low forest cover. And so on.
88 Kaplan et al. 2009, 3019.
89 Kaplan et al. 3028.
person were sufficient for subsistence,90 i.e. population density on cultivable

land lay in the range 50 to 100 persons per km2. Be it noted that at a density
of 20 persons per km2, the population growth conservatively hypothesized
for the Graeco-Roman Mediterranean (see above) would have required the
clearance not of 350,000km2, but of perhaps 1,750,000km2, which is entirely
impossible.
These two criticisms of Kaplan et al. might very roughly cancel each other
out: they underestimate some of the destructive forces tending to produce
deforestation, as far as the Graeco-Roman period is concerned, but they
overestimate others. What gets left out, however, is what is culturally spe-
cific. This matters in two ways (and raises again the question of definition).
No ancient or mediaeval town or city could afford to be reckless about its
fuel needs, and as noted above we have every reason to believe that with a
few readily comprehensible exceptions ancient communities satisfied those
needs within their own territories, which implies that they at worst main-
tained enough scrub, boscaglia, phrygana, and coppiced or pollarded trees
to supply themselves with fuel. The need for long timber was not so man-
ageable, or rather what made it manageable was a combination of trading
networks and imperial power that left large areas denuded of tall trees over,
at least in some cases, long periods. The relative importance of ship-building
and on-land building varied greatly from period to period, and the effects
of these needs on the landscape must have varied greatly as well. In some
cases (fifth- and fourth-century Attica, for instance, and the lower Tiber val-
ley in the high Roman Empire), the effects probably amounted to what by
any standards should count as deforestation; in other cases, the effects must
have been less drastic, with some tall trees surviving but little in the way of
continuous woodland.
I consider therefore that Kaplan et al. probably overestimate to a sig-
nificant degree the relative decline of forest coverage in both central and
western Europe and the eastern regions with consistently low forest cover
in the five centuries from 500bce onwards. Their article is undeniably use-
ful, but it merits further discussion not unqualified citation.
Woodland Management
The totality of the evidence thus seems to suggest in short that there were
probably some deforested areas in classical and Hellenistic Greece and
90 Malanima 2009, 106, with references.

190 w.v. harris
the high Roman Empire. At the height of the Roman Empire, many areas
that had been more or less heavily wooded in 800 bce were farmland with
scattered trees, or scrub-land, or in some cases heavily eroded land. Some
areas that lost their trees were woodland once again by 700ce. There was,
however, no general crisis of timber-supply in antiquity, not only because
woodlands grow again in temperate climates and because demand created
trading mechanisms: landowners also put a great deal of effort into looking
after what they of course knew to be a valuable resource. Such an assertion
does not imply any idealization of the classical world, nor should it obscure
the law of unintended ecological consequences (we know, for example,
about the ecological harm done by the Roman system of drainage in one
section of the valley of the Rhne91).
Our direct evidence about tree- and woodland-care in classical antiquity
consists, admittedly, of advice from highly literate arboricultural experts
such as Columella whose works cannot have been widely circulated: yet
their advice arose ultimately from real experience and can therefore give
some hints about actual practice. In real life, no doubt, practical woodsmen
mostly passed on information and advice by word of mouth.
You can in fact see knowledge and ambition increasing in the surviv-
ing texts. Theophrastus, writing in the late fourth century bce, is already
impressive: consider for example his discussion of the merits and demer-
its of various kinds of wood for charcoal-making,92 which also seems to
reveal the existence of a sophisticated charcoal market. He apparently knew
something about coppicing.93 He gives advice about grafting.94 But while he
envisages the planting of fruit trees, he clearly sees almost all other trees as
simply natural growths; he seems to make an exception for cypresses,95 pre-
sumably cultivated for the long timbers they provided for construction and
shipbuilding.
150 years later, Cato is several degrees more interested in the cultivation
of non-fruit trees; he explains how to plant willows, elms and pine-trees
as well as cypresses (to which he too pays special attention).96 There is
already a seminarium, a nursery, for young olive-trees on his ideal estate;97
91 Such at any rate is the argument of Van der Leeuw et al. 2005, 25.
92 Hist. Plant. 5.9.
93 I take this to be the reference of the word kolobon in Hist. Plant. 5.9.2.
94 Caus. Plant. 1.6, etc. Aristotle knew of this technique.
95 Hist. Plant. 2.7.1.
96 De agr. 9, 28, 151. See further Meiggs 1982, 262263. Other references to woodland
management include chs. 6, 7, 17, 37 end, 38 end, 55.

97 De agr. 4546.
this was a regular Roman institution. He gives detailed instructions about

grafting olives, fig-, pear-, and apple-trees and vines.98 He recommends the
owner of estates that are suburbani to produce firewood for the market
(whereas the aim of planting tall trees is not the market but simply self-
sufficiency).99
By the time we get to Varro, in the 30s bce, there has been another
change: he refers to the Roman habit of planting big trees, specifically pines,
cypresses and elms; he particularly recommends the latterthere is no
better tree for planting; it is extremely profitable.100 But why? Not, as we
might have expected, because these are three of the four woods most used
in ship-building,101 but for more complex reasons: an elm often supports
and gathers a certain number of baskets of grapes, yields most agreeable
foliage for sheep and cattle, and provides stakes for fencing, and also for
hearth and furnace. At all events this passage demonstrates that Varro and
his readers, who will have included substantial Roman landowners with
estates in diverse parts of the Mediterranean, are completely familiar with
the techniques necessary for planting such trees.
Columella, like Cato, was a great enthusiast for vineyards, but he admits
that there are many landowners who preferred pasture land or silva cae-
duatimber-producing land for coppicing;102 he shows once again how
familiar Roman landowners were with techniques for planting large trees.103
He also shows that by modern standards Roman landowners coppiced their
trees on a very short cycle, a reflection of the powerful demand for wood
fuel.104
In the same period the elder Pliny maintains that cypress branches, which
go for a denarius apiece after twelve years of growth, are the most profitable
kind of planting: this is because they can be used as vine-supports.105 He
98 De agr. 4042.
99 De agr. 6 and 7.
100 RR 1.15 (maxime fructuosa, correctly translated by Heurgon 1978 as dexcellent rap-
port). Williams 2003, 97, was wholly mistaken to claim that there were no examples of efforts
to plant trees other than olive trees. On Roman perceptions of the relative desirability of
woodland and other kinds of land see Giardina 1981, 102103.
101 Silver-fir would not come into his calculations in any case, since it is assumed that the
ideal estate is not at a very great altitude.

102 De Re Rustica 3.3.1.
103 De Re Rustica 5.6 and De arboribus (though the latter concerns fruit-trees, and the
author says at the beginning [1.1] that trees that provide materia, construction timber, grow
without human aid).
104 See Meiggs 1982, 268.
105 NH 16.141. They are known as a daughters dowry, he says, twelve being his idea of the
age at which a girl would marry.

192 w.v. harris
takes it for granted that many different varieties of trees other than fruit trees
are regularly propagated by landowners:106
Trees for coppicing, in addition to those which we have mentioned [willow,
chestnut, aesculus oak and cypress] are the ash, laurel, peach, hazel, and
apple, but these shoot more slowly and when fixed in the ground tolerate
the soil, not to mention the damp, with difficulty. The elder, on the contrary,
which is very strong timber for a vine-stake, is grown from cuttings like the
poplar.
Other species of non-fruit trees deemed worthy of cultivation include plane,
elm, ash and alder.107
In short there is every reason to think that by the second century bce
and even more in later times Roman estate-owners regarded many species
of tree as useful assets, to be exploited and therefore to be propagated.
The Impact of Climate Change
In the temperate zone that makes up the majority of the area under discus-
sion, the most important climatic factor that may have affected the lives of
trees is the pattern of precipitation, and not only its volume but its seasonal
distribution.108 Can we identify any Mediterranean climate changes in the
period 800bce to 700ce that are likely to have had large effects on wood-
land, and in particular any changes in patterns of rain- and snow-fall? The
sheer diversity of Mediterranean climates109 is bound to make this difficult.
The papers by Michael McCormick and Sturt Manning in this volume
both review the existing evidence. My reading of their work and of the
work they cite is that as far as archaic Greece is concerned (the eighth
and seventh centuries bce in particular), while it is very possible that pre-
cipitation increased and aridity declined across wide areas of the eastern
Mediterranean, it is too early to assert this as a fact.110 Even if it were true,
it is quite uncertain what the net effects on woodlands would have been,
106 NH 17.151.
107 NH 17.6578.
108 Cf. Perry et al. 2008, 4159, Manning this volume p. 107.
109 Cf. Grove and Rackham 2001, 25, who do not even consider the eastern or southern
shores.
110 A recent review of the literature on Holocene climate in the eastern Mediterranean,
which considered some eighty studies, concluded that during the late Mid-Holocene (2800
1800 yrs bp) diverging records prevent the emergence of a coherent picture (Finn et al. 2011,
3167). (bp for these authors means before 1950). There is still a chronic shortage of data with
high-enough chronological resolution to be useful to a historian.
since the hypothetical change in climate may have been a prime cause of
the demographic expansion that undoubtedly took place in this period, at
least among the Greeks (shades of Malthus). Greeks of this period greatly
increased their production of both metal artefacts and ships. A modest
increase in precipitation may perhaps have caused more loss of woodland
than gain. It is a matter of scale, and we need more numbers.
We seem to be on more solid ground in the third century ce and later.
McCormicks group considers that precipitation declined in Romes eastern
provinces and in north-eastern Gaul in the third century and recovered in
the fourth.111 Another dry period sets in in north-eastern France about 450
and continues for about two centuries.112 In the eastern provinces, according
to this group, a humid climate continues until it changes dramatically
in the sixth century.113 These changes coincide to some extent with the
pollen evidence reviewed earlier (but they appear to be partly based on the
pollen evidence). The most serious difficulty is that north-eastern France
is merely a fragment of the area we are trying to generalize about, while
the eastern provinces of the Roman Empire cover a huge and ecologically
diverse area where we would need data from many different places. And
a recent literature review seems to show that the eastern-Mediterranean
climate grew wetter not drier from 550 to 750ce.114
Conclusions
Much woodland was degraded or disappeared in the Graeco-Roman Medi-

terranean. But no extreme hypothesis about deforestation seems well-
founded, and there is no reason to believe in a generalized crisis (though
Italian wood being used for fuel in Egypt comes close to suggesting that).
In truth, the uncertainties and unknowns are very extensive. Both scien-
tists and historians should work harder to achieve a clearer definition of
deforestation, or a typology of deforestations.115 And, the problem of gen-
eralization is fundamental: we need more discussion of, on the one hand,
111 Bntgen et al. 2011; McCormick, this volume, p. 70.

112 Ibid.
113 McCormick, this volume, p. 71.
114 Finn et al. 2011 (see above, n. 110), 3168 (a widespread period of wetter conditions at
14001200 yrs bp). See their Fig. 1. Their eastern Mediterranean includes Italy, and they have
no data from Syria or Egypt. More precipitation from 550 ce onwards at a site in Switzerland:
Roos-Barraclough et al. 2004.
115 It is depressing to read an account by a classicist who, while citing some scientific
studies in his bibliography, makes no systematic use of them (Thommen 2012, 3741, 8589).
194 w.v. harris
the representativeness of the scientific evidence and on the other the real
value of the textual sources.
Thus the following claims are tentative:
(1) Classical Greek deforestation, in a weak sense of the term (land clear-
ance, with some trees left behind) was widespread, but in a strong
sense of the term it was probably restricted to Attica and the immedi-
ate supply area of a few other cities. Local fuel crises probably occurred
wherever metal-smelting was intense (so for this reason too Attica was
under pressure in the fifth and fourth centuries).
(2) Hellenistic and Roman demand intensified in the third century bce.
This demand was met to some extent by a stronger trading network
and more careful woodland management, but ship-building and fuel-
needs are likely to have deforested some areas quite seriouslymost
of all perhaps in Italy, and left others with much less woodland cov-
erage.
(3) Stress on woodland resources continued to mount all the way through
the high-imperial period of prosperity, perhaps down to Severan times.
(4) Alleviation that came later was highly selective. Some areas regained
woodland, but whatever the reasons were (unfavourable climate
change, poorer management) others did not.
PART FOUR
AREA REPORTS
PROBLEMS OF RELATING ENVIRONMENTAL
HISTORY TO HUMAN SETTLEMENT IN THE
CLASSICAL AND LATE CLASSICAL PERIODS
THE EXAMPLE OF SOUTHERN JORDAN*
Paula Kouki
Introduction
One of the central questions in Mediterranean archaeology is the relation-

ship between humans and their environment. The issue of environmental
change and the factors behind it, as well as its influence on human settle-
ment and land use, is, therefore, a valid topic of research for historians and
archaeologists working in the region.
The dating of archaeological sites is usually based on pottery. For the
Classical and Late Classical periods pottery is closely datable, often within
a century, sometimes even within a few decades, which makes it possible
to reconstruct relatively detailed settlement histories. From ancient writ-
ten sources we may even obtain a picture of the political and economic
conditions of a given region. However, the dating resolution of the results
of palaeoenvironmental and climatological studies, such as sedimentary or
pollen analyses, is generally within a couple of hundred years at best, which
makes it complicated to relate environmental changes to those in human
settlement and land use. Furthermore, different records of environmental
data may produce widely differing pictures, for example, of climatic trends.
Some of these problems may be overcome by means of high-resolution envi-
ronmental data,1 but such data are not yet widely available. There are also
regions where the availability of palaeoenvironmental data is limited due to
local conditions, such as the lack of waterlogged deposits and sedimentary
basins. This necessitates the use of proxies, which creates its own problems.
* Acknowledgements. This research has been carried out under the Finnish Academys
Ancient Greek Written Sources Centre of Excellence in Research. I would like to thank the
following foundations for funding my research: the Jenny and Antti Wihuri Foundation, the
Finnish Cultural Foundation, the Emil Aaltonen Foundation and the Research Foundation
of the University of Helsinki.
1 Manning, this volume.
198 paula kouki
In this article, I address these problems and discuss them in the light
of research material from southern Jordan. After providing a background
description of the environmental conditions and archaeological research
there, I address the problems created by the lack of local palaeoenvironmen-
tal data as well as the limitations of the use of proxies and their reliability.
I then briefly present the results of my study comparing the archaeological
settlement patterns and evidence of climatic change for the Late Hellenistic,
Roman and Byzantine periods in the Petra region, southern Jordan. Finally,
I would like to call into question the notion of favourable/unfavourable cli-
mate, which has been rather indiscriminately used in archaeological and
historical explanation in the past.
Background
The study discussed in this article is part of my recently finished PhD re-
search, in which I studied the settlement and land-use change in the hin-
terland of the ancient city of Petra, southern Jordan (Fig. 1), from c. 300 bc
until c. ad700. The research was carried out under the Finnish Jabal Harun
Project (henceforth the FJHP), an archaeological project under the research
umbrella of the Finnish Academys Ancient Greek Written Sources Centre
of Excellence in Research.2 The FJHP consists of two interrelated parts: the
excavation of a Byzantine monastery/pilgrimage centre on the top plateau
of Jabal Harun, c. 4km SW of Petra, and the intensive archaeological survey
of the immediate surroundings of the mountain.
The present-day climate of the Petra region is semi-arid, with dry, hot
summers and winter rains. There is considerable change in the relief within
a distance of only 15km from the floor of Wadi Araba at c. 100 m asl to the
elevation of 1700m at the highest parts of Jabal ash-Shara on the western
edge of the Jordanian Highland. Rainfall is mostly generated when moist
air masses are lifted over the Jabal ash-Shara range, resulting in the highest
mean annual rainfall, c. 200mm, in the area. Towards the west and east the
rainfall decreases, being only c. 50mm on the floor of Wadi Araba and in
Maan in the eastern pre-desert zone.3 Due to the low rainfall, agriculture is
not possible in most parts of the region without some means of irrigation.
Extensive ancient systems of rainwater collection and runoff cultivation are
known in the region, dating at least from the Nabataean period onwards.
2 20002005, 20062011.
3 Lane and Bousquet 1994, 22; Meteorological Department of the Hashemite Kingdom of
Jordan 2004.
relating environmental history to human settlement 199
Fig. 1. The Petra region, southern Jordan.
At the turn of the Common Era, southern Jordan belonged to the Naba-
taean kingdom. Within the last couple of centuries bc, the formerly nomadic
Nabataeans had established their capital city in Petra and had begun to
settle the surrounding areas. In the 1st to early 2nd century ad an unprece-
dented expansion of sedentary settlement and agriculture took place in the
Petra region, attested in the archaeological record by the appearance of
numerous farmsteads, hamlets and villages. The Nabataean kingdom was
annexed by Rome in ad106, and Petra became the capital of the province of
Arabia. The annexation apparently did not cause any large-scale disruption
in the land-holding system, and the region continued to prosper under the
Roman rule. However, a considerable reduction in the number of rural sites
is visible from the 3rd century ad onwards.4
4 Kouki 2009, 3539.

200 paula kouki
In earlier scholarship, the initial expansion of settlement and agriculture

by the Nabataeans has been related to a favourable climatic phase.5 The
opposite phenomenon observed in the archaeological record for the Late
Roman and particularly the Byzantine period (Table 1) has been explained
by environmental degradation as the result of over-irrigation and erosion
that resulted in the reversal of the population into nomadism.6 The assump-
tion is reasonable, since the entire region can be considered environmen-
tally marginal for agriculture and settlement, and a considerable part of
its population has until recently consisted of mobile or semi-sedentary
Bedouin. However, there has not been much actual research into the ques-
tion. Therefore I made an attempt to find out whether climatic change could
indeed explain the archaeologically observable changes in the settlement
pattern.
Table 1. The relevant archaeological periods in Jordan.
period dating
Hellenistic 33263bc
NabataeanEarly Roman 63bcad106
Late Roman ad106324
Early Byzantine ad324491
Late Byzantine ad491634
Islamic ad634 onwards
Environmental Data
The immediate problem when dealing with environmental change in a

semi-arid to arid area such as southern Jordan is the lack of local sources
for environmental data. No wide-ranging environmental studies have been
carried out in the Petra region, and most of those dealing with certain
localities or sites are concerned with earlier periods. The available data is
mainly based on sedimentary records, and its temporal resolution is low. The
same goes for southern Jordan in general: the data are sporadic, and their use
as a climatic indicator is further complicated by the contexts where they are
5 MacDonald 2001, 375376.

6 Hart 1986, 58; Fiema 2006, 82.
Fig. 2. A modern barrage in the Jabal Harun area in September 2011.
obtained, since these are often affected by human activity. High-resolution

data from a tree-ring series are available from the Dana area, but they cover
only the recent historical period.7
For example, there are pollen series from sediments deposited behind
a large barrage or reservoir in a tributary of Wadi Faynan. The presence of
algae spores in the sediments dated to the Byzantine period has led to the
suggestion that the climate of southern Jordan must have been more humid
at the time to enable the presence of standing water throughout the year.8
The hydrological role and purpose of this structure is uncertain,9 but there
are other such walls in the same area. In the Jabal Harun area we have been
able to observe that one heavy winter storm has been enough to fill a modern
barrage in a wadi at the foot of the mountain to the extent that there is
still shallow water behind the dam in late September (Fig. 2). Furthermore,
the results from agricultural experiments in the Negev indicate that the
7 Touchan and Hughes 1999.

8 Hunt et al. 2007, 13291330.
9 Crook 2009, 2433. The suggestion is that these constructions are related to ore process-
ing rather than agricultural pursuits, see Newson et al. 2007, 163164.
202 paula kouki
mean annual rainfall of c. 100mm can be transformed to c. 300500 mm

of effective precipitation by intensive collection of runoff.10 Supposing that
the Khirbat Faynan barrage system was maintained, it can be questioned
whether the rainfall really had to be significantly more abundant than at
present to provide water behind the barrage year round.
Small-scale environmental studies were also carried out by the FJHP sur-
vey. In the Jabal Harun area the most significant group of structures in the
area consists of those related to runoff farming and terrace cultivation: prac-
tically every possible wadi and slope has been farmed at some point of time.
Although the runoff-cultivation systems are notoriously difficult to date, the
dating of the off-site scatter of pottery and the functional interdependence
of the structures strongly suggest that at least the large barrage system to the
west of the mountain was built as a planned enterprise in the Nabataean
period, in the 1st century ad.11 By what is known about the economics of
contemporary monastic communities in Palestine and Egypt,12 it could rea-
sonably be expected that the use of the field system continued while the
monastery was occupied, that is, from the 5th to the 9th century at least.13
However, very little off-site pottery datable to these later periods was found
in the area of the field system. Therefore we attempted to obtain evidence
of the erosion and sedimentation history of the area in order to recognize
the potential construction phases of the runoff-cultivation structures. Sed-
imentary structures were documented in three wadi terraces and sampled
for OSL dating.14
The results of the OSL dating were disappointing: for the most part, the
sediments were not completely bleached before deposition, and therefore
the dates obtained were not indicative of the timing of the deposition
processes. Furthermore, even in the case where we had apparently reliable
dates from well-bleached sediments, the uncertainties of the dateswithin
a couple of hundred years at bestwere too large to confidently associate
the deposition of the sediment with any one period of human activity. The
sedimentary structures were mostly indicative of very rapid depositional
events, high sediment loads and short transport distances. The situation is
typical of a semi-arid headwater area, which largely explains the poor results
10 Evenari et al. 1971, 109; Bruins 1986, 3844.

11 Lavento et al. 2004, 166167.
12 Hirschfeld 1992, 103104.
13 Fiema 2008, 434436.
14 Grn 2001.
of the OSL dating15 and reduces the possibility of correlating the results with
a wider-scale climate or environmental change.
The unavailability of local palaeoenvironmental data necessitated the
use of proxy data, which are mainly available from the Dead Sea and north-
ern Israel.16 The use of such data is, of course, not straightforward because of
the changes in climatic controls when the distance from the Mediterranean
coast increasesthe climate of Israel is much more affected by the circu-
lation over the Mediterranean than that of southern Jordan, which besides
the Mediterranean circulation is affected by the Red Sea Trough, resulting
in greater climatic instability.17 Nevertheless, since at least the general trends
in the climate over the Levant are considered to have been broadly similar
in the past as in the present, with the coastal and northern areas receiv-
ing more rainfall than the southern parts, I considered it possible to infer
regional-scale climatic changes from the proxies also for southern Jordan.
However the proxy data are not without their problems, especially in the
case of relatively recent periods like the Roman and Byzantine periods. The
emphasis of palaeoenvironmental research in the Near East has largely been
on the Late Pleistocene and earlier Holocene, which means that less infor-
mation is available for the more recent times. The climatic changes of the
historical period have also been of lesser magnitude than those of the Pleis-
tocene and Early Holocene, and thus they are more difficult to discern in
the palaeoclimatic records. A further problem is the resolution of the avail-
able data. While the resolution of a couple of hundred centuries may be an
excellent achievement for palaeoclimatologists, it makes all the difference
when working within the closely dated Classical and Late Classical periods.
Finally, the results of different palaeoclimatic studies are ambiguous, and
the scientists themselves disagree on the timing and scale of the aridifica-
tion in southern Levant. While some propose that it began as early as in the
fourth century ad,18 others push its start as late as the seventh century.19 On
the other hand, the Talmudic sources point towards crop failures, hunger
and pestilence in the third century,20 which is not reflected in the available
environmental data.
15 For more detailed results of the OSL analyses, see Kouki 2006, 147153.
16 E.g., Bar-Matthews et al. 1997, 1999; Bookman et al. 2004; Bruins 1994; Enzel et al. 2003;
Frumkin 1997; Heim et al. 1997; Hirschfeld 2004; Orland et al. 2009.
17 Freiwan and Kadioglu 2008, 525529, 533534.
18 Bruins 1994, 308.
19 Hirschfeld 2004, 133.
20 Sperber 1978, 7099; cf. Bruins 1986, 307308.
204 paula kouki
Furthermore, the reliability of even these data is questionable. It is gen-

erally agreed that pollen is not a very reliable indicator of the climate for the
historical periods in the Near East due to the widely spread and long-lasting
influence of human activity on the landscape. If we consider the Dead Sea,
the reliability of the sedimentological record and lake levels can also be
questioned. It is well known that currently the level of the Dead Sea is lower
than ever before due to the increased consumption of water by the neigh-
bouring countries, which results in very little of the flow actually reaching
the Dead Sea basin. If we consider the scale of agriculture and land use in
the Levant in the Roman and Byzantine periods, could it not be possible
that it also had some, albeit smaller, impact on the Dead Sea levels and the
sedimentary record?
Settlement and Climatic Change in the Petra Region
The pottery sequences from the Petra region are relatively well known
for the Nabataean-Roman and Byzantine periods, although there is some
uncertainty concerning the pottery of the 2nd3rd centuries and the Late
ByzantineEarly Islamic transition (late 6thearly 7th century). Neverthe-
less, the pottery chronology enables the dating of the archaeological sites
within the accuracy of a century or two, which makes it possible to compare
the trends in settlement distribution and climate on a centennial scale.
To establish the potential influence of climate change on settlement pat-
terns, I divided the Petra region into four rough environmental zones (Fig. 3).
The Jabal ash-Shara zone is environmentally the most favourable, with the
highest mean annual rainfall and most springs. It is the only part of the
region where dry farming of cereals is possible under the current climatic
regime. Until recently, the sedentary villages in the region have been mainly
limited to this zone. The narrow Escarpment zone receives less rainfall, but
it is fed water by the large wadis running east-west, and its topography and
soils are particularly well-suited for runoff cultivation,21 for which there is
also abundant archaeological evidence from at least the Nabataean period
onwards. The environment of the Wadi Araba and Eastern Highlands zones
is the most unfavourable, with the lowest mean annual rainfall and the high-
est spatial and temporal diversity in the distribution of rainfall.22 As a result,
21 Bruins 1986, 3943; Yair and Kossovsky 2002, 55.

22 Freiwan and Kadioglu 2008, 529.
Fig. 3. Environmental zones in the Petra region.
these zones are likely to be severely affected by a decrease in rainfall caused

by the aridification of the climate.
I then compared the centennial distribution of settlement through these
zones, combined from three different surveys recently carried out in the
Petra region,23 to the climate history reconstructed from the proxies. My
hypothesis was that in the case of adverse climatic conditions (i.e., aridifica-
tion), settlement and agriculture would withdraw particularly from the envi-
ronmentally marginal zones of Wadi Araba and Eastern Highlands. Accord-
ing to the same principle, the Jabal ash-Shara zone should continuously be
the most densely settled.
The palaeoclimatic evidence for the relevant period is presented in Table
2 together with the settlement trends. The results were somewhat surpris-
ing. The expansion of rural settlement took place mainly in the Jabal ash-
Shara and Escarpment zones as predicted (Fig. 4), but the intensification of
settlement and agriculture lagged behind the improved climatic conditions
23 Amr et al. 1998; Amr and al-Momani 2001; Abudanh 2006; and the unpublished data
from the FJHP survey.

206 paula kouki
by at least two centuries. The increased rainfall24 set good conditions for
the intensification of agricultural production in the Petra region. However,
the intensification of rural settlement lagged behind the beginning of this
humid period by a couple of centuries. The main expansion of rural set-
tlement occurred only in the 1st century ad. If there was a drop in rainfall
already at the turn of the 1st2nd centuries ad, as has been suggested,25 it
apparently did not have an immediate impact on the rural settlement.
Table 2. A reconstruction of the climate history of southern Levant and the rural
settlement in the Petra region.
time climate settlement in Petra region

ad800
climate very arid
further concentration of rural
ad700 settlement into a few large sites
end of Petra as a city?
abandonment of many rural sites in

ad600 gradually increasing aridity the late 6th7th centuries
through the 6th and 7th centuries
increase of rural settlement in the
eastern Jabal ash-Shara and Eastern
Highlands during the 6th century
ad500
aridification of climate starts in the
5th century agricultural sites established in
the Eastern Highlands in the 5th
century
ad400 possible drop in rainfall around
ad400 discontinuity of rural settlement
drop in the number of sites
ad300
a return to more humid conditions
in the 3rd century
widespread abandonment of rural
ad200 settlements by the early 3rd century
ad100 a possible drop in rainfall around

ad100
24 Frumkin 1997, 244; Bookman et al. 2004, 566.

25 Frumkin 1997, 244; Orland et al. 2009, 34.
Fig. 4. The distribution of rural settlement in different environmental zones in the

1st century ad.
time climate settlement in Petra region

proliferation of rural settlement
and intensification of land-use in
the humid climatic phase culmi- the 1st2nd centuries
nates around the turn of the era
first rural settlements elsewhere
100bc climate more humid than at

present
first rural settlements in the Jabal
ash-Sharah area
200bc
climate becomes more humid, first permanent settlement in Petra
probable cooling
300bc
before an arid climatic phase no permanent settlement

208 paula kouki

3rd century ad.
However, although relative humidity apparently increased again in the

late 2nd3rd centuries, a widespread reduction of rural settlement appears
to have taken place in all environmental zones over the same period (Fig. 5).
By the 4th century, most agricultural settlements had withdrawn from both
the western and the eastern peripheries of the region, and rural settlement
had largely become concentrated in the Jabal ash-Shara zone. The palaeo-
climatic data concerning the 4th century are somewhat ambiguous, but
most studies support the interpretation that the climate was still relatively
humid.26
Most palaeoclimatic data point towards increasing aridity from the 5th
century onwards.27 The 5th and particularly 6th centuries witnessed a reduc-
tion of the number of rural settlements in the western part of the Jabal ash-
Shara zone. However, this trajectory was accompanied with an expansion of
26 Frumkin 1997, 244; Orland et al. 2009, 3334; Enzel et al. 2003, 268; Bookman et al. 2004,
566.
27 Frumkin 1997, 244; Orland et al. 2009, 34; Enzel et al. 2003, 268, cf. Bookman et al. 2004,
566.

6th century ad.
settlement and agriculture into the Eastern Highlands zone, with the estab-
lishment of large agricultural settlements in the desert fringe, particularly
in the 6th century (Fig. 6). Such development seems to match poorly with
the evidence for a drying climate, when an opposite phenomenon could be
expected.
Finally, a considerable reduction of settlement again took place in the 7th
century (Fig. 7), with the concentration of settlement into a few large and
medium sites. In the eastern part of the Jabal ash-Shara zone, most of the
small settlements established in the previous century were abandoned. In
the western part of Jabl ash-Shara, the village settlements seem to become
localized in the areas of modern Wadi Musa and at-Tayyiba. However, the
large agricultural settlements in the Eastern Highlands zone show continu-
ity through the 7th century and even beyond, which was not to be expected
in the light of the evidence of further aridification.28
28 Frumkin 1997, 244; Enzel et al. 2003, 268; Bookman et al. 2004, 566567.
210 paula kouki

7th century ad.
Conclusion
As can be seen from the above, I was not able to establish a connection
between the archaeological settlement patterns and climatic change in
the Petra region for the Nabataean-Roman through the Byzantine periods.
Although the initial expansion of rural settlement and agriculture takes
place during a period of favourable climate, it only begins a couple of
centuries later than this wetter climatic phase. On the other hand, the
reduction of rural settlement begins during what can be considered a still
relatively favourable climatic phase. Furthermore, although most scholars
agree that aridification had begun by the later 5th century at the latest,
there is a phenomenon of new agricultural settlements being established
in the eastern part of the Petra region, in the area that lies in the desert
margins, in the 5th and 6th centuries. Actually, the focus of the perma-
nent settlement apparently shifted towards the east in the 5th century, and
this eastern focus became even more pronounced in the 6th and 7th cen-
turies.
Two obvious alternative conclusions can be drawn from these results:
1) The reconstruction of the past climate is correct. The settlement

change observed is not related to climatic change, but to other factors
such as political, economic and social changes.
2) The reconstruction of the past climate is faulty. The resolution of the
available climatic data is too low or the proxies used are unsuitable
for the reconstruction of the climate of southern Jordan, and thus no
conclusions can be made of the relationship of climate and settlement
based on the available data.
A third possible explanation is that the notion of favourable climate is in
itself at fault. To conclude, I would like to consider this possibility a little
more closely.
The environmental data available for the southern Levant is mainly
indicative of changes in rainfall and/or runoff. The periods for which in-
creased rainfall is implied are usually considered favourable, while those
with less rainfall are considered unfavourable. I would like to argue, how-
ever, that the definition of what is favourable climate is not so straight-
forward, and that it is not possible to reduce it into a simple function of
the mean annual rainfall. The timing, spatial distribution and mode of rain-
fall are equally important. The temperature both during and after the rainy
season affects evaporation and thus influences the actual available mois-
ture. Furthermore, local topography, geology and soils influence the out-
come through the differential generation of runoff, absorption of rainwater
into soils and erosion potential. Even short-lived storms can generate a lot
of runoff and potentially catastrophic erosion and hazardous floods in a
sparsely vegetated region, but they also enable the effective collection of
runoff for later use. On the other hand, long-lasting but relatively low inten-
sity rainfall is more useful in areas with thicker soilsprovided that the
temperatures are low enough that the moisture will be retained in the soil
and does not evaporate. The Petra region exemplifies that these conditions
can vary considerably even within a relatively small geographical area.
Finally, and perhaps most importantly, we should not forget that the
response of human communities to changing environmental conditions is
not mechanistic and unchangeable. From the viewpoint of people, the opti-
mal characteristics of rainfall depend on the cultivated plants and available
technologies and techniques used for irrigation and water storage, as well
as on traditions of land use. Thus we need not only more detailed environ-
mental and palaeoclimatic data but also a more thorough picture of ancient
societies to be able to evaluate the impact of climatic change on a given soci-
ety with any confidence.
HUMAN-ENVIRONMENT INTERACTIONS IN
THE SOUTHERN TYRRHENIAN COASTAL AREA:
HYPOTHESES FROM NEAPOLIS AND ELEA-VELIA
Elda Russo Ermolli, Paola Romano and Maria Rosaria Ruello
1. Introduction
Reconstructing past landscapes is a challenging task, especially when one is

dealing with those regions that have witnessed the presence of man since
prehistoric times. In these areas the natural evolution of the environments,
driven by climatic and/or endogenous factors, is strictly and deeply related
to the continuous land use and management by man. On this assumption,
it has become clear that any attempt of landscape reconstruction in archae-
ological contexts has to rely on the systematic integration of archaeology,
geomorphology and other scientific disciplines. Many recent studies1 have
shown, for instance, that humans have influenced large-scale sediment pro-
duction and movement, and can be considered, like climate, as one of the
main agents of sediment movement since the first millennium bc.
Detailed climatic reconstructions of the last 2500 years were recently pro-
vided for central Europe, thanks to a tree ring-based methodology.2 Wet and
warm summers occurred during periods of Roman and medieval prosper-
ity while increased climate variability from ~250 to 600 ad coincided with
the demise of the western Roman Empire and the turmoil of the Migra-
tion Period. Such detailed climatic reconstructions are not yet available for
the Mediterranean area, where the climatic events could have had a differ-
ent character and/or intensity from those recorded in central Europe. So,
in this region, linking the response of geomorphic systems to the interac-
tion between land use and extreme climatic events is not straightforward.
Increasing the acquisition of knowledge about this matter is the only prag-
matic way to come near to a possible solution to the debate, and the present
contribution represents a further step towards the clarification of the long-
lasting question: man, climate or both?
1 e.g. Casana 2008, 431439.

2 Bntgen et al. 2011, 579582.
214 elda russo ermolli, paola romano, maria rosaria ruello
Fig. 1. Location map of the study area.
2. The Study Areas
This study focuses on two archaeological sites of the Greco-Roman period

located along the western coasts of southern Italy. In this part of the penin-
sula, facing the Tyrrhenian Sea, several archaeological sites of human settle-
ments dating back to the period of Greek colonization are present.
2.1. Parthenope-Neapolis
The first site is located in the modern town of Naples where the Greco-
Roman town of Parthenope-Neapolis was situated. The landscape of the area
is characterized by favorable conditions that have facilitated human settle-
ments since Neolithic times: cliffed promontories alternating with narrow
coastal plains offered timber sources and, at the same time, protected land-
ing places. This landscape was inherited by the third epoch (48003800 yr
bp) of activity of the Campi Flegrei volcanic field3 which led to the formation
3 Di Vito et al. 1999, 244.

human-environment interactions: neapolis and elea-velia 215
of tuff ring and cones, as well as of straight fault scarps (Fig. 1). The volcanic
activity of this area started about 50 thousand years ago and is still active
today;4 the last eruption occurred as recently as 1538 and caused in the space
of a few days the emergence of Monte Nuovo. To the east, the landscape
is almost flat and gradually passes to the Vesuvius foot-slope. This volcano,
more than 1,000 meters high, is also of course still active.5
Parthenope was settled by the Greeks during the eighth century bc on
the hilltop of Monte Echia promontory6 as a small outpost belonging to the
northern town of Cuma. Subsequently, at the end of the sixthbeginning
of the fifth century bc,7 the major Greek town of Neapolis was established
at the foot-slope of Monte Echia, leading to the progressive abandonment
of the military outpost which was renamed Palaeopolis. Neapolis developed
as a municipium and then as a colonia under the Roman Empire. Dur-
ing Greco-Roman times port activities developed along the coastal strip of
Neapolis, favored by the presence of a protected bay. The ancient port was
discovered because of construction work for Naples new underground rail-
way.8
2.2. Elea-Velia
The second site is the Greco-Roman town of Elea-Velia on the coast of the
Cilento, about 200km south of Neapolis. This sector of southern Italy is char-
acterized by a gently-sloping hilly landscape, reaching hundreds of meters,
made up by soft silica-clastic sediments of Mesozoic and Cenozoic age.9
Pliocene cemented conglomerates unconformably cover the soft bedrock,
leading to the emergence of some rocky ridges.
The town of Elea-Velia was founded around one of these rocky hills in the
sixth century bc. The acropolis rose on the hilltop of a protruding cliffed
promontory at 200m asl. Subsequently, the town developed in the lowland,
at the mouth of the Frittolo river, where a densely urbanized area, namely
the Southern Quarter, was built.10 The entire history of the Greco-Roman
town was influenced by the Frittolo stream dynamics which caused several
4 De Vivo et al. 2010, 1011.

5 Sigurdsson et al. 1985, 332384.
6 Giampaola 2011.
7 DAgostino and Giampaola 2005, 6372.
8 Giampaola et al. 2006, 51; Amato et al. 2009, 2630; Carsana et al. 2009, 1721.
9 Bonardi et al. 1988.
10 Greco 2003.
alluvial floodings in the Southern Quarter. Episodes of flood were recorded

until late ancient times,11 but the absence of subsequent historical reports
about Velias history points to the towns disappearance beneath sediments
and shrubs during the Middle Ages. Archaeological excavations carried out
during the twentieth century removed some meters of the thick alluvial
strata, unearthing the ruins of the Southern Quarter.
2.3. The Landscape

In both areas the landscape was characterized by a rich vegetation cover.
This conclusion relies on pollen data obtained from the analysis of a sea core
(C106) in the Salerno Gulf.12 The position of this core is equidistant from the
two studied sites (Fig. 1), and moreover this position allows the pollen rain
from a very wide area to be represented in the analyzed sediments. Indeed,
the source area for pollen can be considered the whole catchment basin of
the river Sele and other minor basins.
The pollen diagram of the Salerno core covers the last 30,000 years (Fig. 2).
testifying to the transition from the Last Glacial period to the Holocene.
Before Greek colonization, pollen data clearly show that a deciduous forest
dominated by oaks probably occupied all the slopes surrounding the study
sites. Mediterranean maquis was present as well on the steepest and sun-
niest slopes and close to the coast on thin soils. On mountain tops above
1000m a fir forest was present which has now almost completely disap-
peared from the southern Apennines, to be replaced by beech woods.
3. Methodological Approaches
This study concerned different aspects of both the natural and human land-
scape and involved researchers in geological, environmental and archaeo-
logical sciences.
The main goal was to reconstruct the ways and times of environmental
changes and to understand their causes. For this purpose three steps in the
evolution of these two coastal sectors were chosen, representing the main
detectable changes in the environmental features.
11 Amato et al. 2010, 1415.

12 Russo Ermolli and Di Pasquale 2002, 211219; Di Donato et al. 2008, 153156.
Fig. 2. Pollen diagram from the C106 core in the Salerno Gulf. Taxa percentages are plotted against age (ka = thousand years). AP= arboreal
pollen; LGM= last glacial maximum; B/A= Bolling/Allerod period; YD= Younger Dryas cold spell (modified after Di Donato et al., 2008)
The first step concerns the landscape setting prior to Greek colonization
in the first millennium bc; the second step concerns the Greco-Roman
period; while the third step focuses on the Late Ancient period, from the
third century ad onwards.
The methods used to achieve these goals can be summarized in three
points:
1. the first approach is normally the collection of direct data from the
excavation areas and from boreholes;
2. secondly, these data are integrated to reconstruct the 3-dimensional
arrangement of the strata and their bounding surfaces and
3. finally, all the local reconstructions are integrated to obtain a wider
paleogeographical reconstruction.
4. Evolution of the Neapolis Area
The environmental evolution of the Neapolis region is shown in Fig. 3

through three-dimensional schemes obtained from a digital elevation
model (DEM) and from cross sections.
4.1. Step 1First Millennium bc

In the first step of environmental evolution, the coastal profile of the Neapo-
lis region was located further inland with respect to the modern one (Fig.
3A). Promontories and inlets characterized the coast which was formed by
tufaceous cliffs that in some places were tens of meters high. Slopes were
cut by a dense fluvial network.
The section of Fig. 3A was built along the line indicated in Fig. 3A thanks
to the data coming from boreholes and trenches dug during the construc-
tion of the new underground line of Naples.
In this section, cutting the piazza facing the modern port, it is possible
to visualize the ancient coastline which was about 500 meters inland with
respect to the present one. The tufaceous cliffs directly faced the sea without
any beach strips. The bedrock and the cliffs were made up by the Neapolitan
Yellow Tuff which was emplaced 15,000 years ago from one of the main
eruptions of the Campi Flegrei volcanic fields.13
13 Deino et al. 2004, 168.

Data from boreholes and archeological sections14 allowed the morphol-

ogy of the bedrock to be reconstructed as well as the thickness of marine
deposits mainly concentrated in the depressions.
4.2. Step 2Greco-Roman Period

The second evolutionary step of the coastal environment of Neapolis (Fig.
3B) shows the landscape towards the end of the second century ad, after
the towns foundation. In the sixth century bc Neapolis was founded on the
terraced surface gently dipping towards the sea. Its shape was driven by nat-
ural features: it was bounded by streams at the sides and by a cliff toward
the sea. The city walls more or less followed these natural boundaries. In
Fig. 3B it is also possible to see the main streams and the main Greco-Roman
roadsat least those that have been discovered. They led towards the west-
ern region where other towns were established, such as Puteoli and Cuma
(Fig. 1). The first nucleus of Neapolis now represents the historical center of
Naples where the main streets still follow the traces of the Greco-Roman
streets and where much archaeological material has been found.
Concerning the coastal landscape, a sheltered bay was located at the cliff
base in the western side of the gulf. A small tufaceous promontory protected
the bay from the main sea storms coming from southwest. In this bay, port
activities probably started from the fourth century bc and went on until the
fourth century ad when this part of the bay dried up. The main environmen-
tal difference with respect to the previous stage is the formation of a sandy
beach in the eastern coastal sector, which was occupied by human activities
that developed in this period also outside the city walls. In fact, remains of
a temple and of the gymnasium were found on the ancient beach.15
The section of Fig. 3B refers to this second stage and shows the beach
that formed at the cliff toe and the construction of the port. The latter, as
previously noted, was located in a sheltered inlet, and the promontory which
protected the bay from the southwestern storms is also visible in the section.
The beginning of port activities started between the fourth and the sec-
ond century bc; traces of dredging carried out at this period were found on
the sea floor (Fig. 3B). Dredging was carried out to lower the sea bottom
and to make the inner part of the basin suitable for shipping. A wooden
dock is dated to the first century ad while two shipwrecks were found in the
sediments dated to the turn of the first and second centuries ad (Fig. 3B).
14 Amato et al. 2009, 2530.

15 Giampaola 2004, 44.
Fig. 3. Three-step scheme of the evolution of the Neapolis area. A= Step 1I millen-
nium bc; A= cross section related to Step 1; B= Step 2Greco-Roman period; B=
cross section related to Step 2; C and D= Step 3since the Late Ancient; C and D=
cross sections related to Step 3 (after Russo Ermolli et al., in prep.).
The many repairs to the timber of these ships, together with the absence of
any cargo, suggests that these ships were probably abandoned on purpose.16
Wood analysis of the shipwrecks revealed the systematic use of walnut and
cypress timber. This datum, that is the use of local plants, together with the
modest dimension of the ships, suggests that the location of shipyard has to
be restricted to the central-southern Tyrrhenian coasts.17
4.2.1. The Landscape of Neapolis through Pollen Analysis

The sediments of the port were excavated for about 7 meters. The rich
archeological content allowed the sediments to be dated from the second
century bc to the fifth century ad.18 These deposits were studied in detail
through different methods with the aim of reconstructing the underwater
environment as well as the terrestrial landscape surrounding the site. For
this purpose sediment samples were collected for pollen analysis.
In Fig. 4 the results of pollen analysis are shown in a detailed diagram
where all the recognized taxa are grouped in vegetation associations marked
by different grey shades. It is necessary to underline that pollen spectra from
very urbanized areas are the sum of natural and anthropogenic plants and
thus their interpretation has to be very cautious, especially when one is
dealing with plants that could be either wild or cultivated. This applies, for
instance, to hazelnut, olive and chestnut. Another problem concerns the
herbs which are only determined at the family rank and families such as
Poaceae, Brassicaceae and Fabaceae could include both wild and cultivated
varieties.
On the whole, it is possible to say that from the first century bc to the
second century ad a deciduous oak forest was present on the slopes sur-
rounding the Neapolis harbor and that the Mediterranean maquis probably
occupied the most sunny and rocky sectors, especially close to the sea coast.
These data confirm the landscape reconstruction based on the pollen dia-
gram from the Salerno Gulf (cfr. 2.3). The tree crops mainly consisted of
walnut and secondly of chestnut and grapevine. Concerning the horticul-
tural practices, the presumed cultivated varieties are dominated by the Bras-
sicaceae family, most likely representing cabbage cultivation, which was
rather common in Roman times and represented one of the main plant food
sources for the Romans.
16 Giampaola et al. 2006, 6376.

17 Allevato et al. 2010, 23732374.
Fig. 4. Pollen diagram from the Neapolis port sediments. Taxa percentages are plotted against depth. DF= deciduous forest; TC= tree crops;
MM= Mediterranean maquis; MF= montane forest; HS= herbs; HT= horticultural; WT= water plants; AP= arboreal pollen (after Russo
Ermolli et al., in prep.).
Very few pollen are representative of the mountain forest which was
located rather far from Neapolis, on mountain tops over 1000 meters and
probably also on Vesuvius which was higher than today.
4.3. Step 3Since the Late Ancient

In the third step of environmental evolution at Neapolis (Fig. 3C), the main
episode is the closing of the port area and the formation of a lagoon, first
connected to the sea and then isolated. This closure took place in the fifth
century ad thanks to the growth of a beach ridge. Port activities in this part
of the bay ended even before, during the fourth century, when the town
walls expanded westward. Towards the end of the fifth century the bay was
completely filled up, and the site was used as a farmland from the beginning
of the sixth century.
The section of Fig. 3C shows the lagoon and the beach ridge which
formed above a peak in the tufaceous bedrock. The reduced water depth
and the closing connection to the sea is evidenced by the microfaunal
association which changed from marine to brackish.19
A third shipwreck was found in the sediments dated between the end
of the second and the beginning of the third century. Since the definitive
emergence of the area at the end of the fifth century, the lagoon deposits
have been covered by continental sediments as shown in Fig. 3D. In the
section of Fig. 3D the subsequent anthropogenic filling and the modern
ground level are also shown.
4.3.1. Pollen Results from the Third to the Fifth Century ad

Pollen data from the port sediments show a drastic decrease in cabbage cul-
tivation during the third century ad (Fig. 4). This decline was accompanied
by an increase in Mediterranean maquis and in some elements of deciduous
forest. Such change can be interpreted as a phase of abandonment of veg-
etable gardens in this area of the town and the contemporaneous spread of
wild vegetation. This could mean a decreased human presence and conse-
quent decrease in horticultural activities which favored the increase in tree
crops, requiring less care than vegetables. Indeed, the main peak of chestnut
is recorded in this period. After this moment the situation appears to have
been restored and in the fourth and fifth centuries the vegetation is very
similar to that preceding the third century, apart from a generally higher
19 Bourillon 2005, 38.

presence of oaks and a continuous presence of walnut and grapevine. The

olive peak, restricted to the third century, suggests that it was probably the
wild variety.
It is interesting to observe that at the same time, namely during the third
century, port activity also had a period of stagnation.20
5. Evolution of the Velia Area
The environmental evolution of the Velia region is shown in Fig. 5 through

three-dimensional schemes obtained from a digital elevation model (DEM)
and from cross sections.
5.1. Step 1First Millennium bc

As in the Neapolis region, the coastline of the Elea- Velia region was located
about 800 meters further inland with respect to the modern one (Fig. 5A).
This paleo-coastline position, prior to the development of the Greco-Roman
town, was reconstructed thanks to borehole data which highlighted the
occurrence of marine sediments at ca. 0m asl21 at the cliff base, covering
the fluvial and slope deposits forming the bedrock and the cliffs.
The coastal profile of the area consisted of promontories and inlets with
rocky cliffs made up of cemented fluvial conglomerates. The section of
Fig. 5A, crossing the slope and the lower quarters of the future Greco-Roman
town, is based on data coming from some boreholes stratigraphies and
archaeological excavations.
5.2. Step 2Greco-Roman Period

In the second step of environmental evolution in the Elea Velia area (Fig. 5B)
the main difference with respect to the previous step is the progradation of
the coastline and the formation of sandy beaches at the cliff toes.
Elea-Velia rose in the sixth century bc on a hill surrounded to the west-
ward by the sea. In the fifth century bc it developed downhill, as far as the
narrow emerged littoral plain, with the building of the Southern Quarter.
Some remains of the main structures are still visible, such as the Acropolis
on the top hill, the Temple of Asklepius and the Roman baths, both located
in the thalweg of the Frittolo stream (cfr. section 3.2), and the Southern

21 Ortolani et al. 1991, 167.
Fig. 5. Three-step scheme of the evolution of the Velia area. A= Step 1I millennium
bc; A= cross section related to Step 1; B= Step 2Greco-Roman period; B= cross
section related to Step 2; C= Step 3since the Late Ancient; C= cross section related
to Step 3 (after Romano et al., in prep).
Fig. 6. The road to Porta V at Velia. The wall shows several restoration phases which
were realized in order to compensate for the alluvial deposition at its back.
Quarter with the Necropolis in the plain. As in the Neapolis case, at Velia
too the town walls followed the natural features and in particular the main
divide of the stream basins.
The life in the Southern Quarter of the town was strongly influence by the
torrential regime of the Frittolo stream. In fact, phases of peak in discharge
were followed by alluvial floodings of the town causing damaging to the
urban structures. This influence is clearly shown in the section of Fig. 5B
and in Fig. 6 where it is possible to see the buildings and walls subjected to
several phases of restoration in order to compensate for the ground level
aggradation. This aggradation is represented in section by the layers of
alluvial sediments coming from the stream which alternate with the walls
ruins.
The coastline was probably located very close to the town as testified
by the storm sediments covering some Roman tombs in the Necropolis,22
outside the towns walls.
5.3. Step 3Since the Late Ancient

Starting from the third century ad several new phases of intensive over-
flooding, mainly supplied by the Frittolo stream, completely covered the
Southern Quarter (Fig. 5C and 5C). The cross section of Fig. 5D shows the
22 Ruello 2008, 262.

present situation where thick alluvial levels were removed by the archaeo-
logical excavation carried out in the twentieth century in order to unearth
the town ruins.
The alternation of alluvial phases and restoration phases is clearly shown
in several sectors of the town where the walls of buildings were rebuilt
several times above the strata of alluvial deposits fed by the stream. In
particular, the main rebuilding phases can be dated to the end of the first
century bc, the second century ad, the beginning of the third century ad
and the end of the third century ad.
The sedimentary structures characterizing the alluvial deposits can be
alternately referred to debris flow and flood processes. The thickness of each
alluvial episode (i.e. the thickness of the alluvial strata) clearly rises upward,
testifying to the increase in both sedimentary supply and stream power, par-
ticularly during the third century ad (Fig. 7). These traits are typical of flash
flooding, that is rapid response events to intense precipitation episodes.
The fact that the flooding from the beginning of the third century ad
onwards was more severe is also clearly visible close to the apical zone of the
alluvial deposition, where remnants of the road leading to Porta V (Fig. 6)
are found. The wall beside the road displays levels of elevations dated to
the second and third centuries ad, buried by alluvial strata increasing in
thickness during the third century ad (1.7 meter/100 years). This exceptional
increase in sedimentary inputs starting from the third century is most likely
linked to declining land use management on the slopes behind the town,
especially if we consider the size of the Frittolo basin, the major stream
flowing toward the town, which is very small if compared to the huge
amount of sediments produced.
6. Conclusions
The evidence collected up to now in the two study sites leads to certain
conclusions. Throughout the time period analyzed, namely from the first
millennium bc, both sites recorded a significant coastline progradation,
which can be quantified in more than 500 meters at Neapolis and more than
800 meters at Velia. In the third century ad, evidence of acceleration in both
progradation and aggradation processes is recorded in the two study sites.
At Neapolis, this acceleration caused the formation of a beach ridge which
isolated the port transforming it into a lagoon. At the same time, pollen data
show the abandonment of vegetable gardens, at least in the port area, and
the spread of wild vegetation.
Fig. 7. Section of the alluvial deposits at Velia. The thickness of the alluvial strata
clearly rises in the III century ad testifying to the increase in both sedimentary
supply and stream power.
At Velia, the sediments of the third century ad are the evidence of flash
floods, that is: rapid response events to intense precipitation episodes. The
alluvial strata show an increased thickness with respect to those of the
preceding centuries, testifying to the occurrence of more severe flooding.
It is known from proxy-based reconstruction of the north-central Euro-
pean climate that a higher variability characterized the period from the
third to the seventh century ad, but in the study sites it seems clear that
the presence/absence of man played a crucial role in influencing the slope
dynamics. It is thus rather probable that the effects of particular land use
conditions were enhanced by extreme climatic events creating a window
of opportunity23 in which episodes of severe erosion occurred.
23 Bintliff 2002, 417435.

LARGE-SCALE WATER MANAGEMENT PROJECTS
IN ROMAN CENTRAL-SOUTHERN ITALY*
Duncan Keenan-Jones
Tot aquarum tam multis necessariis molibus pyramidas videlicet otiosas com-
pares aut cetera inertia sed fama celebrata opera Graecorum.1
One may plainly compare such indispensible and enormous structures, car-
rying so much water, with the indolent pyramids and the other useless yet
renowned and celebrated works of the Greeks.
quod si quis diligentius aestumaverit abundantiam aquarum in publico, bali-

neis, piscinis, euripis, domibus, hortis, suburbanis villis, spatia aquae venien-
tis, exstructos arcus, montes perfossos, convalles aequatas, fatebitur nil magis
mirandum fuisse in toto orbe terrarum.2
If anyone were to consider more carefully the vast amount of water for the use
of the public, in baths, in pools, in channels, in homes, in gardens and in sur-
burban villas, the distances traversed by the water, the arcades constructed,
the mountains pierced and the valleys levelled, he would say that there was
nothing more amazing in the whole world.
Extensive alteration of hydrologic systems was a conspicuous feature of
ancient Roman civilization. Water was channelled and directed for a myriad
of different purposes: drinking, aesthetic display, bathing, irrigation, indus-
trial uses, land reclamation and waste disposal. Rows of arcades supporting
water channels were dominant features on the Italian plains. Such arcades
maintained the elevation of water brought from the Apennines so that it
could reach the very highest points of urban centres. Building on Etruscan
practices, marshes and lakes were drained partially or entirely in order to
reclaim land for farming.3
This paper considers two case studies, both involving inter-basin trans-
fers of water, in order to investigate elite Roman attitudes towards the use
* The research for this paper was generously supported by a Macquarie Gale British
School at Rome Fellowship, a Macquarie Gale Travelling Fellowship, a Friends of Hercula-

neum Society Studentship and two Borse di Studio from the Italian Government.
1 Frontinus, De Aquis 1.16.
2 Pliny the Elder, Naturalis Historia 36.24.123.
3 Purcell 1996, 190199.
234 duncan keenan-jones
of water resources, particularly their attitudes towards deleterious conse-

quences of such use on other communities and the wider environment.
Before turning to the case studies, we must first consider how the hydrolog-
ical cycle has changed in Central-Southern Italy since antiquity so that we
can make use of modern data, and also consider the usage and legal control
of water in rural areas.
Hydrological and Climate Change

in Central-Southern Italy
Variables of particular interest to the hydrologic cycle are temperature,

which affects evaporation and transpiration (evaporative loss of water from
plants), and rainfall. These two variables exert a powerful influence on
river flow (including flooding) and groundwater recharge, which, in turn,
controls spring flow rates.4
The basic picture of the Mediterranean climate, that of a marked sea-
sonality with dry summers and abundant rainfall only in winter, seems to
have been begun between 550 and 50bc and has remained until the present
day.5 This seasonal variation, and the resulting reliance on spring water dur-
ing summer, is confirmed for the Apennine springs of Campania in the
early fifth century ad by Paulinus of Nola.6 The numerical values of rain-
fall and temperature in antiquity, however, would not necessarily have been
comparable to those of today. Even over the short period of time since the
beginning of record keeping, annual rainfall in Central-Southern Italy has
decreased by 18%.7 The rainfall pattern has also changed, resulting in fewer,
more intense rain events, which lead to more run-off, decreased infiltration
and hence decreased groundwater recharge.
In general, in Central-Southern Italy over the last 3000 years, the cli-
mate has alternated between warm, dry periods and cool, wet periods.8 The
warm, dry periods, through increased evapo-transpiration and decreased
rainfall, would have seen reduced river and spring flow rates. The opposite
would have been true for the cool, wet periods, which, through increased
rainfall, saw increased flooding and sediment transport in rivers. Giraudi
and co-workers have recently provided the first comprehensive reconstruc-
4 Chen, Grasby, et al. 2004; Fiorillo et al. 2007; Dragoni and Sukhija 2008, 4.
5 Peyron et al. 2011.
6 Carmen 21.752753.
7 Fiorillo and Esposito 2006, 1.
8 Dragoni 1998, 261.
large-scale water management projects 235
tion of climate in Central-Southern Italy over the past 10,000 years.9 They
used Apennine glacial advances and retreats (which basically correlate dur-
ing our period with those of other European mountain chains and with
evidence for large amounts of ice in the North Atlantic) and related pro-
cesses as proxies for temperature. Lake levels and 18O (stable oxygen iso-
tope composition) of carbonate deposits from caves (including stalagmites)
and lakes (shells) were used to estimate the water balance (precipitation
versus evapo-transpiration). These data are believed to represent climatic,
rather than human-induced, changes, although this is less true with regard
to lake levels (see below). The period encompassed by our case studies,
that is to say between the late first century bc and the fifth century ad, lay
between cool and wet periods dated to 850650 bc and ad 550750. Apart
from some evidence for a minor increase in rainfall in the late 2nd/early
3rd century ad discussed below, Central-Southern Italy seems to have been
undergoing a warm, dry climate at this time. The most recent cool, wet
period was between ad14501650, during the period known as the Little Ice
Age. Hence, the late 19th and 20th century were also a comparatively warm
and dry period, happily rendering the period of instrumental records of cli-
mate a reasonable analogue for our period of interest.10 Temperatures and
probably also rainfall, and hence spring discharge, were seemingly compara-
ble to the only data we have available (but see the discussion of flood history
below).
Flood History
The situation regarding the flood history of Central-Southern Italy is more
complex. The flood history of the Tiber River (and of the Po) and the lev-
els of Lakes Fucino (east of Rome) and Accesa (in Tuscany) (Fig. 1) are
well-correlated with Alpine glacial retreat and advance, a proxy for regional
temperature change.11 The exception is the warm period between the 3rd
century bc and the 3rd century ad, a time of increased flooding in the Tiber
despite warmer temperatures. Giraudi claims that this period was one of
increased rainfall, citing increases in the levels of Lake Fucino and Lake
Ledro in northern Italy, and coeval flooding in the Po Basin, while also allow-
ing that human action in the Tiber Basin may have played a major role.12
9 Giraudi et al. 2011.

10 Lamb 1995, 157159.
11 Giraudi 2011, 89.
12 Giraudi 2005; Giraudi 2011, citing Giraudi 1998 for the Lake Fucino data.
Fig. 1. The Italian Peninsula. Based on data provided by ESRI Inc.
The major relevant human activity was probably land clearance and con-
sequent deforestation. The Greco-Roman world witnessed massive land-
clearance and engaged in heavy and prolonged consumption of wood, espe-
cially perhaps in Italy.13 Pollen evidence attests to deforestation in most parts
of Central or Southern Italy starting from around 1700bc or even earlier, with
the process seemingly reaching remote, elevated sited sites such as the Lago
Grande di Monticchio (Fig. 1) around the turn of the era.14 River basins, espe-
cially the Tiber, were particular foci for the wood industry, since the rivers
provided the means to transport the wood to market. Reduction of forest
and other vegetation cover increases the frequency and height of flooding by
reducing the amount of water retained in plants and the soil and increasing
13 See most recently Kaplan et al. 2009, Harris 2011a, and Harris this volume.
14 Ramrath et al. 2000; Allen et al. 2002.
erosion.15 Instead of being retained and slowed in the vegetation, more water
runs off into streams and rivers more quickly. Increased erosion arising from
the lack of vegetation also fills buffers, such as lakes, with silt, reducing their
capacity to slow and minimize flooding. Silt also fills the channels on the
flood-plains, reducing their capacity and preventing the quick movement
of water out of the basin, and raises the height of the flood-plains, again
increasing flood heights.
Of course, similar historical processes to those affecting the Tiber Basin
were also affecting that of the Po at this time. Significant siltation and
progradation at the mouth of the Po testifies to the large sediment loads
carried.16 During the same period, multiple rivers in Basilicata also show
evidence of similar siltation processes that have been attributed to human
activity.17 Slightly later, in the 3rd century ad, we see the beginning of an
acceleration of siltation in rivers at Naples and Velia that has been attributed
to a combination of human activity and climate events.18 Hence, there is
clear evidence for anthropogenic factors that could explain the increased
flooding of the Tiber and the Po, and those of other Mediterranean rivers
such as the Arno (Fig. 1) and the Rhone,19 and we need not posit increased
rainfall on the basis of this evidence.
It is interesting that sediment accumulation in the Ombrone river delta
(not far from Rome, Fig. 1) is well-correlated with those of the Tiber and
Po (and with known climatic cool and wet periods) for the Late Holocene,
except for the Roman period, where there is no large sediment accumula-
tion.20 If this is not due to marine erosion, could it be that there was little
anthropogenic influence on this particular catchment?
The other evidence for increased rainfall during this period is also incon-
clusive. Lake levels are affected by changes in precipitation and evapotran-
spiration, but also by local non-climatic factors, such as changes in land
use.21 As a result, only multiple synchronous movements in lake levels within
a region should be taken as evidence of climatic change. In addition, use of
lake levels as a proxy in Italy in the Holocene is hampered by a lack of com-
prehensive and continuous data at a sufficient number of different sites.
15 Hughes 1994, 8284; Yin and Li 2001; Beven 2003, 529, 538; Walling 2006; Roberts,
Brayshaw, et al. 2011.

16 Hughes 1994, 84.
17 Piccarreta et al. 2011.
18 See the contribution by Russo Ermolli and co-workers to this volume.
19 Magny et al. 2009.
20 Magny et al. 2007, citing Bellotti et al. 2004.
21 Magny 2004; Giraudi et al. 2011.
There are question marks over both lakes cited by Giraudi as evidence.
Regarding Lake Fucino, it is strange that the contemporary publication
by Giraudi and co-workers shows no high stands of Lake Fucino within
approximately 500 years of the turn of the era.22 A possible explanation is
that the increased level shown by Giraudi in his 1998 publication was too
small and/or short-lived to be noted in the longer-term and, hence, lower-
resolution 2011 study by Giraudi and co-workers. This could be because the
increase in level at Fucino cited in Giraudis 1998 article was abruptly cut off
by draining, that is to say by anthropogenic activity. The level of Lake Fucino
could also reflect tectonic activity, adding another potentially confounding
factor.23
Regarding Lake Ledro, there is clear evidence for high lake levels around
the turn of the era.24 How relevant is this site on the southern slopes of the
Alps (Fig. 1) to the climates of Lazio and Campania? If in fact we bring in
a wider survey of lake levels around West-Central Europe, we see that our
period of interest is bracketed by major lake high stands at 800400 bc and
ad750 with a minor high stand at ad150250.25
The third relevant lake is Lake Accesa. Its level steadily declined from
c. 750bc, reaching a low around ad200 before a short-lived increase.26 The
next major high stand was c. ad750. Unfortunately, the data available for
Lake Mezzano are discontinuous at this period, partly due to the human
action of draining the lake.27 There is also some discontinuous archaeologi-
cal evidence that the level of Lake Martignano (near Rome, Fig. 1) was higher
than present in 2bc and fell during the first two centuries ad to a much lower
level.28
With one lake (Fucino) showing level increases, one showing low levels
(Accesa), and some evidence that another (Martignano) dropped in level, it
seems difficult at this stage to maintain that there are multiple synchronous
movements of lake levels in Central-Southern Italy. The fact that larger
closed lakes, such as Lake Fucino, are more likely to reflect climate change
than smaller ones, such as Lake Accesa,29 is irrelevant, since multiple syn-
chronous datasets are required. Given that
22 Giraudi et al. 2011, 109110, Figure 3. The results for Accesa are given in Magny et al. 2007.
25 Magny 2004.
27 Giraudi 2004.
28 Dragoni 1998, 245; Giraudi 1998, 11.
1. the late Republican/Imperial Roman period seems to be the only ex-

ception in the late Holocene to an otherwise good correlation between
low temperatures and higher rainfall, and that
2. the only evidence for this exception is ambiguous and derived from
lake levels (the proxy least reliable, and most susceptible to local
factors such as anthropogenic influence, among those considered by
Giraudi and co-workers),
the explanation of predominantly anthropogenic activity (which is well-
attested for this period), rather than increased rainfall, for the increased
flooding in the Tiber and Po valleys seems to resolve the problem nicely.
This also puts the Central-Southern Italian climate in sync with that of West-
Central Europe, which, with the exception of Lake Ledro, shows low lake
levels at this time. It therefore seems prudent, until further data from more
lakes results in a more univocal testimony of climatic effect on lake levels in
Central-Southern Italy, to continue to regard the Roman period as another
warm and somewhat dry period within the late Holocene history of Central-
Southern Italy.
I will conclude this section by noting that there are numerous problems
in using written sources to reconstruct comparative flood histories. One is
the uneven survival of written sources from different time periods. It has
been suggested the 10th and 11th century ad Italian records are defective.30
Much depends on the floods impact on the literate elite, which may not
correlate with actual flood magnitudes.31 Large floods whose impacts are
felt largely in rural areas with low population density are less likely to be
recorded. Nevertheless, for Rome from the late Republican to the middle of
the imperial period we have some of the best written records of flood history
from pre-modern Europe.
Rural Water Use and Access in Central-Southern Italy
A major use of water here would have been for irrigation.32 Frontinus de-
scribes the careful, time-based allotment of the water of the Aqua Crabra
for irrigation of surrounding villae, including that of Cicero in earlier times,
that was disrupted by the addition of this stream to the Aqua Iulia.33
30 Camuffo and Enzi 1996, 440.

31 Dragoni 1998, 243.
32 Thomas and Wilson 1994, 157158.
33 Frontinus. Aq. 1.9, Cic. Leg. agr. 3.9.
Similar irrigation systems, also from central Italy, were depicted in CIL 6.1261
and CIL 14.3676.34 It is not clear whether they were supplied by rivers (as is
more likely) or by aqueducts. The regulation of the use of water in this way
shows that there was a high demand for water for irrigation and that there
were concerns about scarcity.35 An example of proposed aqueduct irrigation
is preserved in Ciceros Ad Quintum Fratrem 3.1.3. Cicero urges his brother
to expend 45,000 sestertii on the construction of an aqueduct in order to
irrigate 50 iugera of his land, probably near Arpinum (Fig. 1).
Another use of running water in rural Central-Southern Italy was for fish-
ing. Fresh water fishing seems to have been an important and widespread
practice in Roman Italy, although there is little trace of its presence, except
in Roman art.36 Roman law supported public access to most large lakes and
perennial rivers (flumina), which seem to have been treated as a commons.37
This included access for fishing and hauling or drawing off water.
Case Study 1: The Aqua Augusta

The largest and most complex of the aqueducts carrying water from the
Apennines to urban centres was the Aqua Augusta.38 It was Roman water
management on a larger scale than ever before:
1. Rather than being a focused supply for a single urban centre, the
Augusta supplied 8 or 9 different towns plus numerous villas through
a network of branches (Fig. 2).
2. This network had a total length of between 139 and 145 km, making the
Augusta the longest single Roman aqueduct at the time of its construc-
tion in the late Republican or Augustan period, not to be significantly
surpassed until the fifth-century aqueduct of Constantinople.
Simultaneous supply (as it seems) of so many different locations suggests
that the Augusta carried a large amount of water. Some of the towns supplied
were significant in size: Puteoli may have been the third largest town in
Italy in the early Imperial period and the third largest amphitheatre in
Italy was constructed there under the Flavians.39 In addition to the resident
34 Cf. CIL 8.4440 and Pomponius in Dig. 43.20.2.

35 Bannon 2009, 48, citing Bruun 2000, 580581.
36 Toynbee 1973, 209215; Squatriti 2002, 98101. One reference is Varro, Rerum Rusticarum
3.3.5.
37 Taylor 2000, 5565; Bannon 2009, 221.
38 For a more detailed introduction to this aqueduct see Keenan-Jones 2010a, 2010b.
39 DArms 1974, 119 n. 122; Ostrow 1985, 73.
Fig. 2 The Bay of Naples in the Roman Period showing the Aqua Augusta. Based on
data provided by ESRI Inc.
population, as a major port it had many visitors as well as many vessels

requiring water for long voyages. Several sets of public baths, at least 8
public fountains and numerous large cisterns have so far been uncovered
in the city.40 Puteolis demand for water would have been large, and it is
not clear whether Puteolis other aqueduct, the Campanian aqueduct, was
functioning during the life of the Augusta.41
The other towns supplied were also of considerable size. Estimates of
Naples population during the Roman period range up to 35,000.42 On the
poorly-watered western end of the bay were Baiae and Misenum. Baiae,
according to Strabo (5.4.7), was as large as Puteoli. Even allowing for exag-
geration, the closely-spaced yet luxurious villas of the Roman elite for which
Baiae was famous would have had a large (and conspicuous) consumption
40 Known public baths include Via Ragnisco (Republican, Gialanella 1993a, 88) and the
Tempio di Nettuno (early 2nd century ad, Sommella 1978, 32). Fountains: Sommella 1978, 26
(#12) & #39; Gialanella 1993b, 89; De Caro 1999, 651; De Caro and Gialanella 2002, 19; De Caro
2003, 593594. Cisterns of note range from those on the scale of 2 or 3 interconnected vaulted
rooms (Sommella 1978, #1, 11, 14 and 36) to the cavernous Piscina di Lusciano (Sommella 1978,
48), Cento Camerelle (De Caro 2003, 593594) and Piscina Cardito (Sommella 1978, 61).
41 Maiuri (1958, 58) and Dring (2007, 110) suggested it pre-dated the Aqua Augusta. Opus
reticulatum found in an access stair (Dubois 1907, 271) suggests the Campanian aqueduct was
operational at some point during the first centuries bc and ad.
42 Arthur 2002, 22.
of water. Misenum, the home of the western Imperial fleet, had an estimated
population of 10,000 sailors and up to 4000 civilians of citizen rank during
the early Empire, who were catered to by the enormous Piscina Mirabilis
cistern.43
The demand for the water of the Augusta would have been great and it
seems very unlikely that all these towns and villas could have been supplied
to everyones satisfaction.44 This would have been particularly the case in
summer, as we shall see below. One surviving piece of written evidence
suggests greater demand for the Augustas water than could be supplied.
A constitution promulgated in ad399 following a repair of the Augusta
stipulated heavy penalties for use of its water by any private individual.45
The Aqua Augusta represents a conscious decision, almost certainly on
the part of Augustus himself, to transfer a significant quantity of water out
of an Apennine river basin to supply urban centres on the plains around
the Bay of Naples. How large was this transfer and what were the effects
of the removal of this water at the source? Quantifying the flow rate (or
flux) of the Augusta is difficult. Like all ancient aqueducts, modelling or even
estimation of ancient flows is frustrated by the partial survival and limited
accessibility of the channel. In the case of the Augusta, it is complicated
even further by bradyseism, the slow fluctuations in ground level due to the
movement of water and magma beneath volcanoes, such as Vesuvius and
the Campi Flegrei (Fig. 3). This renders the all-important gradient of the
aqueduct very difficult to reconstruct, at least where the aqueduct crosses
the coastal plains.
One starting point in attempting to estimate the flow-rate is the flow
available to the Augusta at its source, i.e. the maximum amount of water
it could have carried. The Augusta was supplied by two main, perennial
springs, Acquaro and Pelosi, and a minor spring, Acquarolo, all located near
Santa Lucia di Serino. These springs were, and still are, fed by the Terminio
aquifer, which is composed of calcareous (calcium carbonate-rich) and
calcareous-dolomite (calcium magnesium carbonate-rich) strata.46 While
travelling from the aquifer to the surface, the water passes through overlying
Quaternary (i.e. more recent) deposits of alluvium and volcanic ash.
43 Starr 1941, 16; Keppie 1984, 103; DArms 2000, 134.

44 Keenan-Jones 2010a, 4, and 6 Fig. 2. This is implicit in the inscription commemorating
the repairs carried out under Constantine, A 1939 no. 151.
45 Codex Theodosianus 15.2.8.
46 Fiorillo and Esposito 2006, 2.
In order to reconstruct the operation of the Acquaro and Pelosi springs

during the time the Augusta functioned, we need to consider the recent
history of the springs. This has been laid out in detail by a team from
the Universit degli Studi del Sannio in two recent publications,47 upon
which this section is largely based. The discharge data for Acquaro and
Pelosi, and for another nearby spring supplied by the Terminio aquifer,
Urciuoli, are available from 1887 to the present day. Rainfall data from the
area are available from 1918 to the present day. Analysis of these data has
revealed that over that period the mean flow of Acquaro was 300800 litres
per second (L/s) and that of Pelosi was 100600 L/s, giving a combined
mean flow rate of 4001400 L/s (34,500m3/day121000 m3/day). The range is
seasonal, as the spring discharge is related to the amount of water recharging
the aquifer. The recharge is a function of the effective rainfall, or the amount
of rainfall over the aquifers catchment area less any evapotranspiration.
As discussed above, this part of Italy has had a standard Mediterranean
climate over this period, where the seasonal cycles of rainfall and evapotran-
spiration (related to temperature and humidity) are almost exactly opposite.
Rainfall is lowest and evapotranspiration highest in July and there is zero
effective rainfall from June to August. After a time delay related to water per-
colation from the surface into and through the aquifer and back to the sur-
face, this leads to minimum spring discharge in November. From September,
rainfall increases, peaking in November, and evapotranspiration decreases,
reaching a minimum in December/January. First, the Quaternary deposits
covering the aquifer are refilled, and once they have reached capacity, the
aquifer is recharged. This and the effects of snowmelt lead to maximum
spring discharge in April and May. The delay in aquifer recharge means that
individual rainstorms do not affect the flow: discharge varies as a smooth
curve throughout the year. The delay is a very useful feature for water sup-
ply, as minimum spring discharge does not coincide with minimum rainfall,
when the spring would be most needed.
It requires an effective rainfall of c. 250 mm to fill the overlying Qua-
ternary deposits and begin to recharge the aquifer. If recharge is less than
spring discharge during the wet season, then the spring discharge continues
to decrease where it would normally increase. This has only happened on
two occasions since records began, 19481949 and 20012002. The discharge
profiles for the Serino springs for these two years differ markedly from
47 Fiorillo and Esposito 2006; Fiorillo et al. 2007.

the norm. In 20012002, the drought was exacerbated by large-scale ground-

water extraction and dramatic temperature increases. Thus 19481949 is a
better indicator of a typical Serino spring drought. The combined monthly
discharge for this year decreased from around 1500 L/s before flatlining
at a minimum figure of around 1000 L/s. This was the minimum even in
20012002 and thus probably represents the maximum effect of a short-term
fluctuation such as a one- or two-year drought. Unfortunately, this figure
does not show the relative contributions of Acquaro/Pelosi and Urciuoli
for these two years. Given that Acquaro and Pelosi are far more sensitive
to fluctuations in rainfall than Urciuoli, they were probably more severely
affected during times of drought. The flow rate of Urciuoli at the beginning
of the 19481949 drought would have been around 1200 L/s. The Acquaro
and Pelosi springs (and the artesian wells near the springs) dried up com-
pletely during the 20012002 drought48 and hence it can be assumed that
the minimum flow rate of Urciuoli is 1000 L/s even in the worst droughts. It
seems that Acquaro and Pelosi were completely dry for 35 months during
the 19481949 drought.49
The return time of these Serino Spring droughts has been calculated to
be 70 years based on the data for 18872007. This return time seems to have
begun to decrease (i.e. droughts have started to become more frequent) dur-
ing the late 20th and early 21st centuries due to the large-scale groundwater
extraction and dramatic temperature increases mentioned above.
Now that some quantification of the water transferred by the Augusta
has been carried out, the effects of this transfer will be considered. None
of the towns supplied by the Augusta were in the Sabato (ancient Saba-
tus) basin, which contained the Acquaro and Pelosi springs, and there is
no archaeological evidence of supply to locations within that basin. The
Augusta functioned as a continuous transfer of water from the Sabato basin
to the Clanius basin (containing Nola, Acerrae and Atella) and the basins
draining into (or, in the case of Cumae, very near to) the Bay of Naples (con-
taining the other towns supplied) via the 6km long Forino tunnel. The long
distance transport of water from a rural source to an urban centre obviously
resulted in less water available at the source, with concomitant social and
environmental impacts.
The Acquaro, Pelosi and Urciuoli springs at Serino are far and away the
largest springs along the course of the Sabato. The other springs, those
48 Personal communication from Azienda Risorse Idriche Napoli staff.

49 Fiorillo et al. 2007, Fig. 5.
at Sorbo Serpico and the Tornola (200400 L/s and 35 L/s respectively)
have only a fraction of the flow of those at Serino.50 The flow of the Serino
springs is currently diverted to Naples for drinking water supply. If the
springs were allowed to flow into the Sabato River, however, the Acquaro
and Pelosi would account for 19% and Urciuoli 30 % of the flow of the Sabato
at Ceppaloni (Fig. 3), more than 20km downstream.51 Thus, they would
together account for almost half the median flow of the river at this point
and a greater share further upstream. The water from the springs would be
especially important to river health due to the time lag between the springs
flow cycle and that of rainfall, described above. Thus, the springs continue
to provide water even during the dry summer period when rainfall is almost
non-existent. The present diversion of the Acquaro and Pelosi often causes
the river to be dry downstream of these springs during the summer.
As discussed above, the modern data are a reasonable approximation
for the Serino springs during the period of the Augustas operation. This
requires the assumption that the vegetation cover above the aquifer was not
drastically different in these two periods, which seems reasonable. Precipi-
tation and hence river flow when the Augusta was constructed seem to have
been similar to that of the 20th Century. Both the sets of springs at Serino
were tapped, and possibly dammed, in the Roman period to supply urban
centres. It is not clear how much of the available water from these springs
was diverted to the aqueducts. Certainly, the diversion was not as compre-
hensive as the modern exploitation of the springs, which also draws water
directly from the aquifer using artesian wells near the Acquaro and Pelosi
springs. Remains of a sluice gate visible at the entrance to the aqueduct in
the 16th century suggest that the proportion of water diverted to the Augusta
in Roman times was variable.52 The summer would have been the period of
highest demand for the water of the Augusta, however, so we should not
expect that extra water was released to the river at this time.
Evidence survives for the use of the water of the Sabato for irrigation.
Upstream of Abellinum, signs of centuriation have been identified in the
fertile valley and adjacent hills.53 The effects of the removal of a significant
proportion of the flow of the river on irrigated agriculture in this area had
the potential to be dramatic, particularly during summer.
50 Regione Campania A.G.C. Sviluppo 2011, 59.

51 Calculated from the figures provided in Provincia di Avellino 2009, 3940, 48.
52 Lettieri 1560, 399.
53 Spadea 1998, 35.
The environmental effects on the river could also have been pronounced.
Rapid reductions in the flow rate of a river can severely disrupt ecosys-
tems that have developed on a longer timescale. This is particularly the case
where certain reaches of the river become disconnected, even if only tem-
porarily. Research on modern rivers around the world has shown that the
fish numbers often drop and some species become extinct in rivers under
these conditions.54
Since the Sabato River was probably a commons in the Roman period,
as discussed above, damage to fish stocks in the Sabato River would have
removed a source of income and/or dietary protein from those who could
ill afford to lose it. In this way, the use of the water of the Acquaro and Pelosi
and Urciuoli springs for urban populations was comparable to an enclosure
of an ager publicus.
Both these impacts, on water available for irrigation and on the river
itself, would have been most severe closest to the springs at Serino. Further
downstream, as the catchment area of the river increased, the impact of the
removal of Serino spring water would have been lessened.
Case Study 2: Tiber Flood Control

In the previous case study, the voices of upstream users of water from
Central-Southern Italian rivers have remained largely mute. One case where
we should be able to recover the voice of these upstream users is in the
proposal to prevent the flooding of the Tiber narrated by Tacitus in his
Annales 1.76 and 79. There is good reason to believe that Tacitus was drawing
his account directly or at least indirectly from the acta senatus, the written
records of senate business55 and thus that he was using records of the actual
speeches of the delegations from these communities.
Also that year (ad15), the rising of the Tiber, swollen by constant rain, inun-
dated the level parts of the city. Large-scale loss of life and building damage
followed the receding of its waters. Asinius Gallus therefore suggested that
the Sibylline books be consulted. Tiberius indicated his denial with a nod
of his head, drawing a veil over both human and divine affairs at the same
time. However, the task of keeping the river within its banks was committed
to Ateius Capito and L. Arruntius. (Annals 1.76)
This flood was towards the end of one of the worst, sustained flooding
episodes in the known history of the Tiber. The 1st centuries bc and ad were
54 Poff et al. 1997; Bunn and Arthington 2002; Richter et al. 2003.
55 Syme 1981, 366; Swan 1987.
second only to the Little Ice Age in terms of flood events,56 with the possible
exception of a spike at the beginning of the 2nd century bc.57 The turn of the
era even exceeded the Little Ice Age in the frequency of exceptionally large
flood events.58 The ad15 flood was the 8th recorded in 45 years.59
There were major incentives for Tiberius and the Senate to seek to rem-
edy the situation. Tacitus highlights the loss of life and damage to build-
ings. Important areas such as the Forum Romanum and the Forum Boarium
would always have been prone to flooding. Moreover, the pressure for monu-
mental space and housing in the growing city had led to an increasing num-
ber of permanent public structures in the Campus Martius, a marshy area
that was one of the first to be flooded when the Tiber rose. Public building
had begun in the southern part of the Campus Martius in the 3rd century bc
with a series of temples associated in some way with, tellingly, water.60 More
temples, a circus and porticoes followed in the late 3rd and 2nd centuries
bc. The late Republic saw the grandiose building program of Pompey, includ-
ing the Theatre of Pompey, and some activity by Caesar,61 but it was in the
Augustan period that building in the Campus really accelerated. The list of
Augustan buildings shows the importance of the area in the Augustan mon-
umental program: the Baths of Agrippa, Ustrinum Agrippae, Saepta, Pan-
theon, Mausoleum of Augustus, Porticus Argonautorum, temple of Bonus
Eventus, Laconicum Sudatorium, Porticus Octavia and Octaviae, theatres of
Marcellus and Balbus, Stagnum Agrippae, Ara Pacis and the horologium.62
In addition, the flooding of the Tiber also caused Romes sewers to over-
flow, inundating low-lying areas such as the Forum and the Campus Martius
with sewage.63 Tacitus may have been alluding to the widespread health
problems this would have caused when he stated that the loss of life fol-
lowed the receding of the waters.
Early attempts to mitigate the flooding of the Tiber included raising the
height of low areas, such as the Forum, by the addition of large quantities
of earth.64 As the monumental character of the Campus Martius grew (and
56 Bersani and Bencivenga 2001, as cited in Giraudi 2011, Fig. 7.

57 Camuffo and Enzi 1995, 1996.
58 Calenda et al. 2009, Fig. 2.
59 Aldrete 2007, 242243.
60 Torelli 2006, 92, 9699.
61 Coarelli 1997, 532590.
62 Shipley 1931; Coarelli 1997, Fig. 140.
63 Hughes 1994, 163.
64 Aldrete 2007, 177178.
also the population of the city), so did the scale of the solutions to remedy
the problem. Julius Caesar had apparently canvassed a plan to divert the
Tiber alongside the Vatican in order to enable further construction on the
Campus Martius.65 Probably in an attempt to do something about the flood-
ing problem that was adversely affecting so many of his new constructions,
Augustus further raised the level of the Forum, widened and deepened the
Tiber and, according to Suetonius, seems to have appointed curatores alvei
et riparum Tiberis (commissioners of the bed and banks of the Tiber).66 Dio,
in discussing Tiberius response to this same flood, mentions his institution
of something very similar: a permanent five-man committee chosen by lot
and tasked with ensuring the even, constant flow of the Tiber.67
Of the two men entrusted with giving advice after the flood of 15 ad,
one at least is known to have possessed some expertise. C. Ateius Capito (a
former consul and a legal expert) had been curator aquarum (water supply
commissioner) for two years by ad15,68 so he was probably the senator with
the most water-related legal and technical expertise. L. Arruntius was also
a former consul but is not known to have possessed any relevant expertise.
We next hear of them from Tacitus a few chapters later:
Next an item was raised in the Senate by Arruntius and Ateius regarding
whether, in order to moderate the floods of the Tiber, the rivers and lakes
through which it was swelled should be diverted. Embassies from towns and
colonies were heard, with the Florentines pleading that the Clanis not be
removed from its normal bed and transferred into the Arno river, so that this
would not bring disaster upon themselves. There was similarity in the matters
that the Interamnates discussed, stating that the most fertile fields of Italy
would be ruined if the river Nar, having been led away through channels (for
this was being prepared), spread over them and formed a lake. Nor were the
Reatines silent, rejecting the damming of Lake Velinus (which there pours
itself out into the Nar) after which it would necessarily overflow into the
surrounding areas. Nature, they argued, which gave to rivers their sources
and their courses, had considered human affairs most carefully: as it gave a
beginning, so it gave limits. They further argued that the cults of the allies,
which have consecrated sanctuaries, sacred groves and altars to the streams
of their native lands, should be respected, and that the Tiber itself did not
want to flow onwards with reduced glory, deprived of its tributaries. Either the
65 Cicero, Ad Atticum 13.33.4.

66 Divine Augustus 37. Cf. Aldrete 2007, 177178, 198199.
67 Dio 57.14.8.
68 Frontinus, De Aquis 102.2. One wonders whether he was related to the Ateius Capito
who told Cicero about Caesars plan to divert the Tiber (above, n. 66).
Fig. 3. The area of the proposed Tiber flood prevention works. Based on data
provided by ESRI Inc.
pleas of the colonies, or the difficulty of the works, or superstition, prevailed

such that at the motion of Piso, who had expressed the opinion that nothing
should be changed, the plan was abandoned. (Annals 1.79)
We are unfortunately given few details about the planned scheme. It in-
volved the diversion of at least two flumina (also described as amnes) and
one lacus (but possibly more) away from the Tiber to lessen its maximal flow.
Writing closer to the event than Tacitus, Pliny regarded the Clanis as one of
the two major tributaries in the upper reaches of the river.69 Figure 3 shows
the Clanis after its comparatively recent diversion into the Arno. Pliny sug-
gests that in his day the Clanis drained land as far north as Arretium into the
Tiber. The Nar was reckoned by Pliny as the second largest of the Tiber trib-
utaries below the Clanis. Hence, these were two of the major tributaries to
69 Naturalis Historia 3.5.5354.

the river, although neither was navigable (of all the Tiber tributaries, only the
Anio claimed this distinction). The damming of the Velinus had ironically
been advocated by the Interamnates in 54bc (presumably to lessen flooding
around Interamna), at which time it had also been strenuously opposed by
the Reatines.70 ad15 saw the two communities making common cause.
The effectiveness of these measures, had they been carried out, is hard
to judge. The diversion of the Clanis, eventually completed c. 1700, was
followed by a drop in the frequency and severity of Tiber flood events.71 The
major cause of this drop, however, was the end of the Little Ice Age, and
numerous damaging floods still occurred after that time.
As we saw above, flumina (larger, perennial rivers) were regarded legally
as public, with public rights of access. Unlike in the first case study, access to
public water does not form the basis of the complaints brought by commu-
nities upstream, however. The communities here faced the opposite prob-
lem: a surfeit rather than a shortage of water.
Given the seriousness of the underlying problem, the communities
needed to present the Senate with strong arguments in favour of the sta-
tus quo. Central to the Roman elites perception of their relationship to the
natural environment was an idea of continual improvement to make land
more productive and profitable.72 Stoic philosophy, an important influence
on Roman aristocratic thinkers, placed an emphasis on such agricultural
improvement, as did Cicero and Strabo, among others.73 The flooding of
prime agricultural land was the antithesis of this process. In particular, the
outflow of Lake Velinus to be dammed had actually been created by Manius
Curius Dentatus centuries before in order to reclaim agricultural land.74
The Reatines appealed to more abstract concepts that were helpful to
their cause: natura, religio and the gloria of the Tiber itself. Their argument
was that natura knows best and should not be tampered with. This line
of reasoning, clearly in tension with the belief in improvement described
above, also stretches back a long way in complex and varying world of Greco-
Roman thought.75 It can also be seen in the writings of the Roman elite, with
Cicero stating a few generations earlier that natures administration has
no component that could be criticised. The best possible outcome has been
70 Cicero, Ad Atticum 4.15.5, Pro Scauro 27.

71 Alexander 1984; Camuffo and Enzi 1996; Giraudi 2011.
72 Hughes 1994, 196; Purcell 1996.
73 Hughes 1994, 68; Pothecary 2002.
74 Cicero, Ad Atticum 4.15.5.
75 Hughes 1994, 69.
produced from the existing elements. If someone were to think it could be

better (although no one ever will), and desired to correct something, they
would either make it worse or seek to do the impossible.76
Scale and the relevance to agriculture seem to have been important to
ancient thinkers in resolving this conflicting tradition. Large-scale projects
criticised by ancient Greek and Roman writers (including Pliny the Elder)
as offending nature and related religious entities, such as bridges and canals
that created islands out of isthmuses,77 had benefits for transportation but
not agriculture. Nevertheless, because of such transportation benefits such
projects were still championed by Roman rulers such as Julius Caesar and
Nero.78 The project put forward by Capito and Arruntius was certainly on
a large scale and would have negatively impacted both agriculture and
transportation. The Tiber was vital in the trade of goods to Rome.79 With
the flow in the Tiber reduced, the navigable distance and navigable season
of the river would also have been reduced. There is evidence of concerns at
this time that occasionally the Tiber might not be navigable in summer. As
we saw above, according to Dio, Tiberius five-man committee was to ensure
also that the Tiber might not fail in summer.80
No doubt the Reatines knew what they were doing when they appealed
to religio. Sacred groves, in theory at least, were strongly protected.81 While
Livy cites a widespread abandonment of the ominous conception of floods,82
it remained a persistent theme in history and other writing. Livy himself
ascribes divine causation to 5 of the 8 floods he describes.83 Julius Obse-
quens (unsurprisingly), Aurelius Victor and Claudian (once each) as well
as the Historia Augusta (twice) also see divine forces at work in floods. Dio
Cassius, in his brief account of the ad15 flood, notes that many at the time
76 De Natura Deorum 2.34.8687: cuius quidem administratio nihil habet in se quod
reprehendi possit; ex his enim naturis quae erant quod effici optimum potuit effectum est
.
77 Pliny, Naturalis Historia 4.4.10, Hughes 1994, 5051, 69.
78 Julius Caesar, Caligula and Nero favoured a canal though the Isthmus of Corinth (Sue-
tonius, Divine Julius 44, Gaius 21, Nero 19; Pliny, Naturalis Historia 4.4.10), while canals linking
Rome to better harbours at Terracina (Suetonius, Divine Julius 58.4) and Puteoli (Tacitus,
Annals 15.42, Suetonius, Nero 31, Pliny, Naturalis Historia 14.8.61) were put forward by Cae-
sar and Nero respectively.
79 Cicero, De Republica 2.10; Pliny, Naturalis Historia, 3.54, Brunt 1980, 92, esp. n. 60.
80 Dio 57.14.8. Compare Ciceros description of the Tiber as a perennis amnis (De Republica
2.10).
81 Lucan, Pharsalia, 3.433437; Hughes 1994, 169177.
82 43.13.
83 Aldrete 2007, 221223, 296 n. 18, which is also the source for the following authors.
(like Asinius Gallus) considered it a portent,84 as Dio himself does for 10

of the 12 floods he records. According to Pliny, Tiber floods made people
more observant of religio.85 One first century ad Roman was sufficiently con-
cerned about divine intervention while constructing an aqueduct tunnel
through a mountain important to the Bona Dea that he repaired her tem-
ple after its completion, probably making good on an earlier vow.86
The political impacts of Capitos and Arruntius project were also prob-
ably slightly against it. Clearly, there was political mileage to be made as
the saviour of Rome from floods. Since the houses of the wealthy and pow-
erful were concentrated on the hills out of reach of the floodwaters,87 the
constituency was largely the masses. Augustus and Tiberius (and before
them, Caesar), as well as the Senate, had a vested interest in keeping the
numerous poor occupants of Rome satisfied as a way of maintaining pub-
lic order. In addition to the corn dole, one method was by maintaining un-
or semi-skilled employment in public building projects.88 These projects,
of course, also distributed wealth to individuals across the socio-economic
spectrum in Rome,89 developing patronage networks with political benefits.
Augustus was clearly seeking to underscore this patronage in his Res Ges-
tae, and it is telling that all the construction projects mentioned in the work
were located at Rome.90 While presumably the considerable cost of the ad 15
project would technically have been paid from funds under the control of
the Senate (the aerarium),91 it is clear even from the passage discussed above
that Tiberius could have decided the projects future, if he had wished. This
reality was surely known to those receiving the funds.
Julius Caesars proposal for flood control would have secured a grateful
Rome (if it had actually alleviated the flooding problem) but had the added
advantage of conspicuously spending the funds in the city itself. The cre-
ation of the Aqua Augusta, considered above, while not benefiting the city
of Rome, did spend its exorbitant funds at a critical place at a critical time.
The proposal of Capito and Arruntius may have alleviated flooding, but the
84 Dio 57.14.78.
85 Pliny, Naturalis Historia 3.55.
86 CIL 14.3560.
87 Aldrete 2007, 211217.
88 Brunt 1980; Wilson 2002a, 4; Wilson 2006, 230231, citing Suetonius, Vespasian 18;
Aldrete 2007, 297 n. 23.

89 Geiger 2009.
90 With the exception of the Via Flaminia (20.5), which at least started at Rome.
91 Jones 1950.
funds would not have been spent in Rome or even in some other critical
area. It should be added that this form of patronage does not seem to have
been particularly important to Tiberius. He spent little on public building
projects, especially when compared to his predecessor Augustus, preferring
to accumulate funds.92
The project would also have antagonized communities upstream. Floren-
tia, Interamna and Reate probably had patrons in the Senate to act on their
behalf. How much weight they carried we cannot know.
If the reasons behind the decision of the Senate not to proceed could not
be determined a century later by Tacitus, himself a senator, we can hardly
hope to do so now, given the chronological and cultural gulf separating us.
Of course, the decision-making may not have been rational. There is, how-
ever, at least a plausible rationale to the decision, whether or not it was
the proximate cause. A number of different modes of analysis (philosophi-
cal/religious and political) seem to have contributed to this line of reason-
ing: impacts upstream outweighed the benefits to the urban centre.
Conclusion: The Return of Modernity
Judging from these two case studies, it seems that the Roman elite around
the turn of the era was prepared to consider, and to execute, plans involving
large-scale alteration to the hydrologic landscape in order to benefit coastal
urban centres and to serve imperial aims. At the same time, they show some
mindfulness of the impacts of such projects on communities upstream, par-
ticularly municipia and coloniae, and of religious and cultural impacts. There
is also evidence of a tension between a desire for productive alterations to
the landscape and an unwillingness to upset the existing natural balance.
These considerations could sometimes be exploited by upstream commu-
nities seeking to avoid adverse effects on their way of life from proposed
changes.
The confident attitude to remaking the hydrologic landscape shown by
the Roman elite prefigured, and, at least in Central-Southern Italy, served as
a model for, the modernity of the 18th and following centuries.93 The modern
ideals of progress seemingly behind such Roman projects contrast with
92 Thornton 1986; Thornton and Thornton 1990.

93 On modernity and water management, see Allan 2006; Balali et al. 2009; Molle et al.
2009.
other, often earlier, conceptions of regress in history, particularly human

history.94 Many of the later engineers who were involved in this work made
explicit reference to Roman civilization. In the history of water management
in Central-Southern Italy, the early Roman Empire is more similar to the 18th
and 19th centuries than to any period between the two. There is, however,
some continuity in the intervening period, with much effort being expended
in maintaining more modest water supply systems.
The last recorded repair of the Aqua Augusta was shortly before ad 399
and it seems to have finally broken down even before the major eruption
of Somma-Vesuvius in the late 5th/early 6th century buried much of its
channel. A Roman aqueduct at Abella (Fig. 3), long out of commission,
was repaired around ad406 at the instigation of Paulinus of Nola.95 The
upheavals in the area during the 5th and 6th centuries probably prevented
any further repairs. The construction of a new aqueduct at Salerno during
the 9th century was unusual. Almost all aqueduct construction at this time
was repair of Roman systems and the Salerno aqueduct was probably the
last wholly new aqueduct built in Italy for several hundred years.96 At Rome,
the last ancient aqueducts had been maintained, intermittently, until the
11th century due to Papal repairs.97
For a long time after the final breakdown of the Roman aqueducts, gen-
erally only very low-level aqueducts seem to have been maintained or con-
structed. The Acquedotto della Bolla probably supplied Naples during this
time,98 but the Neapolitans accessed its water through wells, rendering it
vulnerable to contamination. At Romavecchia, where the Aqua Claudia and
Anio Novus had soared up to 32m above ground in superimposed, covered
stone channels, the Acqua Marrana Mariana, constructed to supply Rome
in 1122,99 was an uncovered canal running at ground level.
The remains of the Roman aqueducts were sometimes reused in later
ones. An early initiative was the reuse of a 1 km section of the Aqua Clau-
dia and of parts of the Aqua Alexandrina at Rome in the construction of
the Acqua Marrana Mariana.100 A further step was taken with the repair of
94 Hughes 1994, 69. These include the widespread metallic ages mythological motif found
in Hesiod, the Book of Daniel and Ovid.

95 Paulinus of Nola, Carmen 21.706750.
96 Squatriti 2002, 17.
97 Coates-Stephens 2003, 169; Hostetter et al. forthcoming.
98 Rasulo 2002; Riccio 2002.
99 Motta 1986b, 203204.
100 Ashby 1935, 222223.
the Aqua Virgo at Rome by a succession of Popes between 1453 and 1570.101
Subsequent Papal projects at Rome were the reuse of several ancient aque-
ducts in order to build the Acqua Felice (completed in 1589) and of the
Aqua Traiana to build the Acqua Paola (completed in 1613).102 Sufficient cen-
tralized authority to undertake such large-scale repairs and constructions
seems to have reappeared at Naples rather later. The Spanish viceroy of
Naples commissioned a four-year study (completed in 1560)103 of the feasibil-
ity of repairing the Aqua Augusta, but the repair was never undertaken. The
same viceroy had, however, repaired the ancient aqueduct supplying Puteoli
known as the Acquedotto Campano in 1540.104 Further repairs to this aque-
duct are attested in 1640, 1695 and 1863..105 Outside Central-Southern Italy,
the reunification of Italy seems to have given impetus to such projects, with
the reactivation, or investigations to do so, of Roman aqueducts at Bologna
and Forl (1880) and Corfinium (18881900).106
Pietrantonio Abate was an engineer in the Kingdom of Naples in the
mid-19th century and a leading proponent of the repair of the Aqua Augusta.
After more than two decades of research into the feasibility of the project,
he introduced his final report in this way:
The art under consideration (water supply), the origin of which goes back
to furthest antiquity, has two hey-days: classical Rome and the present. A
protracted period of decline and misery, which extends into the early years
of the current century, separates these two outstanding epochs.107
Abates proposal of the repair of the Aqua Augusta was also not taken up.
Instead an entirely new aqueduct, the Acquedotto di Serino (completed
in 1885)108 was built, which was symptomatic of the increasing technical
confidence and capability in Central-Southern Italy (with British assistance
in this case). The springs used for the Aqua Augusta were again transferred
out of their drainage basin to supply Naples in 1938 when they were added
to Acquedotto di Serino.109 The ancient Romans had been surpassed but
101 Ashby 1935, 170; Evans 2002, 72.

102 Cancellieri 1986; Motta 1986a.
103 Lettieri 1560.
104 Scherillo 1844, 101103.
105 De Criscio 1881, 5869.
106 Bologna and Forl (Santarelli 1882), Corfinium (De Nino 1900, 642). Evidence outside
Italy also comes from Nimes in the 19th Century (Esprandieu 1926; Fabre et al. 1991).
107 Abate 1864, 2.
108 Naples Water Works Company Ltd 1885.
109 Lettieri 1560; Sgobbo 1938.
not forgotten: a new water system inaugurated in 1923 was commemorated

by the addition of another Latin inscription in the already crowded Porta
Maggiore.110
The diversion of the Clanis (Chiana) from the Tiber into the Arno,
planned in ad15, was actually carried out piecemeal between about 1500 and
the eighteenth century. By that time, the purpose was not flood-control, but
land reclamation.111 Land reclamation on the Roman scale had to wait until
the mid-19th century and it continued under Mussolini. The partial drain-
ing of Lake Fucino in the 1st century ad, mentioned above, was carried out
again (and completed) between 1861 and 1875.112 Lake Mezzano was similarly
drained during the Roman period and then partially drained again at some
point after the 17th century.113
Thus the history of water management in Central-Southern Italy shows
the persistent influence of Roman technical achievements and thought right
through to the early 20th century.
110 Aicher 1995, 168.

111 Alexander 1984, 547.
112 Giraudi 1998, 2.
113 Giraudi 2004.
PART FIVE
FINALE
THE MEDITERRANEAN ENVIRONMENT IN ANCIENT HISTORY:
PERSPECTIVES AND PROSPECTS*
Andrew Wilson
The papers at the conference from which this volume is derived concen-
trated chiefly on energy, climate and climate change, and environmental
questions of land use, deforestation, mountains and rivers. This reflects per-
haps the current focus of historical research on the ancient environment
but we should recall that several important facets of the environment,
including the marine environment, pollution, and natural disasters such
as earthquakes and volcanic eruptions (on which more below) are absent
from such treatments. Overall, they reflect not only a current interest in
mankinds impact on the natural environment, but also renewed interest, in
the face of advances in climate science, in how the environment influences
human action, and history.1
Energy
Paolo Malanimas chapter stresses the importance of understanding the

energy budget when considering ancient economies. His questions are
important and go to the heart of debates over the environment and ancient
economic performance. Although he presents no real data for the ancient
world, his model suggests some of the constraints on pre-industrial societies
and explains why climate change would have had such a big impact on the
Roman world and other pre-industrial societies. But the approach remains
very general, with sweeping assumptions and questionable proxies. The
graph of femur length in fig. 3 shows a rise initially because most British
samples were excluded from the period ad 149 as Britain lay outside the
Roman Empire for most of this period, but were included in the ad 5099
* I am very grateful to William Harris for inviting me to be a respondent at the conference
in Rome, and to all the participants for stimulating debate and ideas. I thank especially Sturt
Manning, Mike McCormick and Kyle Harper for helpful discussion and references.
1 This is not to be confused with environmental determinism, which is the notion that
the physical environment determines human character and intelligence.

260 andrew wilson
and later periods; the rise is partly thus due to the inclusion of more north-
ern European samples from populations with a high degree of pastoralism
and thus milk and meat consumption, which typically correlate with greater
stature. Archaeological wood remains cannot straightforwardly be used to
estimate the extent of forest clearance, contra his fig. 6. Moreover, the dis-
cussion ignores cultural specifics, which for the Roman world in particular,
would include the need for fuel generated by the widespread practice of
public bathing in large, heated bathhouses. Nor can we safely assume that
fuels different from firewood represented a negligible share of the total:
archaeological evidence from North Africa, Syria and Italy shows that olive
pressing waste was used to fire kilns and even bakers ovens at various sites,
some but not all of which lay in wood-scarce regions.2 There is a method-
ological danger in borrowing figures (for fuel consumption per capita) from
other periods or places and then using these to establish Roman perfor-
mance or consumption or whateverit denies us the possibility of finding
out what was specific to or characteristic of the Roman world, which is pre-
cisely what we want to know. For example, the assumption that sailing ships
were not more numerous in the Roman Empire than in medieval and early
modern Europe is not self-evidently secure, nor were ships of 400 tons rare;
there were clearly fewer of them than 100-ton ships, but that is not the same
thing. Nor is it quite the case that water-mills and sailing ships between them
provided 100% of the mechanical energy supplied by non-biological con-
verters. Besides grain mills, water-power was also used to drive saws and
ore-crushing devices, and also water-lifting wheels for irrigation. A consid-
eration of the energy budget ought also to include the use of flowing water
for downstream river transport, which led to significant differences between
the costs of upstream and downstream transport.3 The potential for down-
stream river transport was of course regionally limited, but had a substantial
effect on shaping regional economies of major river valleys, especially the
Nile (100-ton barges were not uncommon), the Rhne and the Rhine.4 As
to the role of animals in the ancient economy, we should remember that
they were not used only for ploughing and transport, but for driving irriga-
tion machines, at least from the third century bc onwards.5 There is clearly
2 Wilson 2012b, 149150. Kilns at Leptiminus (Lamta, Tunisia): Smith 2001, 434435; kiln
at Androna (al-Andarin, Syria): Mango 2011, 108; bakeries at Pompeii: Monteix 2009, 78. Cf.
McCormick 2012, 73.
3 Russell 2009, 113114.
4 Cf. Franconi 2013.
5 Oleson 1984; 2000; Wilson 2012b; Malouta and Wilson 2013.
the mediterranean environment in ancient history 261
much more to be done in teasing out the similarities and differences in the
availability and use of the energy budget in different pre-industrial societies,
Rome included.6
Climate Reconstructions
An important study in Science published in February 2011 presented a cli-

mate reconstruction for central Europe over the last 2,400 years, based on
northern European tree-rings, including AprilMayJune precipitation lev-
els and JuneJulyAugust averaged temperatures normalised to the twen-
tieth-century averages.7 The study noted some suggestive parallels with
major historical events: the period of Roman conquest coincided with a
marked dip in summer temperatures in the late first century bc; the Roman
Empire saw, by and large, favourable rainfall and temperatures, followed
by a period of much greater climatic instability, including lower tempera-
tures, in the late third to late sixth centuries. Given Malanimas emphasis on
the significant economic effect that a rise or fall in average temperatures of
even 1C could have, it therefore becomes important to ask: to what extent,
and how, do these northern European data relate to the Mediterranean? Are
they part of a more global trend in climate, that affected the Mediterranean
world, and, in its time, the Roman Empire, as a whole?
Interestingly, increasing precipitation in the fourth to sixth centuries
shown in the northern European data seems to parallel a phase of increased
runoff and erosion which has been identified (but not closely dated) in sites
along the Tunisian coast by the Franco-Tunisian coastline survey which
identified a late Roman erosive peak which the researchers attributed to
climate rather than anthropic causes.8 But the continental European and
Mediterranean climates are different and should not be expected to track
each other precisely; and McCormicks paper in this volume reminds us
that there is not a close correlation between the northern European data
and Mediterranean written records. We need much more work on Mediter-
ranean climate datalarge-scale analyses of tree-rings, to compare with the
northern European study, and a coherent programme of pollen analyses,
especially from regions around the southern shores of the Mediterranean.
Nonetheless, a new synthetic review of multiple series of proxy climate
6 Cf. Wikander 2008; Wilson 2012b.

8 Slim et al. 2004, 251253.
262 andrew wilson
data, while identifying regional variation, underscores some broad correla-

tions between climate and general economic performance, and some spe-
cific events.9 In particular, it notes that the period from the first century bc
to early third century ad, coinciding with the peak of the Roman impe-
rial expansion and apparent prosperity (from the archaeological record),
was unusually favourable from a climatic point of view. Sturt Mannings
reworking of some of the northern European dendrochronological data (his
figs. 1213), shows rises in both precipitation and temperature which appear
to coincide with a peak in the expansion (numbers and sizes) of villas in
Britain in the fourth century; this relationship is worth exploring further.
When considering the possible inter-relationships between climate and
history, attention has until recently largely been focused largely on catastro-
phes and problems, especially the fall of the Roman Empire, and the large-
scale population migrations of late antiquity, for example the migrations
of the Huns and Avars discussed in Ed Cooks chapter in this volume.10 But
the tree-ring-based climatic reconstruction by Bntgen et al., and the wider
multi-proxy syntheses in this book by Sturt Manning, and elsewhere by
McCormick et al., remind us of the significance of the Roman warm period
in Europe, which coincided with a period of general economic prosperity in
the Roman world from the first to the mid-third century ad. Clearly climate
is not a monocausal explanation for the success of the Roman Empire at its
peak, but it does rather look like one of many contributing factors. The fall
of the Roman Empire was not of course not a single event, but more of a
process. What emerges as remarkable from the new climate evidence, if it
is applicable to the Mediterranean too, is how resilient the Roman Empire
was through the third century ad, because it now seems as though, in addi-
tion to everything else thrown at it in the course of that century, the Roman
Empire also had to contend with a marked deterioration in climate which
will have put agricultural production, and therefore the economy at large,
under additional stress.
Climate, Settlement and the Economy
The relationship of the expansion and contraction of settlement, as detect-

able by archaeological survey, to climatic change is still very unclear. Do
we see favourable climatic conditions leading to expansion, or was settle-
9 McCormick, Bntgen et al. 2012.

10 E.g., in a popular vein, Keys 1999.
ment expansion underpinned by better irrigation technology and greater

capital investment? Were past societies at the mercy of natural events, men
hopping ineffectually on the earths crust powerless in the grip of unshake-
able natural forces, or were people in some control of their environment,
and able to control their destiny by shaping it, at least to a degree? Paula
Koukis paper on the Jabal Harun is ambivalent on this point. So too the long-
running debate over Roman North Africa is still not fully settled; its origins
date back to the nineteenth century, when some argued that Roman Africas
prosperity must have been facilitated by a more favourable climate,11 while
others argued that it was achieved at the price of extensive investment in
artificial irrigation and water management systems.12 What relatively little
work has been done on the subject in recent decades has not resolved the
question: we have a few, inconclusive pollen studies,13 and even the UNESCO
Libyan Valleys Study concluded that it was unable to answer the question
definitively; the climate in the Roman period was thought to be broadly
similar to that today, although any detectable variation in ancient vegeta-
tion cover may have been due either to cultivation or to climate change,
but the environment was so marginal that potentially even a small varia-
tion could have made a big difference.14 Whether there was such a variation,
and whether it did make a difference, the project could not say.
As Malanima points out, a change of one degree centigrade in the average
temperature affects cultivation altitudes by 100 m. Warmer climates thus
mean that more upland areas can be cultivated, and cooler climates restrict
the cultivated surface. Of course, things are rather more complicated than
that because mountainsides are not totally abandoned when the climate
becomes cooler; rather their landuse would change from vines or olives to
forest, which might still be used for fuel, or timber.
What we do see, in parts of North Africa at least, is a considerable expan-
sion of the cultivated surface as marginal lands were increasingly brought
into cultivation. This is particularly apparent in the Kasserine Survey, where
the Roman period saw intensive exploitation of nearly all hillsides and
mountainsides, which were terraced to enable their systematic cultivation
with vines and olive trees.15 Pottery from the terraced slopes dated from the
first to seventh centuries ad; there was no sign of the slopes being used
11 E.g. du Coudray la Blanchre 1895, 4.

12 Gsell 1921, 5699; cf. Shaw 1981, 380, 383, 388390; Wilson 1997, 3335.
13 Rouvillois-Brigol 1985; 1986; cf. discussion in Wilson 1997, 34.
14 Barker et al. 1983, 8284; Hunt et al. 1985, 13; Hunt et al. 1986, 1416; Hunt et al. 1987, 46.
15 Hitchner 1988, 28; 1989; 1990; 1992; 1995b.
264 andrew wilson
before the Roman period, and little afterwards. Bruce Hitchner has argued
that these developments reflect the kind of incentivisation to expand the
cultivated area that we can see on imperial estates in the first and early sec-
ond centuries ad, in the lex Manciana and the lex Hadriana de rudibus agris,
which provided for rent-free periods on uncultivated land that was brought
under exploitation, while newly planted trees and vines came to maturity.16
The significance to the present debate is that this expansion was happen-
ing in a climatically favourable period. Climatic warming was not necessary
to enable cultivation of these hillsides, although it remains an open ques-
tion whether this exploitation of marginal lands was at least facilitated by
(slightly?) increased rainfall. It is possible, therefore, that the cultivation of
these more marginal lands is to be explained as a result of population pres-
sure, and perhaps economic growth, taking advantage of expanded or better
connected markets, rather than as a result of climate change.
Debates over Roman economic growth are tending to countenance sce-
narios where there was some per capita growth from the reign of Augus-
tus to the mid second century ad, at which point it was stopped by an
exogenous shock. There are dissenting voicesWalter Scheidel, for exam-
ple, argues that such growth as there was from the late Republic to the
second century ad was neither sustained or sustainable but rather a by-
product of the peace dividend following the end of the Civil Wars and
the unification of the Mediterranean under direct or indirect Roman con-
trol, and was terminated by Malthusian constraints.17 Yet the growth/shock
scenario has found considerable favour,18 and even Scheidel has accepted
that the economy received a major shock in the later second century. So far
the Antonine Plague has figured in debate as the best candidate for such a
shock, and this is accepted by a number of historians, though by no means
all; there remains in fact considerable debate over the impact of the Anto-
nine Plague.19 But Mike McCormicks contribution presents data for Nile
flood events extracted from papyrological records20 which suggest a marked
16 Hitchner 1995a.
17 Scheidel 2002; 2009, 6770.
18 E.g. Temin 2006; Jongman 2007a.
19 Severe: Duncan-Jones 1996; Scheidel 2002. Less severe: Bagnall 2000; 2002; Greenberg
2003; Bruun 2003; 2007. See now the various papers in Lo Cascio 2012, where the majority is
in favour of the severe opinion.
20 McCormick (this volume); see also McCormick, Bntgen et al. 2012; full dataset (on
which the following relies) available online as McCormick, Harper et al. 2012: Geodatabase
of Historical Evidence on Roman and Post-Roman Climate at http://darmc.harvard.edu/icb/
icb.do?keyword=k40248&pageid=icb.page496495 (accessed 16 December 2012).
change in the climate-related agricultural potential for Egypt in the mid-

second century. Although the pattern of average and somewhat better than
normal floods was much the same across the period, before ad 155 excep-
tionally good floods occurred on average one year in five, but afterwards only
once every twelve and a half years; while poor or failed floods occurred every
five years or so before the mid second century, but more frequentlyevery
three yearsafterwards. This deterioration in the pattern of Nile floods
would have depressed agricultural production, and exposed the Egyptian
populace to greater risk of famine and malnutrition, especially as the state
insisted on extracting annona grain even in poor years. McCormick chooses
ad155 as the year dividing the two different sets of flood regimes, but ad 162
or 166 would do equally well, i.e. at around the same time as the great epi-
demic. Is the deterioration of the Nile flood regime an additional, or alterna-
tive, perhaps even better, explanation, for the economic troubles observed
in Egypt than depopulation caused by the plague? Did the epidemic ravage
a population already to some extent weakened by poorer harvests? Or are
the observed discontinuities in the proxy data from Egypt which have been
used to argue for the effects of the plague in fact due not to the plague but
to the effects of poor harvests consequent on bad or failed Nile floods? In
general, and not just in the period of the epidemic, there is in fact a striking
apparent correlation with recorded incidents of flight from tax-collection
to take only the examples mentioned by Gilliam in his discussion of whether
or not the Antonine Plague had a serious effect,21 the anachoresis mentioned
in a document of ad55/5922 would follow a poor decade: three consecutive
years of poor or very poor flooding ad4547, four years of average floods
(ad4952), three years of very poor floods (ad 5355), a normal flood in
ad57 and a very poor flood in ad59, with no good recorded floods in the
years 4559. Further evidence for flight from villages in the Fayyum in ad 162
correlates with a year in which the Nile flood was low;23 the depopulation of
villages in the Mendesian nome in ad168/169 and the revolt of the Boukoloi
in ad172 or 173 can be seen to have followed a series of bad years:24 a very poor
flood in 166, an average flood in 167, three successive poor years in 168170,
and an average flood again in 171 (data for 172 and 173 themselves are not
available).25
21 Gilliam 1961, 240242.

22 P. Graux 2 = Sammelbuch IV, 7462.
23 P. Berl. Leihg. 7.
24 Mendesian nome: BGU 903.
25 Data from McCormick, Harper et al. 2012.
266 andrew wilson
The Nile flood records, which ultimately reflect rainfall in the Niles head-
water catchment area in Ethiopia and Central and East Africa, remind us
that events affecting the environment outside the Mediterranean world
may have repercussions for the Mediterranean tooeither directly through
transport of precipitation, or less directly through their effects on neigh-
bouring populations. Bntgen et al. have drawn renewed attention to the
links between climate and the migrations of late antiquity in northern
Europe.26 Recent research on the Garamantes of the Libyan Sahara has
emphasised how the incipient urbanism, oasis agriculture and long-
distance trading contacts with the Roman world were underpinned by irri-
gation from subterranean foggara systems tapping groundwater based on
foggara irrigation; and when these irrigation systems failed in late antiquity
the nexus of urban settlement and long-distance trade collapsed.27 The pre-
cise chronology remains uncertain, as does the question of to what extent
this failure was due to over-exploitation of a fossil, non-renewable, aquifer,
or to decreased local rainfall as a result of climate change;28 but the envi-
ronmental changes that led to the break-up of Garamantian power over the
Libyan Sahara seem to have had broader repercussions, and may be linked to
migrations out of oases in the northern Sahara, and the incursions of Libyan
tribes into late Roman, Vandal and Byzantine North Africa in the fourth to
sixth centuries ad.29
Ancient Understanding of the Hydrological Cycle
To explore the extent to which past societies might have been able to control
and manage their environment, it it perhaps useful to review Roman under-
standing of the hydrological cycle.30 Vitruvius (fl. c. 5026 bc) gives a descrip-
tion of the hydrological cycle based on Greek sources (he cites Theophras-
tus, Timaeus, Posidonius, Hegesias, Herodotus, Aristides and Metrodorus).

27 Mattingly 2003; Wilson and Mattingly 2003; Mattingly and Wilson 2003; Mattingly and
Wilson 2010; Wilson 2012a.
28 Cremaschi et al. 2006 argue for a dramatic drop in Saharan rainfall in the Ghat region
around ad 450; but their dendro-record is based on a single tree for large stretches of the
reconstruction, and cannot be regarded as reliable without cross-matching (Sturt Manning,
pers. comm.). Moreover, rainfall in the Ghat region may have had little connection with that
in the Wadi al-Ajal, the Garamantian heartlands, 400 kms. to the north-east.
29 Fentress and Wilson forthcoming.
30 The following section is based on Wilson 1997, 3539, which treats the sources at greater
length.
Rainfall is caused by the air acquiring moisture from the ground and from
springs, rivers, marshes and the sea, which is evaporated by the heat of the
sun. It is carried upwards in the form of clouds, and is precipitated when the
clouds strike mountains (de Architectura 8.2.14). Long-distance transfer of
water resources is described in the case of south winds, which desiccate hot
areas and redeposit the moisture elsewhere (8.2.5). Water (including melt-
water) infiltrates underground and emerges as springs (8.1.7).
Seneca (c. 5/4 bcad 65) devotes Book 3 of his Naturales Quaestiones to
a discussion of water, rivers and springs, and two partially preserved books
(4A and 4B) to the Nile and to hail and snow. A lost work of Theophras-
tus seems to have been one of his main sources. He discusses a number of
observed phenomena, but his explanations for them are influenced by Stoic
notions of cosmic flux and are often less accurate than those of some of his
predecessors. He plays down the role of groundwater, believing that rain
never penetrates the earth to a depth of more than ten feet, and most of it
is carried off to the sea by rivers (Nat. 3.7.1). He notes that in arid regions
wells may be driven 200 or 300 feet before encountering water, and calls
this aqua viva, living water, as it cannot have come from rainwater infil-
tration (3.7.34). Rivers are formed not by groundwater but by condensa-
tion of air in hollows within the earth (3.9.23); rain does not cause rivers,
but only makes them flow faster (3.11.6). He does not give an account of
the hydrologic cycle, and states that the sea has its own springs (3.14.3).
However, some observations are more accurate; deforestation causes the
emergence of springs because water is no longer consumed by vegetation
(3.11.35).
Pliny (ad 23/2479) indicates an awareness of seasonal variations in
groundwater levels; he says that wells generally run dry about the rising of
Arcturus (17th September), and that springs vary their discharge according
to season (Nat.Hist. 31.xxviii.5051). Earthquakes may cause springs to dry
up or appear (31.xxx.54). Springs can also arise as a result of deforestation,
and tree cover inhibits erosive streams (31.xxx.53). Pausanias (fl. c. ad 160)
displays awareness of the effects of agriculture on increasing erosion and the
concomitant downstream deposition of alluvial fans (8.24.1011). Certain
ancient writers therefore display a basic understanding of the hydrologi-
cal cycle, the formation and behaviour of groundwater, and of the effects
of agriculture and deforestation on erosion. Knowledge was available to
tap aquifers, locate springs, use dams to recharge groundwater reserves,
and exploit mineral or thermal properties of water. The principle that sub-
terranean reservoirs could store winter rains for discharge in the summer
months was at least theoretically understood, even if Aristotle had refuted it
268 andrew wilson
as the explanation for the origin of rivers (Mete. 349b315). There was clearly
much interest in different types of water, their properties and their effects
on health.
We are familiar with the fact that Near Eastern, Greek and Roman hydrau-
lic engineering was capable of remedying local deficiencies in water avail-
ability by the transfer of water resources from elsewhere via aqueducts,
which could in the Roman and Byzantine periods achieve extraordinary
lengths98km for the second-century aqueduct feeding Carthage from the
springs at Zaghouan, later increased to 132km by the addition of tributaries
in the Severan period; and the aqueduct that supplied Constantinople, com-
pleted in ad373, had a main channel of 227km with a tributary of 41 km.31
Duncan Keenan-Jones discussion of the Serino aqueduct in Campania in
this volume emphasises the capability of Roman engineers not only to effect
inter-basin transfers of water (something common in fact to many Roman
aqueducts), but also to think on a regional scale in terms of supplying several
cities. The highly urbanised region of the Bay of Naples demanded extraordi-
nary solutions to the problem of supplying such a large and nucleated urban
population with water.
Case Study: Controlling Tiber Floods

A further illustration of the developing ancient understanding of total river
basin management, which enables us to go further than the information
given by the ancient authors mentioned above, is provided by the history of
Roman attempts at controlling the flooding of the Tiber. Duncan Keenan-
Jones chapter in this volume discusses two of the big proposed flood relief
schemes, under Caesar and Tiberius; here I want to consider them in the
light of the overall chronological development of attempts at Tiber flood
control. In the last thousand years the Tiber has flooded to levels of 1013
masl on average every year, and above 13 masl on average every two years.32
As Keenan-Jones notes, the wetter conditions suggested by climate recon-
structions for between c. 100bc and ad200 would have exacerbated this pat-
tern in the Roman period. The lowest-lying areas in antiquity, most exposed
to flooding, were the Campus Martius, the Velabrum, forum, the Circus
31 Generally: Wilson 2008a. Constantinople: Crow et al. 2008; Crow 2012, 40.
32 Le Gall 1953, 6265; Lugli 1953, 6169; Ammerman 1990, 637; Belati 1999; Aldrete 2007.
Maximus and Emporium areas below the Aventine, and Trastevere. The
Pons Sulpicius had to be rebuilt on numerous occasions following damage
by flooding.33
From the period of the Etruscan kings through to the High Empire, pro-
gressively more ambitious and effective means of flood control were consid-
ered, some of which were not carried out for political reasons. Coring work
in the forum and the Velabrum by Albert Ammerman has shown that in
the sixth century bc the ground level in these areas was artificially raised by
between 1 and 4m by dumps containing gravel, tuff fragments and anthropic
material, as a landfill scheme intended to bring the ground surface above the
level of many floods.34 The ground surface would have needed to be raised
some 2m in the centre of the valley, to bring it to around the 9 m contour
level that would put it above all but the worst floods. Ammerman estimates
that at least c. 10,000m3 of fill would have been needed; possibly twice that
amount. Such an exercise implies considerable manpower resources, and
may have occurred over a period of several years; Ammerman notes parallels
with other major landscaping works under the Etruscan kings. If the con-
struction of the Cloaca Maxima to drain the forum area went hand in hand
with raising of its surface above the level of the Tiber floods, the emphasis
in the ancient writers on the arduous nature of the corve work becomes
easier to understand.35
These early solutions, involving considerable landscape transformations
of the forum valley, made possible the creation of a public space in what was
to become the heart of Rome, and laid the foundations for the later urban
development of the city. Impressive and labour-intensive as they were, their
effect was to raise certain areas only above the levels of many, but not all
floods; the problem of Tiber flooding had been alleviated, but not solved.
As Rome expanded in the mid and late Republic, Tiber floods began
to affect a larger inhabited area. The development of the Campus Mar-
tius, started by Pompey and Caesar and further continued under Augustus,
compounded the problem.36 There were six severe floods in the reign of
Augustus, and a further, disastrous, flood in ad 15, the year after his death.
Tiberius, instead of consulting the Sibylline books as advised, appointed a
33 Tac. Ann. I.86 (23bc); SHA Antoninus Pius 8; Dio Cassius 1.3.20; 27.58 and 50.8; Desnier
1998, 515, 520.

34 Ammerman 1990; 1998.
35 Ammerman 1990, esp. 636645.
36 Belati 1999, 13.
270 andrew wilson
permanent commission of five senators, the curatores alvei Tiberis et ripa-

rum, who had responsibility for managing the banks of the Tibertheir
duties included ensuring the dredging of the river channel and canalising
its banks (marked by cippi). These measures contained some of the lesser
floods, but still did not remove the problem; floods continued almost annu-
ally. Part of a bridge pier belong to the ancient Pons Aurelius, found below
the Ponte Sisto in the nineteenth century, seems to have been marked with
a Tiber flood gauge; it was graduated in Roman feet in reverse order, appar-
ently therefore showing the height remaining until the river reached a dan-
ger or warning level.37
Up to this point, then, we have seen the development of attempts to con-
tain flooding within the urban zone. Fundamentally different were efforts to
tackle the cause of the problem, which we see for the first time in a scheme
considered by Caesar, but never put into effect. This was the construction
of a relief channel leaving the Tiber near the Mulvian bridge and running
around the back of the Janiculum hill, to feed back into the Tiber below the
city.38 Caesars assassination prevented the scheme being carried out.
Serious flooding recurred almost annually, and throughout the course of
the first century ad increasingly ambitious attempts started to deal with the
root of the problem, rather than simply to contain the symptoms. After a
major flood under Tiberius, a scheme was proposed to divert the Chiana,
one of the Tibers tributaries on the right bank upstream from Rome, into
the Arno to reduce the amount of water flowing through Rome, and that
the waters of another tributary, the Nar, should be dispersed throughout
its floodplain, and the artificial channel feeding the Nar from the Veline
lakes be dammed up. This provoked a senatorial debate and the scheme
was successfully opposed by the local communities (Florence, Interamna,
and Reate) who would have been affected by this move, using a variety of
arguments both practical and religious.39
At the downstream end of the Tiber Valley, Claudius works on the new
harbour at Portus included the digging of relief channels to link the Tiber
to the sea, above the sharp bend by Ostia which acted as a bottleneck and
caused the waters to back up, affecting the river as far upstream as Rome.
Claudius inscription makes explicit the link between the works at Portus
and the risk of flooding at Rome (ILS 207 = CIL XIV.85):
37 Marchetti 1892.
38 Cicero, Att. 13.33.4; Aldrete 2007, 182184.
39 Tacitus, Annals I.79. Aldrete 2007, 185188; Campbell 2012, 118119; Keenan-Jones (this
volume).
Ti. Claudius Drusi f. Caesar fossis ductis a Tiberi operis portu caussa emis-
sisque in mare urbem inundationis periculo liberavit
Tiberius Claudius Caesar, son of Drusus, freed the city from the danger of
flooding by leading canals from the Tiber into the sea in connection with the
building of the harbour.
This scheme shows that Claudius engineers evidently understood the sys-
temic nature of flood control, and that it might be necessary to effect works
many miles distant from the point at which one desired to control the floods.
An additional effect of Claudius works must also have been to reduce the
extent to which Ostia was exposed to regular and severe flooding. Claudius
boast that he had liberated the city from the danger of flooding was cruelly
shown to be premature by the severe flood of ad. 69, recorded by Tacitus
(Hist. I, 86.2), and also by the inscription commemorating Trajanic works
(below).
There was further raising of the ground level in the Campus Martius in
the later first century ad: under Domitian the pavement of Augustus sundial
was repaved at a level 1.5m higher than the original. Indeed, it may be that
the flooding of the Tiber in this area caused the obelisk to settle out of true,
explaining why Pliny says the sundial was inaccurate only 50 years after
its construction. The Trajanic or Hadrianic pavement of the Pantheon lies
1.85m above that of its Agrippan predecessor.40
Trajan constructed further exit channels from the Tiber to the sea near
Ostia and Portus; if the reconstruction of some of the missing text is correct,
Claudius measures had not been entirely effective (ILS 5797):
[Imp. Caesar divi] | Ne[rvae fil. Nerva] | Tra[ianus Aug. Germ.] | Dac[icus trib.
pot.] | im[p. , cos. p. p.] | fossam [fecit | q[ua inun[dationes Tiberis |
a]dsidue u[rbem vexantes | rivo] peren[ni instituto arcerentur]
Imperator Caesar Nerva Traianus Augustus Germanicus, son of the deified
Nerva, made a canal by which the floods of the Tiber which continually
troubled the city were kept away with the construction of a permanent chan-
nel.
Post-Trajanic measures seem to have been limited largely to repairs to the
Tiber banks and dredging of its channel, for example under Aurelian (SHA
Aur. 47, 13).
The larger flood relief schemes of the first and early second centuries
ad imply a recognition that effective solutions to the problem of the Tiber
40 Belati 1999, 16.

272 andrew wilson
floods required a landscape-wide solution which addressed the whole

catchment area of the river valley, and its estuary. However, political con-
siderations prevented the most ambitious schemes (Caesars relief channel
and the Chiana diversion), which were also the most likely to have provided
an effective longer-term solution, from being put into action.
The Environment of Production
The differential costs of land, riverine and maritime transport influenced

the geography of production in a manner clearly seen along coastlines and
major river valleys. In the ancient Mediterranean the preferred container for
long-distance transport was for centuries the amphora, although it gradually
gave ground to the barrel. Ceramic amphorae were heavy, however, and their
production was preferentially located along navigable rivers and near the
coast. In Baetica, a major olive-oil producing region, the kilns which made
the Dressel 20 amphorae in which the oil was shipped were located along the
Guadalquivir and its main tributary, the Genil, and the oil was transported
in skins from the farms to the kiln sites for bottling and riverine transport in
amphorae down to Hispalis (Seville) and then loaded onto maritime ships
for export to Rome and other distant markets. In the other major olive
oil exporting regions, modern Tunisia and north-west Libya (Tripolitania),
amphora production was chiefly concentrated at coastal sites which acted
as the bottling centres for the produce of the hinterland.
Amphora production, and ceramic production in general, was also
strongly influenced by the availability of fuel, especially important where
production had been scaled up for large-scale long-distance export. The
otherwise puzzling location of the large-scale southern Gaulish samian pro-
duction centre at La Graufesenque may be explicable to a large degree by the
ready available of firewood (the clay deposits were some kilometres distant).
In Tunisia, where firewood was scarce, and the waste from olive pressings
was used as fuel, some co-location of olive oil production and amphora
and cookware manufacture, often at coastal sites, is observable.41 Coastal
sites also had the advantage that firewood could be imported by boat, as
depicted in a mosaic from Sousse.42 Confirmation of the importance of a
41 Olive pressings as fuel at Leptiminus, Tunisia: Smith 2001, 434435. Co-location of
amphora and cookware production: Leitch 2011.

42 du Coudray la Blanchre and Gauckler 1897; Meiggs 1982, 10, Pl. I.6; Meiggs 1982,
529530; Wilson 2012b, 149.

long-distance maritime trade in wood fuel comes from a second- or third-

century papyrus from Tebtunis, referring to the import of Italian firewood.43
William Harris contribution to this volume discusses the extent of pos-
sible deforestation in the Greek and Roman periods, suggesting hesitantly
that such deforestation as there was in classical times was probably more
in the nature of land clearance, and most visible in Attica; deforestation
probably increased in the Hellenistic and Roman periods but the effects of
demand were to some extent offset by improved woodland management
and increased long-distance maritime trade that mitigated the impact on
their immediate hinterlands of the fuel demand created by large cities. His
discussion emphasises the scale of ancient fuel consumption and the trade
both in firewood and in construction timbers.44 For firewood, it is not fully
clear whether fuel wood was imported to certain regions because it had to
be (i.e. was there serious deforestation in North Africa, Egypt, and the vicin-
ity of Rome?), or because it could be, cheaply, for example as return cargoes.
Robyn Veals chapter on fuels stresses the use of charcoal as well as firewood,
and shows the potential for ancient fuel studies, but we should also con-
sider the possible uses of non-wood fuelsas mentioned above, Pompeii,
North Africa and Syria have all yielded evidence suggesting that the use of
olive pressing waste as a fuel in commercial ovens and kilns was common;
and the use of animal dung as a fuel for purposes not requiring very high
temperatures, such as domestic hearths and bread ovens, has a long history
in Asia Minor and the Near East, and has been identified in Numidian lev-
els at Althiburos (Tunisia).45 The scale of the trade in construction timber
will evidently need to be taken into account when we get to the stage of
doing the tree-ring analysis for the Roman Mediterranean. What mixing of
the dendro-record might it create? Can we use the tree-ring research to iden-
tify the sources of timber?
Future Directions
Our understanding of the scale and effects of ancient pollution still has a
long way to go. Exemplary regional studies such as that in the Wadi Fay-
nan, Jordan, have shown the local effects of Roman and Byzantine copper
43 P. Tebt. II.686 (cited by Harris, this volume).

44 Harris (this volume); cf. Wilson 2012b: 139140.
45 W. Brown 1820, 297300. Althiburos (dung in an oven of the 4th2nd century bc):
Portillo and Albert 2011, 3232.

274 andrew wilson
smelting and mining activities there, and how the soil still remains mas-
sively polluted, with long-term effects on animals and the human popula-
tion of the region even today.46 The identification of anthropogenic pollution
in Greenland ice cores in the 1990s, principally from copper and lead/silver
working, has attracted much attention, but the original studies used rela-
tively few samples and produce graphs with very spiky peaks and troughs.47
Much work on ice-cores in both the Arctic and Antarctica has been done
since the original studies, and we can soon expect to have a high-resolution
record of atmospheric pollution over the last 3,000 years, with annual or sub-
annual resolution and an absolute dating precision of 2 years, that could
enable further investigation of levels of ancient mining and metallurgical
activity, and their environmental impact.
Most of the papers in this volume concentrate on processes and long-
term trends, rather than catastrophic events. Volcanic eruptions or their
effects are mentioned in passing in several papers (McCormick, Keenan-
Jones), but earthquakes are not mentioned at all. These perhaps deserve
greater consideration in environmental historynot to overemphasise
catastrophe in itself, but rather to assess the relative effects of sudden and
long-term change. There is much work to be done here, ranging from assem-
bling more comprehensive lists of ancient earthquakes (including not just
those mentioned in ancient sources but also those inferred from archaeo-
logical excavations) and volcanic eruptions (again, including those detected
principally through tephra or sulphate deposits in ice cores), to the consid-
eration of large socio-economic effects.48 To what extent were ancient states
able to respond to natural disasters such as earthquakes or volcanic erup-
tions, to afford disaster relief, or rebuilding, and how far did they try to do
so? The clustering of earthquakes in the fourthsixth centuries ad known to
geologists as the Early Byzantine Tectonic Paroxysm, deserves greater atten-
tion in considerations of Byzantine history.49 Does a particular clustering of
earthquakes within this period and around the big one of 21 July ad 365 with
its epicentre off western Crete, help to explain the notable peak in rebuild-
46 Barker et al. 2007.

47 Original studies: Hong, Candelone, Patterson and Boutron 1994; Hong, Candelone, and
Boutron 1994; Hong, Candelone, Patterson and Boutron 1996; Hong, Candelone, Soutif, and
Boutron 1996; Rosman et al. 1997. Relation to ancient economic history: McCormick 2001, 53;
Wilson 2002a, 2529; 2007; 2009; de Callata 2005; Jongman 2007a, 188189; Scheidel 2009.
48 Lists of earthquakes: Guidoboni 1989; Guidoboni et al. 1994; Ambraseys et al. 2005; cf.
Ambraseys 1994; 2006; Nur 2010. Much archaeological evidence remains to be assessed and
added (e.g. a probable earthquake at Euesperides, Benghazi, 262250bc: Wilson 2003).
49 Stiros 2001.
ing inscriptions between ad360 and 375, especially in North Africa?50

The ice core record, which has trapped sulphates emitted by volcanic
eruptions, can also help with reconstructing links between ancient erup-
tions and climate. A recent study of Scandinavian settlement and tree rings
argued that the Norse myth of the Fimbulwinter, the harsh winter that pre-
cedes the end of the world, with three years without summer, and the break-
down of social order, with successive wars and brothers killing brothers,
reflects the dust veil event of ad536, observable in the dendrochronologi-
cal record of Scandinavia and in the abandonment of numerous settlements
in the mid sixth century.51 The dust veil event had a global reach, affecting
the Mediterranean (Procopius speaks of the sun shining only weakly for 18
months) and China.52 The Greenland ice core record shows extremely high
sulphate levels for ad5335342, which suggests that the dust veil event of
536 was caused by a volcano,53 possibly the Tierra Blanca Joven (TBJ) event
of the Ilopango volcano in El Salvador.54 Through a combination of ice core
studies, volcanology, archaeology, and the analysis of Norse poetry, we are
now able to draw a link between the mythological motif of the Fimbulwinter
and, possibly, a volcano in Latin America.
And lastly, what of the marine environment? The effects of eustatic and
relative sea-level change have been identified along the coasts of Italy and
Tunisia,55 and at other localised pointschiefly harbours and major
ports,56 but systematic study of the Mediterranean coastline, and espe-
cially Algeria and Libya where relatively little work has been done, is needed
to assess the regional impact of sea-level change on coastal settlement
and economies, including the effects on the viability of ports. Conversely,
although a certain amount of attention has been paid by historians and
archaeologists to human impact on the terrestrial environment, what of the
impact on the marine environment and fish stocks. Did the large-scale fish-
ing practised in the Roman period have any measurable impact on the size
of particular fish species, or on fish stocks or populations?57
50 Cf. Wilson 2011, 163165.

51 Grslund and Price 2012.
52 Gunn 2000.
53 Larsen et al. 2008.
54 Dull et al. 2001; Dull et al. 2010.
55 Schmiedt 1972; Slim et al. 2004.
56 Blackman 1973.
57 Cf. Marzano 2013 on the scale of Roman fishing.
276 andrew wilson
The highest resolution reconstructions of ancient climate currently avail-

able pertain to northern Europe where the dendrochronological record
is best, but the situation for the Mediterranean is improving, with multi-
proxy reconstructions now being produced, albeit often reliant on lower-
resolution data. There is a need for more data and higher-resolution studies
for the Mediterranean, North Africa and the Near East if we are properly
to address the question of how far climate affected history and settlement,
and explore to what extent were ancient populations able to combat adverse
climate change with adaptive agricultural technologies and water manage-
ment systems, such as the runoff farming in the pre-desert of Tripolitania
or in the Negev. Nevertheless, exciting advances have been made in recent
years at an accelerating rate in the fields of ancient climate reconstruction,
dendrochronology, and the studies of environmental pollution trapped in
ice cores, and the increasing interplay between these kinds of science and
historical research is opening up new possibilities for profoundly deepen-
ing, and perhaps even transforming, our understanding of the relationship
between history and the environment.
BIBLIOGRAPHY
Abate, F. 1864. Studi sullacquedotto Claudio e progetto per fornire di acqua potabile
la citta di Napoli. Naples.
Abudanh, F. 2006. Settlement Patterns and Military Organisation in the Region of
Udhruh (southern Jordan) in the Roman and Byzantine Periods III. Unpublished
PhD Dissertation, University of Newcastle upon Tyne.
Aicher, P.J. 1995. Guide to the Aqueducts of Ancient Rome. Wauconda, IL.
Akkemik, ., and A. Aras. 2005. Reconstruction (16891994ad) of AprilAugust
precipitation in the southern part of central Turkey. International Journal of
Climatology 25: 537548.
Akkemik, ., N. Dagdeviren, and A. Aras. 2005. A preliminary reconstruction (A.D.
16352000) of spring precipitation using oak tree rings in the western Black Sea
region of Turkey. International Journal of Biometeorology 49: 297302.
Aldrete, G.S. 2007. Floods of the Tiber in ancient Rome. Baltimore.
Alexander, D. 1984. The Reclamation of Val-di-Chiana (Tuscany). Annals of the
Association of American Geographers 74(4): 527550.
Alexandre, P. 1987. Le climat en Europe au Moyen ge: contribution lhistoire des
variations climatiques de 1000 1425, daprs les sources narratives de lEurope
occidentale. Paris.
Allan, J.A. 2006. Millennial water management paradigms: making IWRM work.
http://www.mafhoum.com/press/53aE1.htm.
Allen, J.R.M., W A. Watts, E. McGee, and B. Huntley. 2002. Holocene environmen-
tal variabilitythe record from Lago Grande di Monticchio, Italy. Quaternary
International 88: 6980.
Allevato, E., E. Russo Ermolli, G. Boetto, and G. Di Pasquale. 2010. Pollen-wood
analysis at the Neapolis harbour site (1st3rd century ad, southern Italy) and its
archaeobotanical implications. Journal of Archaeological Science 37: 23652375.
Alley, R.B. 2000. The Younger Dryas cold interval as viewed from central Greenland.
Quaternary Science Reviews 19: 213226.
Alley, R.B., and A.M. gstsdttir. 2005. The 8k event: Cause and consequences
of a major Holocene abrupt climate change. Quaternary Science Reviews 24:
11231149.
Amato, L., V. Carsana, A. Cinque, V. Di Donato, D. Giampaola, C. Guastaferro, G.
Irollo, C. Morhange, P. Romano, M.R. Ruello, E. Russo Ermolli. 2009. Ricostru-
zioni geoarcheologiche e morfoevolutive nel territorio di Napoli (Italia): levolu-
zione tardo pleistocenica-olocenica e le linee di riva di epoca storica. Mditer-
rane 112:2331.
Amato, L., G. Bisogno, L. Cicala, A. Cinque, P. Romano, M.R. Ruello, E. Russo Ermolli.
2010. Palaeo-environmental changes in the archaeological settlement of Elea-
Velia: climatic and/or human impact signatures? In Scienze naturali e archeolo-
gia. Il paesaggio antico: interazione uomo/ambiente ed eventi catastrofici, edited
by A. Ciarallo and M.R. Senatore, 1316. Rome.
278 bibliography
Ambraseys, N.N. 1994. Material for the Investigation of the Seismicity of Libya.
LibStud 25: 722.
2006. Earthquakes and archaeology. Journal of Archaeological Science 33.7:
10081016.
Ambraseys, N.N., C.P. Melville, and R.D. Adams. 2005. The Seismicity of Egypt, Arabia
and the Red Sea: A Historical Review. Cambridge.
Ammerman, A.J. 1990. On the origins of the Forum Romanum. AJA 94.4: 627
645.
1998. Environmental archaeology in the Velabrum, Rome: interim report.
JRA 11: 213223.
Amr, K., A. al-Momani, S. Farajat, and H. Falahat. 1998. Archaeological survey
of the Wadi Musa Water Supply and Wastewater Project Area. Annual of the
Department of Antiquities of Jordan 42: 503548.
Amr, K., and A. al-Momani. 2001. Preliminary report on the archaeological com-
ponent of the Wadi Musa water supply and wastewater project (19982000).
Annual of the Department of Antiquities of Jordan 45: 253285.
Andreadis, K.M., E.A. Clark, A.W. Wood, A.F. Hamlet, and D.P. Lettenmaier. 2005.
Twentieth-century drought in the conterminous United States. Journal of
Hydrometeorology 6: 9851001.
Arjava, A. 2005. The mystery cloud of 536CE in the Mediterranean sources. Dum-
barton Oaks Papers 59: 7393.
Arnaud-Fassetta, G., N. Carcaud, C. Castanet, and P.-G. Salvador. 2010. Fluviatile
palaeoenvironments in archaeological context: Geographical position, method-
ological approach and global changeHydrological risk issues. Quaternary
International 216: 93117.
Arthur, P. 2002. Naples, from Roman town to city-state: an archaeological perspective.
Rome/Lecce.
Ashby, T. 1935. The aqueducts of ancient Rome. Oxford.
Asouti, E. 2004a. Charcoal Analysisa Short History of the Discipline. London.
Available from http://www.ucl.ac.uk/~tcrneas/History of charcoal analysis.htm.
Asouti, E. 2004b. Factors Affecting the Formation of an Archaeological Wood Char-
coal Assemblage. London. http://www.ucl.ac.uk/archaeology/staff/profiles/
asouti.htm.
Atherden, M. 2000. Human Impact on the Vegetation of Southern Greece and
Problems of Palynological Interpretation: a Case Study from Crete. In Landscape
and Land Use in Postglacial Greece, edited by P. Halstead and C. Frederick, 6278.
Sheffield.
Bach, A. 19521956. Deutsche Namenkunde. Heidelberg.
Bagnall, R.S. 2000. P.Oxy. 4527 and the Antonine plague in Egypt: death or flight?
JRA 13: 288292.
2002. The effects of plague: model and evidence. JRA 15: 114120.
Bagnall, R.S., A. Cameron, S.R. Schwartz, and K.A. Worp. 1987. Consuls of the later
Roman Empire. Atlanta.
Baillie, M.G.L. 1994. Dendrochronology raises questions about the nature of the
AD536 dust-veil event. The Holocene 4: 212217.
2008. Proposed re-dating of the European ice core chronology by seven years
prior to the 7th century AD. Geophysical Research Letters 35: L15813.
bibliography 279
2010. Volcanoes, ice-cores and tree-rings: one story or two? Antiquity 84:
202215.
Bakels, C. 1999. Archaeobotanical investigations in the Aisne valley, northern
France, from the Neolithic up to the early Middle Ages. Vegetation History and
Archaeobotany 8: 7177.
Balali, M.R., J. Keulartz, and M. Korthals. 2009. Reflexive water management in arid
regions: the case of Iran. Environmental Values 18(1): 91112.
Bang, P.F. 2009. The Ancient Economy and New Institutional Economics. JRS 99:
194206.
Bannon, C.J. 2009. Gardens and Neighbors: Private Water Rights in Roman Italy. Ann
Arbor, MI.
Bar-Matthews, M., and A. Ayalon. 2004. Speleothems as paleoclimate indicators,
a case study from Soreq Cave located in the Eastern Mediterranean region,
Israel. In Past Climate Variability through Europe and Africa, Vol. 6, edited by
R.W. Battarbee, F. Gasse, and C.E. Stickley, 363391. Dordrecht.
Bar-Matthews, M., and A. Ayalon. 2011. Mid-Holocene climate variations revealed
by high-resolution speleothem records from Soreq Cave, Israel and their corre-
lation with cultural changes. The Holocene 21: 163171.
Bar-Matthews, M., A. Ayalon, and A. Kaufman. 1997. Late Quaternary paleoclimate
in the eastern Mediterranean region from stable isotope analysis of speleothems
at Soreq Cave, Israel. Quaternary Research 47: 155168.
Bar-Matthews, M., A. Ayalon, and A. Kaufman. 1998. Middle to Late Holocene
(6500 years period) paleoclimate in the Eastern Mediterranean region from
stable isotopic composition of speleothems from Soreq Cave, Israel. In Water,
Environment and Society in times of Climate Change, edited by A.S. Issar and
N. Brown, 203214. Dordrecht.
Bar-Matthews, M., A. Ayalon, M. Gilmour, A. Matthews, and C.J. Hawkesworth.
2003. Sea-land oxygen isotopic relationships from planktonic foraminifera and
speleothems in the Eastern Mediterranean region and their implication for
paleorainfall during interglacial intervals. Geochimica et Cosmochimica Acta 67:
31813199.
Bar-Matthews, M., A. Ayalon, A. Kaufman, and G.J. Wasserburg. 1999. The eastern
Mediterranean paleoclimate as a reflection of regional events: Soreq cave, Israel.
Earth and Planetary Science Letters 166: 8595.
Bard, E., G.M. Raisbeck, F. Yiou, and J. Jouzel. 1997. Solar modulation of cosmogenic
nuclide production over the last millennium: comparison between 14C and 10Be
records. Earth and Planetary Science Letters 150: 453462.
Barker, G. 2002. A Tale of Two Deserts: Contrasting Desertification Histories on
Romes Desert Frontiers. World Archaeology 33: 488507.
Barker, G.W.W., D.D. Gilbertson, C.M. Griffin, P.P. Hayes, and D.A. Jones. 1983. The
UNESCO Libyan Valleys Survey V: sedimentological properties of Holocene wadi
floor and plateau deposits in Tripolitania, north-west Libya. LibStud 14: 69
85.
Barker, G.W.W., D.D. Gilbertson, and D.J. Mattingly. 2007. Archaeology and desertifi-
cation. The Wadi Faynan Landscape Survey, Southern Jordan. Oxford.
Beckh, H. 1895. Geoponica sive Cassiani Bassi scholastici De re rustica eclogae. Leip-
zig.
280 bibliography
Beer, J. 2000. Long-term indirect indices of solar variability. Space Science Reviews
94: 5366.
Beer, J., A. Blinov, G. Bonani, R.C. Finkel, H.J. Hofmann, B. Lehmann, H. Oeschger,
A. Sigg, J. Schwander, T. Staffelbach, B. Stauffer, M. Suter, and W. Wlfli. 1990. Use
of 10Be in polar ice to trace the 11-year cycle of solar activity. Nature 347: 164
166.
Beer, J., U. Siegenthaler, G. Bonani, R.C. Finkel, H. Oescheger, M. Suter, and W. Wlfli.
1988. Information on past solar activity and geomagnetism from 10Be in the
Camp Century ice core. Nature 331: 675679.
Beer, J., S. Tobias, and N. Weiss. 1998. An Active Sun Throughout the Maunder
Minimum. Solar Physics 181: 237249.
Beer, J., M. Vonmoos, and R. Muscheler. 2006. Solar variability over the past several
millennia. Space Science Reviews 125: 6779.
Behre, K.-E. 1992. The History of Rye Cultivation in Europe. Vegetation History and
Belati, M. 1999. Le inondazioni del Tevere nellantichit. Forma Urbis 4.7/8: 1217.
Bellotti, P., C. Caputo, L. Davoli, S. Evangelista, E. Garzanti, F. Pugliese, and P. Valeri.
2004. Morpho-sedimentary characteristics and Holocene evolution of the emer-
gent part of the Ombrone River delta (southern Tuscany). Geomorphology 61(1
2): 7190.
Beltrn Lloris, F. 2006. An Irrigation Decree from Roman Spain: the Lex Rivi Hibe-
riensis. JRS 96: 147197.
Berggren, A.-M., J. Beer, G. Possnert, A. Aldahan, P. Kubik, M. Christl, S.J. Johnsen,
J. Abreu, and B.M. Vinther. 2009. A 600-year annual 10Be record from the NGRIP
ice core, Greenland. Geophysical Research Letters 36: L11801.
Bernand, E. 1993. Une figure de la papyrologie: Danielle Bonneau. ZPE 98: 97102.
Bersani, P., and M. Bencivenga. 2001. Le piene del Tevere a Roma dal V secolo a.C.
allanno 2000. Presidenza del Consiglio dei Ministri, Dipartimento per i Servizi
Tecnici Nazionali, Servizio Idrografico, Mareografico Nazionale.
Betancourt, J.L., H.D. Grissino-Mayer, M.W. Salzer, and T.W. Swetnam. 2002. A Test
of Annual Resolution in Stalagmites Using Tree Rings. Quaternary Research 58:
197199.
Beven, K. 2003. Surface runoff generation. Handbook of weather, climate, and
water: atmospheric chemistry, hydrology, and societal impacts. In Handbook
of weather, climate and water: atmospheric chemistry, hydrology, and societal im-
pacts, edited by T.D. Potter and B.R. Colman, 527542. Hoboken, N.J.
Bintliff, J. 2002. Time, process and catastrophism in the study of Mediterranean
alluvial history: a review. World Archaeology 33 (3): 417435.
Blaauw, M. 2010. Methods and code for classical age-modelling of radiocarbon
sequences. Quaternary Geochronology 5: 512518.
Blaauw, M., B. van Geel, I. Kristen, B. Plessen, A. Lyaruu, D.R. Engstrom, J. van
der Plicht, and D. Verschuren. 2011. High-resolution 14C dating of a 25,000-year
lake-sediment record from equatorial East Africa. Quaternary Science Reviews
30: 30433059.
Blaauw, M., B. van Geel, and J. van der Plicht. 2004. Solar forcing of climatic change
during the mid-Holocene: indications from raised bogs in The Netherlands. The
Holocene 14: 3544.
bibliography 281
Black, E. 2011. The influence of the North Atlantic Oscillation and European circu-
lation regimes on the daily to interannual variability of winter precipitation in
Israel. International Journal of Climatology 32(11): 16541664.
Black, E., D. Brayshaw, S. Black, and C. Rambeau. 2011. Using proxy data, his-
torical climate data and climate models to investigate aridification during the
Holocene. In Water, Life and Civilisation: Climate, Environment and Society in the
Jordan Valley, edited by S. Mithen and E. Black, 105112. Cambridge.
Blackman, D.J. 1973. Evidence of Sea Level Change in Ancient Harbours and Coastal
Installations. In Marine archaeology (Colston Symposium Papers 23), edited by
D.J. Blackman, 115139. London.
Blender, R., K. Fraedrich, and B. Hunt. 2006. Millennial Climate Variability: GCM-
Simulation and Greenland Ice Cores. Geophysical Research Letters 33: 24.
Blockley, S.P.E., C.B. Ramsey, C.S. Lane, and A.F. Lotter. 2008. Improved age mod-
elling approaches as exemplified by the Revised Chronology for the Central Euro-
pean Varved Lake, Soppensee. Quaternary Science Reviews 27: 6167.
Boardman, J. 2011. Fish and the Mediterranean: The Nourishing Sea. Ancient West
and East 10: 19.
Bonardi, G., B. DArgenio, V. Perrone. 1988. Carta Geologica dellAppennino merid-
ionale in scala 1: 250.000. 74 Congresso della Siociet Geologica Italiana, Sorrento,
1317 settembre 1988. Memorie della Societ Geologica, 41. Rome.
Bond, G., B. Kromer, J. Beer, R. Muscheler, M. Evans, W. Showers, S. Hoffman,
R. Lotti-Bond, I. Hajdas, and G. Bonani. 2001. Persistent Solar Influence on North
Atlantic Surface Circulation During the Holocene. Science 294: 21302136.
Bond, G.C., and R. Lotti. 1995. Iceberg Discharges into the North Atlantic on Mil-
lennial Time Scales During the Last Glaciation. Science 267: 10051010.
Bond, G., W. Showers, M. Cheseby, R. Lotti, P. Almasi, P. deMenocal, P. Priore,
H. Cullen, I. Hajdas, and G. Bonani. 1997. A pervasive millennial-scale cycle in
North Atlantic Holocene and glacial climates. Science 278: 12571266.
Bonneau, D. 1964. La crue du Nil, divinit gyptienne, travers mille ans dhistoire (332
av.-641 ap. J.-C.) daprs les auteurs grecs et latins, et les documents des poques
ptolmaque, romaine et byzantine. tudes et commentaires, 52. Paris.
1966. Utilisation des documents papyrologiques, numismatiques et pi-
graphiques pour la dtermination de la qualit de la crue du Nil, chaque anne
de lpoque grco-romaine. Atti dellXI Congresso internazionale di papirologia,
Milano 28 settembre 1965, 1: 379395. Milan.
1971. Le fisc et le Nil. Incidences des irrgularits de la crue du Nil sur la fiscalit
foncire dans lgypte grecque et romaine. Paris.
Bookman (Ken-Tor) R., Y. Enzel, A. Agnon, and M. Stein. 2004. Late Holocene lake
levels of the Dead Sea. Geological Society of America Bulletin 116: 555571.
Bootle, K.R. 2004. Wood in Australia. 2nd ed. Sydney.
Boswijk, G., A. Fowler, A. Lorrey, J. Palmer, and J. Ogden. 2006. Extension of the New
Zealand kauri (Agathis australis) chronology to 1724BC. The Holocene 16: 112.
Bottema, S., G. Entjes-Nieborg, and W. van Ziest, eds. 1990. Mans role in the shaping
of the Eastern Mediterranean landscape. Rotterdam.
Bourillon, J. 2005. Etude paloenvironnementale du port antique de Naples: le site
de Piazza Municipio. Mmoire de Master 2 Gographie Physique, Institut de
Gographie, Universit de Provence, France.
282 bibliography
Bradley, R.S., K.R. Briffa, J. Cole, M.K. Hughes, and T.J. Osborn. 2003. The climate
of the last millennium. In Paleoclimate, Global Change and the Future, edited by
K.D. Alverson, R.S. Bradley and T.F. Pedersen, 105149. Berlin.
Bradley, R.S., and P.D. Jones. 1993. Little Ice Age summer temperature variations:
Their nature and relevance to recent global warming trends. The Holocene 3:
367376.
Brzdil, R., C. Pfister, H. Wanner, H. von Storch, and J. Luterbacher. 2005. Historical
climatology in EuropeThe state of the art. Climatic Change 70: 363430.
Bresson, A. Ecology and Beyond: the Mediterranean Paradigm. In Harris 2005a,
94114.
2006. La machine dHron et le cot de lnergie dans le monde antique.
In Innovazione tecnica e progresso economico nel mondo romano, edited by E. Lo
Cascio, 5580. Bari.
Briffa, K.R. 2000. Annual climate variability in the Holocene: interpreting the mes-
sage of ancient trees. Quaternary Science Reviews 19: 87105.
Briffa, K.R., T.J. Osborn, F.H. Schweingruber, I.C. Harris, P.D. Jones, S.G. Shiyatov, and
E.A. Vaganov. 2001. Low-frequency temperature variations from a northern tree
ring density network. Journal of Geophysical Research 106: 29292941.
Bronk Ramsey, C. 2008. Deposition models for chronological records. Quaternary
Science Reviews 27: 4260.
Bronk Ramsey, C., J. van der Plicht, and B. Weninger. 2001. Wiggle matching
radiocarbon dates. Radiocarbon 43: 381389.
Broodbank, C. 2000. An Island Archaeology of the Early Cyclades. Cambridge.
2011. The Mediterranean and the Mediterranean World in the Age of Andrew
Sherratt. In Interweaving Worlds: Systemic Interactions in Eurasia, 7th to 1st Mil-
lennia bc, edited by T.C. Wilkinson, S. Sherratt and J. Bennet, 2736. Oxford.
Brooks, N. 2006. Cultural responses to aridity in the Middle Holocene and increased
social complexity. Quaternary International 151: 2949.
Brown, P. 1971. The world of late antiquity: from Marcus Aurelius to Muhammad.
London.
Brown, W. 1820. Antiquities of the Jews, Carefully Compiled from Authentic Sources
and Their Customs Illustrated from Modern Travels, Vol. 2. London.
Bruins, H.J. 1986. Desert Environment and Agriculture in the Central Negeb and
Kadesh-Barnea during Historical Times. Ph.D. dissertation, The Agricultural Uni-
versity of Wageningen. Nijkerk.
1994. Comparative chronology of climate and human history in the south-
ern Levant from the Late Chalcolithic to Early Arab Period. In Late Quaternary
Chronology and Paleoclimates of the Eastern Mediterranean, edited by O. Bar-
Yosef and R.S. Kra, 301314. Tucson.
Brun, J-P. 2006. Lnergie hydraulique durant lempire Romain: quel impact sur
lconomie agricole? In Innovazione tecnica e progresso economico nel mondo
romano, edited by E. Lo Cascio, 101131. Bari.
Brunt, P.A. 1971. Italian Manpower, 225BC14AD. Oxford.
1980. Free Labour and Public Works at Rome. JRS 70: 81100.
Bruun, C.F.M. 2000. The Roman World. In Handbook of Ancient Water Technology,
edited by O. Wikander, 675604. Leiden.
2003. The Antonine plague in Rome and Ostia. JRA 16: 426434.
bibliography 283
2007. The Antonine Plague and the Third-Century Crisis. In Crises and
the Roman Empire, (Impact of Empire 7), edited by O. Hekster, G. de Kleijn, and
D. Slootjes, 201217. Leiden.
Buckley, B.M., K.J. Anchukaitis, D. Penny, R. Fletcher, E.R. Cook, M. Sano, L.C. Nam,
A. Wichienkeeo, T.T. Minh, T.M. Hong. 2010. Climate as a contributing factor
in the demise of Angkor, Cambodia. Proceedings of the National Academy of
Sciences 107: 67486752.
Buisman, J., and A.F.V. van Engelen, ed. 1995. Duizend jaar weer, wind en water in de
Lage Landen. Franeker.
Bunn, S.E., and A.H. Arthington. 2002. Basic principles and ecological conse-
quences of altered flow regimes for aquatic biodiversity. Environmental Man-
agement 30: 492507.
Bntgen, U., D.C. Frank, D. Nievergelt, and J. Esper. 2006. Summer temperature
variations in the European Alps, AD 7552004. Journal of Climate 19: 56065623.
Bntgen, U., W. Tegel, K. Nicolussi, M. McCormick, D. Frank, V. Trouet, J.O. Kaplan,
F. Herzig, K.-U. Heussner, H. Wanner, J. Luterbacher, and J. Esper. 2011. 2500 Years
of European Climate Variability and Human Susceptibility. Science 331: 578582.
Bntgen, U., V. Trouet, D. Frank, H.H. Leuschner, D. Friedrichs, J. Luterbacher, and
J. Esper. 2010. Tree-ring indicators of German summer drought over the last
millennium. Quaternary Science Reviews 29: 10051016.
Burton, R.F. 1875. Notes on Rome. Macmillans Magazine 31: 126134.
Butzer, K.W. 2003. Review of A. Grove and O. Rackham, The Nature of Mediterranean
Europe. Annals of the Association of American Geographers 93: 494498.
2005. Environmental history of the Mediterranean world: cross-disciplinary
investigation of cause-and-effect for degradation and soil erosion. Journal of
Archaeological Science 32: 17731800.
2011. Geoarchaeology, climate and sustainability: a Mediterranean Perspec-
tive. In Geoarchaeology. Climate and Sustainability, Special Paper 476, edited by
A.G. Brown, L.S. Basell, and K.W. Butzer, 112. Denver.
2012. Collapse, Environment, and Society. Proceedings, National Academy of
Sciences USA 109: 36323639.
Butzer, K.W., and S.E. Harris. 2007. Geoarchaeological Approaches to the Environ-
mental History of Cyprus: Explication and Critical Evaluation. Journal of Archae-
ological Science 34: 19321952.
Butzer, K.W. and D.M. Helgren. 2005. Livestock, land cover and environmental his-
tory: the Tablelands of New South Wales, Australia, 18201920. Annals, Associa-
tion of American Geographers 95: 80111.
Cacho, I., J.O. Grimalt, M. Canals, L. Sbaffi, N.J. Shackleton, J. Schnfeld, and R. Zahn.
2001. Variability of the western Mediterranean Sea surface temperature during
the last 25,000 years and its connection with the Northern Hemisphere climatic
changes. Paleoceanography 16: 4052.
Cadell, H. 1976. P. Genve 60, B.G.U. II 456 et le problme du bois en Egypte.
Chronique dgypte 51: 331348.
Calenda, G., C.P. Mancini, and E. Volpi. 2009. Selection of the probabilistic model
of extreme floods: The case of the River Tiber in Rome. Journal of Hydrology
371(14): 111.
Cameron, A. 1993. The Mediterranean world in late antiquity, AD 395600. London.
284 bibliography
Camp, J. McK. 1979. A Drought in the Late Eighth Century B.C. Hesperia 48: 397411.
Campbell, J.B. 2012. Rivers and the power of ancient Rome. Chapel Hill, NC.
Camuffo, D., and S. Enzi. 1995. Climatic features during the Sporer and Maunder
Minima. PaloklimaforschungPalaeoclimate Research 16: 105124.
1996. The analysis of two bi-millenial series: Tiber and Po river floods.
In Climatic variations and forcing mechanisms of the last 2000 years, edited by
P.D. Jones, R.S. Bradley and J. Jouzel, 432450. Berlin; Heidelberg.
Cane, M.A. 2005. The evolution of El Nio, past and future. Earth and Planetary
Sciences Letters 230: 227240.
Cancellieri, C. 1986. Lacquedotto Paolo (sec. XVIIXX). In Il trionfo dellacqua:
Acque e acquedotti a Roma IV Sec. a.C.XX Sec., edited by G. Pisani Sartorio and
A. Liberati Silverio, 225231. Rome.
Carmichael, G.A. 1993. St. Anthonys Fire. The Cambridge World History of Human
Disease. Cambridge.
Carrin, Y. 2006. Tres Montes (Navarra, Spain): Dendrology and wood uses in
an arid environment. In Charcoal Analysis: New Analytical Tools and Methods
for Archaeology. Papers from the Table-Ronde held in Basel 2004, edited by A.
Dufraisse, 8394. Oxford.
Carsana, V., S. Febbraro, D. Giampaola, C. Guastaferro, G. Irollo, and M.R. Ruello.
2009. Evoluzione del paesaggio costiero tra Parthenope e Neapolis. Mditerrane
112: 1522.
Casana, J. 2008. Mediterranean valleys revisited: Linking soil erosion, land use and
climate variability in the Northern Levant. Geomorphology 101: 429442.
Caseldine, C.J., and C. Turney. 2010. The bigger picture: towards integrating palaeo-
climate and environmental data with a history of societal change. Journal of
Quaternary Science 25: 8893.
Cavarzere, A., ed. 2003. Decimus Magnus Ausonius: Mosella. Introduzione, testo,
traduzione e commento. Amsterdam.
Cawkwell, G.L. 1992. Early Colonisation. Classical Quarterly 42: 289303.
Certini, G., and R. Scalenghe. 2011. Anthropogenic soils are the golden spikes for the
Anthropocene. The Holocene 21: 12691274.
Chabal, L., L. Fabre, J.-F. Terral, and I. Thry-Parisot. 1999. LAnthracologie. In
La Botanique, edited by C. Bourquin-Mignot, J-E. Brochier, L. Chabal, S. Crozat,
L. Fabre, F. Guibal, P. Marinval, H. Richard, J-F. Terral, and I. Thry-Parisot, 43104.
Paris.
Chakrabarty, D. 2009. The Climate of History: Four Theses. Critical Inquiry 35:
197222.
Chambers, F.M., D. Mauquoy, S.A. Brain, M. Blaauw, and J.R.G. Daniell. 2007. Glob-
ally synchronous climate change 2800 years ago: proxy data from peat in South
America. Earth and Planetary Science Letters 253: 439444.
Chen, D., M. Cane, A. Kaplan, S. Zebiak, and D. Huang. 2004. Predictability of El
Nio over the past 148 years. Nature 428: 733736.
Chen, Z., S.E. Grasby, and K.G. Osadetz. 2004. Relation between climate variability
and groundwater levels in the upper carbonate aquifer, southern Manitoba,
Canada. Journal of Hydrology 290(12): 4362.
Cheng, H., D. Fleitmann, R.L. Edwards, X.F. Wang, F.W. Cruz, A.S. Auler, A. Mangini,
Y.J. Wang, S.J. Burns, and A. Matter. 2009. Timing and structure of the 8.2 ky event
bibliography 285
inferred from 18O records of stalagmites from China, Oman and Brazil. Geology
37: 10071010.
Chew, S.C. 2002. Globalisation, ecological crisis, and Dark Ages. Global Society 16:
333356.
Chierici, R. 1911. I boschi nelleconomia generale dItalia. Loro stima. Caserta.
Christiansen, B., and F.C. Ljungqvist. 2011. Reconstruction of the Extratropical NH
Mean Temperature over the Last Millennium with a Method that Preserves
Low-Frequency Variability. Journal of Climate 24: 60136034.
Chrzavzez, J., I. Thry-Parisot, J.-F. Terral, A. Ducom, and G. Fiorucci. 2011. Differen-
tial preservation of anthracological material and mechanical properties of wood
charcoal, an experimental approach of fragmentation. In 5th International Meet-
ing of Charcoal Analysis. The charcoal as cultural and biological heritage, edited
by E. Badal, Y. Carrin, E. Grau, M. Macas, and M. Ntinou, 2930. Valncia.
Clausen, H.B., C.U. Hammer, C.S. Hvidberg, D. Dahl-Jensen, J.P. Steffensen, J. Kipfs-
tuhl, and M. Legrand. 1997. A comparison of the volcanic records over the past
4000 years from the Greenland Ice Core Project and Dye 3 Greenland ice cores.
Journal of Geophysical Research 102 C12: 26,707726,723.
Cline, W.R. 2007. Global Warming and Agriculture: Impact Estimates by Country.
Washington, D.C.
Coarelli, F. 1997. Il Campo Marzio: dalle origini alla fine della repubblica. Rome.
Coates-Stephens, R. 2003. The Water Supply of Rome from Late Antiquity to the
Early Middle Ages. Acta ad Archaeologiam et Artium Historiam Pertinentia 17:
165186.
Coldstream, J.N. 1977. Geometric Greece. London.
Collaborative Partnership on Forests. 2011. Energy grows on trees: Global forest
institutions highlight vital role of forests to cope with future demands for bioen-
ergy, but call for balanced approaches. International Union of Forest Research
Organisations (IUFRO), United Nations Forum on Forests Secretariat, Interna-
tional Year of Forests New Releases, 8 November, 2011. Available at: http://www
.cpfweb.org/75548/en/.
Conedera, M., P. Krebs, W. Tinner, M. Pradella, and D. Torriani. 2004. The cultivation
of Castanea sativa (Mill.) in Europe, from its origin to its diffusion on a continen-
tal scale. Vegetation History and Archaeobotany 13: 161179.
Cook, E. 1971. The Flow of Energy in an Industrial Society. Scientific American 225,
3: 135142.
1976. Man, Energy, Society. San Francisco.
Cook, E.R., K.J. Anchukaitis, B.M. Buckley, R.D. DArrigo, G.C. Jacoby, and W.E.
Wright. 2010. Asian monsoon failure and megadrought during the last millen-
nium. Science 328(5977): 486489.
Cook, E.R., R. DArrigo, and M.E. Mann. 2002. A Well-Verified, Multiproxy Recon-
struction of the Winter North Atlantic Oscillation Index since A.D. 1400. Journal
of Climate 15: 17541764.
Cook, E.R., and R.L. Holmes. 1999. USERS MANUAL for Program ARSTAN. February
1999. Tucson, AZ.
Cook, E.R., and L.A. Kairiukstis, eds. 1990. Methods of Dendrochronology: Applica-
tions in the Environmental Sciences. Dordrecht.
286 bibliography
Cook, E.R., C.A. Woodhouse, C.M. Eakin, D.M. Meko, and D.W. Stahle. 2004. Long-
Term Aridity Changes in the Western United States. Science 306: 10151018.
Corella, J.P., A. Moreno, M. Morelln, V. Rull, S. Giralt, M.T. Rico, A. Prez-Sanz,
and L. Valero-Garcs. 2011. Climate and human impact on a meromictic lake
during the last 6,000 years (Montcorts Lake, Central Pyrenees, Spain). Journal
of Paleolimnology 46: 351367.
Corretti, A., and M. Firmati. Metallurgia antica e medievale allisola dElba: vecchi
dati e nuove acquisizioni. In Archeometallurgia: dalla conoscenza alla fruizione,
edited by C. Giardino, 229241. Bari.
Cokun, A. 2002a. Die gens Ausoniana an der Macht: Untersuchungen zu Decimius
Magnus Ausonius und seiner Familie. Oxford.
2002b. Ein geheimnisvoller gallischer Beamter in Rom, ein Sommerfeldzug
Valentinians und weitere Probleme in Ausonius Mosella. Revue des tudes anci-
ennes 104: 401431.
Cremaschi, M., M. Pelfini, and M. Santilli. 2006. Cupressus dupreziana: a dendrocli-
matic record for the middle-late Holocene in the central Sahara. The Holocene
16.2: 293303.
Crook, D. 2009. Hydrology of the combination irrigation system in the Wadi Faynan,
Jordan. Journal of Archaeological Science 36: 24272436.
Crow, J. 2012. Ruling the waters: managing the water supply of Constantinople,
AD 3301204. Water History 4.1: 3555.
Crow, J., J. Bardill, and R. Bayliss. 2008. The water supply of Byzantine Constantinople
(Journal of Roman Studies Monograph). London.
Crutzen, P.J., and E.F. Stoermer. 2000. The Anthropocene. Global Change Newslet-
ter 41: 1718.
Cuffey, K.M., and G.D. Clow. 1997. Temperature, accumulation, and ice sheet eleva-
tion in central Greenland through the last deglacial transition. Journal of Geo-
physical Research 102: 2638326396.
DAgostino, B., and D. Giampaola. 2005. Osservazioni storiche e archeologiche sulla
fondazione di Neapolis. In Noctes Campanae, studi di storia antica e archeologia
dellItalia pre-romana e romana in memoria di Martin W. Frederiksen, edited by
W.V. Harris and E. Lo Cascio, 6372. Naples.
DArms, J.H. 1974. Puteoli in the Second Century of the Roman Empire: A Social and
Economic Study. JRS 64: 104124.
2000. Memory, Money and Status at Misenum: Three New Inscriptions from
the Collegium of Augustales. JRS 90: 126144.
Dai, A.G., and T.M.L. Wigley. 2000. Global patterns of ENSO-induced precipitation.
Geophysical Research Letters 27: 12831286.
Dalby, A., trans. 2011. Geoponika. Farm Work: a modern translation of the Roman and
Byzantine farming handbook. Totnes, Devon.
Dalfes, H.N., G. Kukla, and H. Weiss, eds. 1997. Third Millennium BC Climate Change
and Old World Collapse. NATO ASI Series I, vol. 49. Berlin.
Danti, M.D. Late Middle Holocene Climate amd Northern Mesopotamia: Varying
Cultural Responses to the 5.2 and 4.2 ka Aridification Events. In Mainwaring,
Giegengack, and Vita-Finzi 2010, 139172.
Dark, P. 2000. The environment of Britain in the first millennium AD. London.
bibliography 287
de Callata, F. 2005. The Graeco-Roman economy in the super long-run: lead,

copper and shipwrecks. JRA 18: 361372.
De Caro, S. 1999. Lattivit della Soprintendenza archeologica di Napoli e Caserta
nel 1998. In LItalia meridionale in et tardo antica. Atti del trentottesimo convegno
di studi sulla Magna Grecia. Taranto 26 ottobre 1998, edited by A. Stazio and
S. Ceccoli, 635661. Taranto.
2003. Lattivit archeologica a Napoli e Caserta nel 2002. In Ambiente e
paesaggio nella Magna Grecia. Atti del quarantaduesimo convegno di studi sulla
Magna Grecia. Taranto 58 ottobre 2002, 2: 569621. Taranto.
De Caro, S., and C. Gialanella, eds. 2002. Il Rione Terra di Pozzuoli. Naples.
de Criscio, G. 1881. Notizie istoriche archeologiche topografiche dellantica citta di
Pozzuoli e dei suoi due acquidotti Serino e Campano. Naples.
de la Ville de Mirmont, H. 1889. La Mosella dAusone. dition critique et traduction
franaise prcdes dune introduction, suivies dun commentaire explicatif. Bor-
deaux.
De Nino, A. 1900. PentimaIndagini circa il percorso dellantico acquedotto corfi-
niese. NSA: 642643.
De Vivo, B., P. Petrosino, A. Lima, G. Rolandi, and H.E. Belkin. 2010. Research
progress in volcanology in the Neapolitan area, southern Italy: a review and some
alternative views. Mineralogy and Petrology 99: 128.
De Vleeschouwer, F., L. Grard, C. Goormaghtigh, N. Mattielli, G. Le Roux, and
N. Fagel. 2007. Atmospheric lead and heavy metal pollution records from a
Belgian peat bog spanning the last two millennia: Human impact on a regional
to global scale. Science of the Total Environment 377: 282295.
Debussche, M., J. Lepart, and A. Dervieux. 1999. Mediterranean landscape changes:
evidence from old postcards. Global Ecology and Biogeography 8: 315.
Deino, A.L., G. Orsi, S. de Vita, and M. Piochi. 2004. The age of the Neapolitan
Yellow Tuff calderaforming eruption (Campi Flegrei CalderaItaly) assessed
by 40Ar/39Ar dating method. Journal of Volcanology and Geothermal Research 133:
157170.
DeLaine, J. 1996. Hypocaust. In The Oxford Classical Dictionary, 3rd edn, edited by
S. Hornblower and A. Spawforth. Oxford.
deMenocal, P.B. 2001. Cultural Responses to Climate Change During the Late Holo-
cene. Science 292: 667673.
Denton, G., and W. Karln. 1973. Holocene climatic variationstheir pattern and
possible cause. Quaternary Research 3(2): 155205.
Deser, C. 2000. On the Teleconnectivity of the Arctic Oscillation. Geophysical
Research Letters 27: 779782.
Desnier, J.L. 1998. Les dbordements du Fleuve. Latomus 57: 513522.
Desprat, S., M.F. Snchez-Goi, and M.F. Loutre. 2003. Revealing climatic variability
of the last three millennia in northwestern Iberia using pollen influx data. Earth
and Planetary Science Letters 213: 6378.
Devillers, B., and N. Lecuyer. 2008. Le Petit ge Glaciaire en milieu semi-aride: le
bassin versant du Gialias (Chypre) et ses relations avec loccupation des sols.
Zeitschrift fr Geomorphologie 52(2): 207224.
Di Donato, V., P. Esposito, E. Russo Ermolli, A. Scarano, and R. Cheddadi. 2008. Cou-
pled atmospheric and marine palaeoclimatic reconstruction for the last 35 kyr
288 bibliography
in the Sele Plain-Gulf of Salerno area (southern Italy). Quaternary International

190: 146157.
Di Vito, M.A., R. Isaia, G. Orsi, J. Southon, S. de Vita, M. DAntonio, L. Pappalardo,
and M. Piochi. 1999. Volcanism and deformation since 12,000 years at the Campi
Flegrei caldera (Italy). Journal of Volcanology and Geothermal Research 91: 221
246.
Di Rita, F., and D. Magri. 2009. Holocene Drought, Deforestation, and Evergreen
Vegetation Development in the Central Mediterranean: a 5500 Year Record from
Lago Alimini Piccolo, Apulia, Southeast Italy. The Holocene 19: 295306.
Diaz, H.F., M.P. Hoerling, and J.K. Eischeid. 2001. ENSO variability, teleconnections,
and climate change. International Journal of Climatology 21: 18451862.
Diaz, H.F., and D.W. Stahle. 2007. Climate and cultural history in the Americas: an
overview. Climatic Change 83: 18.
Diaz, H.F., R. Trigo, M.K. Hughes, M.E. Mann, E. Xoplaki, and D. Barriopedro. 2011.
Spatial and Temporal Characteristics of Climate in Medieval Times Revisited.
Bulletin of the American Meteorological Society 92: 14871500.
Dincauze, D.F. 2000. Environmental archaeology. Principles and practice. Cambridge.
Dring, M. 2007. Rmische Hfen, Aqudukte und Zisternen in Campanien. Bestands-
aufnahme der antiken Wasserbauten. Darmstadt.
Dragoni, W. 1998. Some considerations on climatic changes, water resources and
water need in the Italian region south of the 43N. In Water, Environment and
Society in Times of Climatic Change, edited by S. Issar and N. Brown, 241271.
Dordrecht; Boston.
Dragoni, W., and B.S. Sukhija. 2008. Climate change and groundwater: a short
review. In Climate change and groundwater, edited by W. Dragoni and B.S.
Sukhija, 112. London.
Drake, B.L. 2012. The influence of climatic change on the late Bronze Age Collapse
and the Greek Dark Ages. Journal of Archaeological Science 39: 18621870.
Drew-Bear, M. 1995. Le bois en gypte daprs les papyrus dpoque romaine. In
Larbre et le fort: le bois dans lantiquit, edited by J.-C. Bal, 39. Paris.
Drinkwater, J.F. 1999. Re-Dating Ausonius War Poetry. AJPh 120: 443452.
Drysdale, R.N., G. Zanchetta, J.C. Hellstrom, R. Maas, A.E. Fallick, M. Pickett, I.
Cartwright, and L. Piccini. 2006. Late Holocene drought responsible for the
collapse of Old World civilizations is recorded in an Italian cave flowstone.
Geology 34: 101104.
du Coudray la Blanchre, R.M. 1895. Lamnagement de leau et l installation rurale
dans lAfrique ancienne. Paris.
du Coudray la Blanchre, R.M., and P. Gauckler. 1897. Catalogue du Muse Alaoui
(Muses et collections archologiques de lAlgrie et de la Tunisie). Paris.
Dubois, C. 1907. Pouzzoles antique. Paris.
Dull, R., J.R. Southon, S. Kutterolf, A. Freundt, D. Wahl, and P.D. Sheets. 2010. Did the
TBJ Ilopango eruption cause the AD 536 event? In American Geophysical Union,
Fall Meeting 2010, abstract #V13C-2370. Washington.
Dull, R.A., J.R. Southon, and S. Payson. 2001. Volcanism, Ecology and Culture: A
Reassessment of the Volcn Ilopango Tbj eruption in the Southern Maya Realm.
Latin American Antiquity 12.1: 2544.
Duncan-Jones, R.P. 1996. The impact of the Antonine plague. JRA 9: 108136.
bibliography 289
Eastwood, W.J., N. Roberts, and H.F. Lamb. 1998. Palaeoecological and archaeologi-
cal evidence for human occupance in southwest Turkey: the Beyhir Occupation
Phase. AS 48: 6986.
Eddy, J.A. 1976. The Maunder Minimum. Science 192: 1189202.
1977. Climate and the changing sun. Climate Change 1: 173190.
Ehrmann, W., G. Schmiedl, Y. Hamann, T. Kuhnt, C. Hemleben, and W. Siebel.
2007. Clay minerals in late glacial and Holocene sediments of the northern and
southern Aegean Sea. Palaeogeography, Palaeoclimatology, Palaeoecology 249:
3657.
Emeis, K.C., U. Struck, H.M. Schulz, R. Rosenberg, S. Bernasconi, H. Erlenkeuser,
T. Sakamoto, and F. Martinez-Ruiz. 2000. Temperature and salinity variations
of Mediterranean Sea surface waters over the last 16,000 years from records of
planktonic stable oxygen isotopes and alkenone unsaturation ratios. Palaeo-
geography, Palaeoclimatology, Palaeoecology 158: 259280.
English Heritage. 2000. Thatch and thatching: a guidance note. http://www.english
-heritage.org.uk/publications/thatch-and-thatching/.
Enzel, Y., R. Bookman, D. Sharon, H. Gvirtzman, U. Dayan, B. Ziv, and M. Stein. 2003.
Late Holocene climates of the Near East deduced from Dead Sea level variations
and modern regional winter rainfall. Quaternary Research 60: 263273.
Ermolli, E.R., and G. Di Pasquale. 2002. Vegetation Dynamics of South-western
Italy in the Last 28kyr Inferred from Pollen Analysis of a Tyrrhenian Sea Core.
Vegetation History and Archaeobotany 11: 211219.
Esper, J., E.R. Cook, F.H. Schweingruber. 2002. Low-Frequency Signals in Long
Tree-Ring Chronologies for Reconstructing Past Temperature Variability. Science
295: 22502253.
Esper, J., E.R. Cook, P.J. Krusic, K. Peters, and F.H. Schweingruber. 2003. Tests of the
RCS method for preserving low-frequency variability in long tree-ring chronolo-
gies. Tree Ring Research 59: 8198.
Esper, J., D.C. Frank, U. Bntgen, A. Verstege, J. Luterbacher, and E. Xoplaki. 2007.
Long-term drought severity variations in Morocco. Geophysical Research Letters
34.
Esprandieu, E. 1926. Le Pont du Gard et lAqueduc de Nmes. Paris.
Evans, H.B. 2002. Aqueduct Hunting in the Seventeenth Century: Rafaello Fabrettis De
aquis et aquaeductibus veteris Romae. Ann Arbor, MI.
Evenari, M., L. Shanan, and T. Naphtali. 1971. The Negev: The Challenge of a Desert.
Cambridge, MA.
Fabre, G., J.-L. Fiches, and J.-L. Paillet. 1991. Interdisciplinary research on the aque-
duct of Nimes and the Pont du Gard. JRA 4: 6388.
Fagan, B.M. 2000. The little ice age: how climate made history 13001850. New York.
Falconer, S., and C.L. Redman, eds. 2009. Polities and Power: Archaeological Perspec-
tives on the Landscapes of Early States. Tucson, AZ.
Fall, P.L., S.E. Falconer, L. Lines, and M.C. Metzger. 2004. Environmental Impacts
of the Rise of Civilization in the Southern Levant. In The Archaeology of Global
Change: the Impact of Humans on their Environment, edited by C.L. Redman,
S.R. James, P.R. Fish, and J.D. Rogers, 141157. Washington.
Fentress, E.W.B., and A.I. Wilson. Forthcoming. The Saharan Berber Diaspora and
the Southern Frontiers of Vandal and Byzantine North Africa. In Rome Re-
290 bibliography
imagined: Byzantine and Early Islamic Africa, AD 500800, (Dumbarton Oaks

Byzantine Symposia and Colloquia), edited by J. Conant and S. Stevens. Wash-
ington.
Fernandes S.D., N.M. Trautmann, D.G. Streets, C.A. Roden, T.C. Bond. 2007. Global
Biofuel Use, 18502000. Global Biogeochemical Cycles 21: 115.
Ferrari Fontana, I., B.I. Menozzi, and C. Montanari. 2008. Charcoal as Environmen-
tal and Ethnological Evidence from Medieval Archaeological Sites in NW Italy.
In Fiorentino and Magri 2008, 105109.
Ferrio, J.P., N. Alonso, J.B. Lpez, J.L. Araus, and J. Voltas. 2006. Carbon isotope
composition of fossil charcoal reveals aridity changes in the NW Mediterranean
Basin. Global Change Biology 12: 12531266.
Field, C.V., G.A. Schmidt, D. Koch, and C. Salyk. 2006. Modeling production and
climate-related impacts on 10Be concentration in ice-cores. Journal of Geophysi-
cal Research 111: D15107.
Fiema, Z.T. 2006. City and countryside in Byzantine Palestine. Prosperity in ques-
tion. In Settlements and Demography in the Near East in Late Antiquity. Proceed-
ings of the colloquium, Matera 2729 October 2005. Biblioteca di Mediterraneo
Antico 2, edited by A.S. Lewin and P. Pellegrini, 6788. Pisa; Rome.
2008. The Concluding Remarks. In PetraThe Mountain of Aaron I. The
Church and the Chapel, edited by Z.T. Fiema and J. Frsn, 425441. Helsinki.
Finn, M., K. Holmgren, H.S. Sundqvist, E. Weiberg, and M. Lindblom. 2011. Cli-
mate in the Eastern Mediterranean and Adjacent Regions, during the Past 6000
YearsA Review. Journal of Archaeological Science 38: 31533173.
Fiorentino, G., and D. Magri, eds. 2008. Charcoals from the Past: Cultural and Palaeo-
environmental Implications (BARIS 1807). Oxford.
Fiorillo, F., and L. Esposito. 2006. The Serino Spring Discharge: Analyses of an Ultra-
Centenarian Hydrological Series, Southern Italy. BALWOIS Conference on Water
Observation and Information Systems For Decision Support, http://balwois.mpl
.ird.fr/balwois/administration/full_paper/ffp-632.pdf. [But apparently no longer
accessible as of June 9, 2013].
Fiorillo, F., L. Esposito, and F.M. Guadagno. 2007. Analyses and forecast of water
resources in an ultra-centenarian spring discharge series from Serino (Southern
Italy). Journal of Hydrology 336: 125138.
Fisher, C.T., J.B. Hill, and G.M. Feinman, eds. 2009. The Archaeology of Environmental
Change: Socionatural Legacies of Degradation and Resilience. Tucson, AZ.
Fleitmann, D., S.J. Burns, A. Mangini, M. Mudelsee, J. Kramers, I. Villa, U. Neff,
A.A. Al-Subbary, A. Buettner, D. Hippler, and A. Matter. 2007. Holocene ITCZ
and Indian monsoon dynamics recorded in stalagmites from Oman and Yemen
(Socotra). Quaternary Science Reviews 26: 170188.
Fleitmann, D., S.J. Burns, M. Mudelsee, U. Neff, J. Kramers, A. Mangini, and A. Matter.
2003. Holocene Forcing of the Indian Monsoon Recorded in a Stalagmite from
Southern Oman. Science 300: 17371739.
Fleitmann, D., H. Cheng, S. Badertscher, R.L. Edwards, M. Mudelsee, O.M. Gktrk,
A. Fankhauser, R. Pickering, C.C. Raible, A. Matter, J. Kramers, and O. Tysz.
2009. Timing and climatic impact of Greenland interstadials recorded in stalag-
mites from northern Turkey. Geophysical Research Letters 36: L19707.
Forni, G., and A. Marcone, eds. 2002. Storia dellagricoltura italiana, I. Firenze.
bibliography 291
Food and Agricultural Organization of the United Nations, Land and Water devel-
opment division, Crop Water Management, Wheat, Water Supply and Crop
Yield: http://www.fao.org/landandwater/aglw/cropwater/wheat.stm#supply,
consulted January 4, 2012.
2011. Forestry. Key Facts. Global Forestry Resources Assessment. Food and
Agriculture Organisation of the United Nations. Available from: http://www.fao
.org/docrep/014/am859e/am859e08.pdf.
Fowler, A.M. 2008. ENSO history recorded in Agathis australis (kauri) tree rings.
Part B: 423 years of ENSO robustness. International Journal of Climatology 28:
2135.
Fowler, A.M., J. Palmer, J. Salinger, and J. Ogden. 2000. Dendroclimatic interpreta-
tion of tree-rings in Agathis australis (kauri). 2. Evidence of a significant relation-
ship with ENSO. Journal of the Royal Society of New Zealand 30: 277292.
Fowler, A.M., G. Boswijk, J. Gergis, and A. Lorrey. 2008. ENSO history recorded in
Agathis australis (kauri) tree rings. Part A: kauris potential as an ENSO proxy.
International Journal of Climatology 28: 120.
Franconi, T. forthcoming 2013. The power of rivers in the world of ancient Rome.
JRA 23.2.
Frank, D., J. Esper, and E.R. Cook. 2007. Adjustment for proxy number and coher-
ence in a large-scale temperature reconstruction. Geophysical Research Letters
34: L16709.
Frederiksen, M.W. 1984. Campania. London.
Freiwan, M., and M. Kadioglu. 2008. Spatial and temporal analysis of climatological
data in Jordan. International Journal of Climatology 28: 521535.
Friedrich, W.L., B. Kromer, M. Friedrich, J. Heinemeier, T. Pfeiffer, and S. Talamo.
2006. Santorini eruption radiocarbon dated to 16271600B.C. Science 312: 548.
Frumkin, A. 1997. The Holocene history of Dead Sea levels. In The Dead Sea. The
Lake and Its Setting. Oxford Monographs on Geology and Geophysics 36, edited
by T. Niemi, Z. Ben-Avraham, and J.R. Gat, 237248. New York; Oxford.
Gales, B., A. Kander, P. Malanima, and M. Rubio. 2007. North versus South: Energy
Transition and Energy Intensity in Europe over 200 Years. European Review of
Economic History 11: 219253.
Gallant, T.W. 1991. Risk and Survival in Ancient Greece. Cambridge.
Garca, M.J.G., M.B.R. Zapata, J.I. Santisteban, R. Mediavilla, E. Lpez-Pamo, and
C.J. Dabrio. 2007. Late holocene environments in Las Tablas de Daimiel (south
central Iberian peninsula, Spain). Vegetation History and Archaeobotany 16: 241
250.
Gardiner, A. 1961. Egypt of the Pharaohs: an Introduction. Oxford.
Garnsey, P. 1988. Famine and food supply in the Graeco-Roman world: responses to risk
and crisis. Cambridge.
Geiger, J. 2009. Rome and Jerusalem: Public Building and the Economy. Herod and
Augustus. In Papers presented at the IJS conference, 21st23rd June 2005, edited by
D.M. Jacobson and N. Kokkinos, 157169. Leiden.
Gelabert, L.P., E. Asouti, and E.A. Mart. 2011. The ethnoarchaeology of firewood
management in the Fang Village of Equatorial Guinea, central Africa: Impli-
cations for the interpretation of wood fuel remains from archaeological sites.
Journal of Anthropological Archaeology 30: 375384.
292 bibliography
Gerasimidis, A. 2000. Palynological Evidence for Human Influence on the Vege-

tation of Mountain Regions in Northern Greece: the Case of Lailias, Serres. In
Halstead and Frederick 2000, 2837.
Giachi, G., S. Lazzeri, M. Mariotti Lippi, N. Macchioni, and S. Paci. 2003. The Wood
of C and F Roman Ships Found in the Ancient Harbour of Pisa (Tuscany, Italy):
the Utilisation of Different Timbers and the Probable Geographical Area which
Supplied them. Journal of Cultural Heritage 4: 269283.
Gialanella, C. 1993a. Il Rione Terra alla luce dei nuovi scavi archeologici. Rione
Terra, Pozzuoli. Bollettino di Archeologia 22: 8491.
1993b. La Topografia di Puteoli. In Puteoli, edited by F. Zevi, 7398. Napoli.
Giampaola, D. 2004. Dagli studi di Bartolomeo Capasso agli scavi della Metropoli-
tana: ricerche sulle mura di Napoli e sullevoluzione del paesaggio costiero.
Napoli Nobilissima 5(III): 3556.
2011. Archeologia e opere pubbliche: lo scavo della Stazione Chiaia della Linea
6 della Metropolitana. Conferenza Museo Archeologico Nazionale di Napoli, 20
ottobre 2011.
Giampaola, D., V. Carsana, G. Boetto, M. Bartolini, C. Capretti, G. Galotta, G. Giachi,
N. Macchioni, M.P. Nugari, B. Pizzo. 2006. La scoperta del porto di Neapolis: dalla
ricostruzione topografica allo scavo e al recupero dei relitti. Archeologia Marit-
tima Mediterranea, International Journal on Underwater Archaeology 2, 4791,
Istituti Editoriali e Poligrafici Internazionali MMVI. Pisa; Rome.
Giardina, A. 1981. Allevamento ed economia della selva in Italia meridionale: tras-
formazioni e continuit. In Societ romana e produzione schiavistica, edited by
A. Giardina and A. Schiavone, I, 87113. Rome; Bari.
Giesecke T., W.O. van der Knaap, and F. Bittmann. 2010. Towards quantitative
palynology: using pollen accumulation rates and models of pollen dispersal.
Gilliam, J.F. 1961. The plague under Marcus Aurelius. AJPh 82.3: 225251.
Giraudi, C. 1998. Late Pleistocene and Holocene lake-level variations in Fucino Lake
(Abruzzo, central Italy) inferred from geological, archaeological and historical
data. In Palaeohydrology as Reflected in Lake-Level Changes as Climatic Evidence
for Holocene Times, edited by S.P. Harrison, B. Frenzel, U. Huckried, and M. Weiss,
117. Stuttgart.
2004. Le oscillazioni di livello del Lago di Mezzano (Valentino-VT): vari-
azioni climatiche e interventi antropici. Il Quaternario 17: 221230.
2005. Late-Holocene alluvial events in the Central Apennines, Italy. The
Holocene 15(5): 768773.
2011. The sediments of the Stagno di Maccarese marsh (Tiber river delta,
central Italy): A late-Holocene record of natural and human-induced environ-
mental changes. The Holocene 22: 14611471.
Giraudi, C., M. Magny, G. Zanchetta, and R.N. Drysdale. 2011. The Holocene climatic
evolution of Mediterranean Italy: A review of the continental geological data.
The Holocene 21(1): 105115.
Gktrk, O.M., D. Fleitmann, S. Badertscher, H. Cheng, R.L. Edwards, M. Leuen-
berger, A. Fankhauser, O. Tysz, and J. Kramers. 2011. Climate on the southern
Black Sea coast during the Holocene: implications from the Sofular Cave record.
bibliography 293
Goudie, A.S. 2006. The Human Impact on the Natural Environment. 6th edition.
Oxford.
Goudsblom, J. 1992. Fire and Civilization. London.
Graser, E.R. 1959. The Edict of Diocletian on Maximum Prices (Appendix). In An
Economic Survey of Ancient Rome. Rome and Italy of the Empire, Vol. 5, edited by
T. Frank, 305422. Paterson, NJ.
Grslund, B., and N. Price. 2012. Twilight of the gods? The dust veil event of AD 536
in critical perspective. Antiquity 86.322: 428443.
Gray, L.J., J. Beer, M. Geller, J.D. Haigh, M. Lockwood, K. Matthes, U. Cubasch,
D. Fleitmann, G. Harrison, L. Hood, J. Luterbacher, G.A. Meehl, D. Shindell, B. van
Geel, and W. White. 2010. Solar Influences on Climate. Reviews of Geophysics 48:
RG4001.
Greco, G. 2003. Le nuove ricerche nel quartiere meridionale. In Velia, le nuove
ricerche, Atti del convegno di studi, Napoli, 14 dicembre 2001. Quaderni del Centro
Studi Magna Grecia 1, Pozzuoli.
Green, R.P.H. 1978. The minence grise of Ausonius Moselle. Res publica litterarum
1: 8994.
1991. Decimus Magnus Ausonius: The works of Ausonius. Oxford.
1997. On a recent redating of Ausonius Moselle. Historia 46: 214226.
Greenberg, J. 2003. Plagued by doubt: reconsidering the impact of a mortality crisis
in the 2nd c. A.D. JRA 16: 413425.
Greene, K. 2000. Technological Innovation and Economic Progress in the Ancient
World: M.I. Finley reconsidered. Economic History Review 53: 2959.
1986. The Achaeology of the Roman Economy. Berkeley-Los Angeles.
Griggs, C.G., A. DeGaetano, P.I. Kuniholm, and M.W. Newton. 2007. A regional
high-frequency reconstruction of MayJune precipitation in the north Aegean
from oak tree rings, A.D.10891989. International Journal of Climatology 27: 1075
1089.
Griggs, C.B., P.I. Kuniholm, M.W. Newton, J.D. Watkins, and S.W. Manning. 2009.
A 924-year Regional Oak Tree-ring Chronology for North Central Turkey. In
Tree-Rings, Kings and Old World Archaeology and Environment: Papers Presented
in Honor of Peter Ian Kuniholm, edited by S.W. Manning and M.J. Bruce, 7179.
Oxford.
Grinter, P. 2007. Seeds of destruction: conflagration in the grain stores of Dichin. In
The transition to late antiquity: on the Danube and beyond, edited by A.G. Poulter,
281285. Oxford.
Grissino-Mayer, H. 1996. A 2129-year reconstruction of precipitation for northwest-
ern New Mexico, U.S.A. In Tree Rings, Environment and Humanity, edited by
J.S. Dean, D.M. Meko, and T.W. Swetnam, 191204. Tucson, AZ. Data are archived
at the World Data Center for Paleoclimatology, Boulder, CO.
Grocock, C., and S. Grainger, eds and trans. 2006. The Texts. In Apicius. A Critical
Edition with an Introduction and an English Translation of the Latin Recipe Text
Apicius, 125326. Blackawton, Devon.
Grove, A.T. 2001. The Little Ice Age and its geomorphological consequences in
Mediterranean Europe. Climatic Change 48: 121136.
Grove, A.T., and O. Rackham. 2001. The Nature of Mediterranean Europe: an Ecologi-
cal History. New Haven; London.
294 bibliography
Grove, J.M. 1988. The Little Ice Age. London.

Grudd, H., K.R. Briffa, W. Karln, T.S. Bartholin, P.D. Jones, and B. Kromer. 2002. A
7400-year tree-ring chronology in northern Swedish Lapland: natural climatic
variability expressed on annual to millennial timescales. The Holocene 12: 657
665.
Grn, R. 2001. Trapped charge dating (ESR, TL, OSL). In Handbook of Archaeologi-
cal Sciences, edited by D.R. Brothwell and A.M. Pollard, 4762. Chichester.
Gsell, S. 1921. Histoire ancienne de lAfrique du Nord, vol. 1. Paris.
Guibal, F., and P. Pomey. 2009. Ancient Shipwrecks, Naval Architecture and Den-
drochronology in the Western Mediterranean. In Between the Seas: Transfer and
Exchange in Nautical Technology. Proceedings of the 11th International Sympo-
sium on Boat and Ship Archaeology, edited by R. Bockius, 219226. Mainz am
Rhein.
Guidoboni, E. 1989. I Terremoti prima del Mille in Italia e nellarea mediterranea
(Storia, Archeologia, Sismologia). Bologna.
Guidoboni, E., A. Comastri, and G. Traina. 1994. Catalogue of ancient earthquakes in
the Mediterranean area up to the 10th century, vol. 1. Rome.
Guiot, J., C. Corona, and ESCARSEL members. 2010. Growing Season Temperatures
in Europe and Climate Forcings Over the Past 1400 Years. PLoS ONE 5(4): e9972.
Gunn, J.D., ed. 2000. The years without summer. Tracing A.D. 536 and its aftermath.
(BAR International Series 872). Oxford.
Haas, J. 2006. Die Umweltkrise des 3. Jahrhundert n. Chr. im Nordwesten des Imperium
Romanum: interdisziplinre Studien zu einem Aspekt der allgemeinen Reichskrise
im Bereich der beiden Germaniae sowie der Belgica und der Raetia. Geographica
historica 22. Stuttgart.
Haas, J.N., I. Richoz, W. Tinner, and L. Wick 1998. Synchronous Holocene climatic
oscillations recorded on the Swiss Plateau and at timberline in the Alps. The
Holocene 8: 301309.
Hgg, R., ed. 1983. The Greek Renaissance of the Eighth Century BC: tradition and
innovation. Stockholm.
Hall, I.R., G.G. Bianchi, and J.R. Evans. 2004. Centennial to millennial scale Holo-
cene climate-deep water linkage in the North Atlantic. Quaternary Science Re-
views 23: 15291536.
Haller, J.S., Jr. 1993. Ergotism. The Cambridge World History of Human Disease.
Cambridge.
Halsall, G. 2007. Barbarian Migrations and the Roman West, 376568. Cambridge.
Halstead, P., and C. Frederick, eds. 2000. Landscape and Land Use in Postglacial
Greece. Sheffield.
Hammond, N. 2010. Climate, Crisis, Collapse, and the Ancient Maya Civilization:
an Enduring Debate. In Mainwaring et al. 2010, 189196.
Harris, W.V. 2000. Trade. In Cambridge Ancient History XI (second ed.), 710739.
Cambridge.
ed. 2005a. Rethinking the Mediterranean. Oxford.
2005b. The Mediterranean and Ancient History. In Harris 2005a, 142.
2011a. Bois et dboisement dans la Mditerrane antique. Annales HSS 66,
1: 105140.
2011b. Plato and the Deforestation of Attica. Athenaeum 99: 479482.
bibliography 295
2012. The Great Pestilence and the Complexities of the Antonine-Severan

Economy. In Limpatto della peste antonina, edited by E. Lo Cascio, 331338.
Bari.
forthcoming. The Indispensable Commodity: Notes on the Economy of
Wood in the Roman Mediterranean.
Harris, W.V., and K. Iara, eds. 2011. Maritime Technology in the Ancient Economy: Ship-
Design and Navigation. Journal of Roman Archaeology, Supplementary Series
Number 84. Portsmouth, R.I.
Hart, S. 1986. Some preliminary thoughts on settlement in southern Edom. Levant
18: 5158.
Hass, H.C. 1996. Northern Europe climate variations during late Holocene: evidence
from marine Skagerrak. Palaeogeography Palaeoclimatology Palaeoecology 123:
121145.
Hassan, F.A. 1981. Historical Nile floods and their implications for climatic change.
Science 212: 11421144.
2007. Extreme Nile floods and famines in Medieval Egypt (AD 9301500) and
their climatic implications. Quaternary International 173174: 101112.
Haury, J., and G. Wirth, ed. 1963 Procopius, Opera Omnia (revised edn), 3 vols. Leipzig.
Healy, J.F. 1978. Mining and Metallurgy in the Greek and Roman World. London.
Heather, P. 1998. Goths and Huns, c. 320425. In Cambridge Ancient History XIII,
487515. Cambridge.
2000. The Western Empire, 425476. In Cambridge Ancient History XIV, 132.
Cambridge.
Hegerl, G., J. Luterbacher, F.J. Gonzlez-Rouco, S. Tett, T. Crowley, and E. Xoplaki.
2011. Influence of human and natural forcing on European seasonal tempera-
tures. Nature Geoscience 4: 99103.
Heikkil, U., J. Beer, J.A. Abreu, and F. Steinhilber. 2011. On the Atmospheric Trans-
port and Deposition of the Cosmogenic Radionuclides (10Be): A Review. Space
Science Reviews.
Heim, C., N.R. Nowaczyk, J.F.W. Negendak, S.A.G. Leroy, and Z. Ben-Avraham. 1997.
Near East desertification: evidence from the Dead Sea. Naturwissenschaften 84:
398401.
Henning, J. 1987. Sdosteuropa zwischen Antike und Mittelalter: archologische Bei-
trge zur Landwirtschaft des 1. Jahrtausends u. Z. Schriften zur Ur- und Frh-
geschichte 42. Berlin.
2009. Revolution or relapse? Technology, agriculture and early medieval
archaeology in Germanic Central Europe. In The Langobards before the Frankish
conquest: an ethnographic perspective, edited by G. Ausenda, P. Delogu, and
C. Wickham, 149173. Woodbridge, UK.
Herman, I.P. 2007. Physics of the Human Body. Berlin.
Heurgon, J., ed. 1978. Varron, conomie rurale I. Paris.
Hirschfeld, Y. 1992. The Judean Desert Monasteries in the Byzantine Period. New
Haven.
2004. A climatic change in the Early Byzantine period? Some archaeological
evidence. Palestine Exploration Quarterly 136: 133149.
Hitchner, R.B. 1988. The Kasserine archaeological survey, 19821986. Antiquits
africaines 24: 741.
296 bibliography
1989. The organisation of rural settlement in the Cillium-Thelepte region.

In LAfrica Romana 6. Atti del VI convegno di studio, Sassari, 1618 dicembre 1988,
vol. 1, edited by A. Mastino, 387402. Sassari.
1990. The Kasserine archaeological survey1987. Antiquits africaines 26:
231259.
1992. The Kasserine Archaeological Survey 19821985. Africa. Fouilles, mon-
uments et collections archologiques en Tunisie 1112: 158198.
1995a. Historical text and archaeological context in Roman North Africa: the
Albertini Tablets and the Kasserine Survey. In Methods in the Mediterranean.
Historical and archaeological views on texts and archaeology, edited by D.B. Small,
124142. Leiden.
1995b. Irrigation, terraces, dams and aqueducts in the region of Cillium (mod.
Kasserine). The role of water works in the development of a Roman-African
town and its countryside. In Productions et exportations africaines: actualits
archologiques en Afrique du Nord antique et medivale. VI e colloque international
sur lhistoire et larchologie de lAfrique du Nord (PAU, octobre 1993118e congrs),
edited by P. Trousset, 143157. Paris.
Holzhauser, H., M. Magny, and H.J. Zumbhl. 2005. Glacier and lake-level vari-
ations in west-central Europe over the last 3500 years. The Holocene 15: 789
801.
Hong, S., J.P. Candelone, and C.F. Boutron. 1994a. Greenland ice history of the
pollution of the atmosphere of the northern hemisphere for lead during the last
three millenia. Analusis 22: M38M40.
Hong, S., J.P. Candelone, C.C. Patterson, and C.F. Boutron. 1994b. Greenland ice
evidence of hemispheric lead pollution two millennia ago by Greek and Roman
civilizations. Science 265: 18411843.
Hong, S., J.P. Candelone, C.C. Patterson, and C.F. Boutron. 1996a. History of ancient
copper smelting pollution during Roman and medieval times recorded in Green-
land ice. Science 272: 246249.
Hong, S., J.P. Candelone, M. Soutif, and C.F. Boutron. 1996b. A reconstruction of
changes in copper production and copper emissions to the atmosphere during
the past 7000 years. The Science of Total Environment 188: 183193.
Horden, P., and N. Purcell. 2000. The Corrupting Sea: a Study of Mediterranean
History. Oxford.
Four Years of Corruption: a Response to Critics. In Harris 2005a, 348375.
Hostetter, E., B.W. Fouke, and C.C. Lundstrom. Forthcoming. The Last Flow of
Water to the Baths of Caracalla: Age, Temperature and Chemistry. Acts of the
Terme di Caracalla conference, 12 March 2007, Soprintendenza Archeologica di
Roma/Istituto Storico Austriaco.
Hoyt, D.V., and K.H. Schatten. 1998a. Group sunspot numbers: a new solar activity
indicator. Part 1. Solar Physics 181: 189219.
1998b. Group sunspot numbers: a new solar activity indicator. Part 2. Solar
Physics 181: 491512.
Hsiang, S.M., K.C. Meng, and M.A. Cane. 2011. Civil Conflicts Are Associated with
the Global Climate. Nature 476: 438441.
Hughes, J.D. 1983. How the Ancients Viewed Deforestation. JFA 10: 437445.
bibliography 297
1994. Pans travail: environmental problems of the ancient Greeks and Romans.
Baltimore.
2011. Ancient Deforestation Revisited. Journal of the History of Biology 44:
4357.
Hughes, M.K. 2002. Dendrochronology in climatologythe state of the art. Den-
drochronologia 20: 95116.
Hughes, M.K., P.I. Kuniholm, J. Eischeid, G. Garfin, C.B. Griggs, and C. Latini. 2001.
Aegean Tree-Ring Signatures Explained. Tree-Ring Research 57: 6773.
Hughes, M.K., T.W. Swetnam, and H.F. Diaz. 2011. Dendroclimatology: Progress and
Prospects. Developments in Paleoenvironmental Research, Vol. 11. Heidelberg.
Humlum, O., J.-E. Solheim, and K. Stordahl. 2011. Identifying natural contributions
to late Holocene climate change. Global and Planetary Change 79: 145156.
Hunt, C.O., S.J. Gale, and D.D. Gilbertson. 1985. The UNESCO Libyan Valleys Sur-
vey IX: Anhydrite and limestone karst in the Tripolitanian pre-desert. LibStud
16: 113.
Hunt, C.O., D.D. Gilbertson, and H.A. El-Rishi. 2007. An 8000-year history of lan-
scape, climate, and copper exploitation in the Middle East: the Wadi Faynan and
the Wadi Dana National Reserve in southern Jordan. Journal of Archaeological
Science 34: 13061338.
Hunt, C.O., D.D. Gilbertson, R.D.S. Jenkinson, M. van der Ween, G. Yates, and P.C.
Buckland. 1987. ULVS XVII: Paleoecology and agriculture of an abandonment
phase at Gasr Mm 10, Wadi Mimoun, Tripolitania. LibStud 18: 113.
Hunt, C.O., D.J. Mattingly, D.D. Gilbertson, J.N. Dore, G.W.W. Barker, J.R. Burns,
A.M. Fleming, and M. van der Ween. 1986. ULVS XIII: Interdisciplinary ap-
proaches to ancient farming in the Wadi Mansur, Tripolitania. LibStud 17: 7
47.
Hunt, C.O., G. Rushworth, D.D. Gilbertson, and D.J. Mattingly. 2001. Romano-Libyan
Dryland Animal Husbandry and Landscape: Pollen and Palynofacies Analyses of
Coprolites from a Farm in the Wadi el-Amud, Tripolitania. Journal of Archaeo-
logical Science 28: 351363.
Huntington, E. 1907. The Pulse of Asia: A Journey in Central Asia Illustrating the
Geographic Basis of History. Boston.
Hurrell, J.W. 1995. Decadal trends in the North Atlantic Oscillation: regional tem-
peratures and precipitation. Science 269: 676679.
1996. Influence of variations in extratropical wintertime teleconnections on
Northern Hemisphere temperature. Geophysical Research Letters 23: 665668.
Hurrell, J.W., and H. van Loon. 1997. Decadal Variations in Climate associated with
the North Atlantic Oscillation. Climatic Change 36: 301326.
Ineson, S., A.A. Scaife, J.R. Knight, J.C. Manners, N.J. Dunstone, L.J. Gray, and J.D.
Haigh. 2011. Solar forcing of winter climate variability in the Northern Hemi-
sphere. Nature Geoscience 4: 753757.
Institute for Energy Research. 2010. A Primer on Energy and the Economy.
Available from http://www.instituteforenergyresearch.org/2010/02/16/a-primer
-on-energy-and-the-economy-energys-large-share-of-the-economy-requires
-caution-in-determining-policies-that-affect-it/.
IPCC. 2007. Climate Change 2007: Synthesis Report. Contribution of Working
Groups I, II and III to the Fourth Assessment Report of the Intergovernmental
298 bibliography
Panel on Climate Change. Core Writing Team, edited by R.K. Pachauri and A. Rei-
singer. Geneva: IPCC. Available from http://www.ipcc.ch/publications_and_
data/publications_ipcc_fourth_assessment_report_synthesis_report.htm.
Issar, A.S. 2003. Climate Changes during the Holocene and their Impact on Hydrolog-
ical Systems. Cambridge.
Issar, A.S., and N. Brown eds. 1998. Water, environment and society in times of climatic
change. Dordrecht.
Issar, A.S., and D. Yakir. 1997. Isotopes from Wood Buried in the Roman Siege Ramp
of Masada: The Roman Periods Colder Climate. Biblical Archaeologist 60 (2):
101106.
Issar, A.S., and M. Zohar. 2007. Climate Change: Environment and History of the Near
East. 2nd edition. Berlin.
Jahns, S. 2005. The Holocene History of Vegetation and Settlement at the Coastal
Site of Lake Voulkaria in Acarnania, Western Greece. Vegetation History and
Jex, C.N., A. Baker, I.J. Fairchild, W.J. Eastwood, M.J. Leng, H.J. Sloane, L. Thomas, and
E. Bekarolu. 2010. Calibration of speleothem 18O with instrumental climate
records from Turkey. Global and Planetary Change 71: 207217.
Ji, J., J. Shen, W. Balsam, J. Chen, L. Liu, and X. Liu. 2005. Asian monsoon oscillations
in the northeastern Qinghai-Tibet Plateau since the late glacial as interpreted
from visible reflectance of Qinghai Lake sediments. Earth and Planetary Science
Letters 233: 6170.
Jiang, H., J. Eirksson, M. Schulz, K.-L. Knudsen, and M.-S. Seidenkrantz. 2005.
Evidence for solar forcing of sea-surface temperature on the North Icelandic
Shelf during the late Holocene. Geology 33: 7376.
Jones, A.H.M. 1950. The Aerarium and the Fiscus. JRS 40: 2229.
Jones, A.H.M., J.R. Martindale, and J. Morris. 19711992. The prosopography of the
later Roman Empire. Cambridge.
Jones, M.D., C.N. Roberts, M.J. Leng, and M. Trkes. 2006. A high-resolution late
Holocene lake isotope record from Turkey and links to North Atlantic and mon-
soon climate. Geology 34: 361364.
Jones, P.D., K.R. Briffa, T.J. Osborn, J.M. Lough, T.D. van Ommen, B.M. Vinther,
J. Luterbacher, E.R. Wahl, F.W. Zwiers, M.E. Mann, G.A. Schmidt, C.M. Ammann,
B.M. Buckley, K.M. Cobb, J. Esper, H. Goosse, N. Graham, E. Jansen, T. Kiefer,
C. Kull, M. Kttel, E. Mosley-Thompson, J.T. Overpeck, N. Riedwyl, M. Schulz,
A.W. Tudhope, R. Villalba, H. Wanner, E. Wolff, and E. Xoplaki. 2009. High-
resolution palaeoclimatology of the last millennium: a review of current status
and future prospects. The Holocene 19: 349.
Jongman, W.M. 2007a. Gibbon was Right: the Decline and Fall of the Roman Econ-
omy. In Crises and the Roman Empire. Proceedings of the Seventh Workshop of the
International Network Impact of Empire, edited by O. Hekster, G. de Kleijn, and
D. Slooties, 183199. Leiden.
2007b. The Early Roman Empire: Consumption. In Cambridge Economic
History of the Greco-Roman World, edited by W. Scheidel, I. Morris, and R. Saller,
592618. Cambridge.
Kaal, J., Y. Carrion Marco, E. Asouti, M. Martin Seijo, A. Martinez Cortizas, M. Costa
Casais, and F. Criado Boado. 2011. Long-term Deforestation in NW Spain: linking
bibliography 299
the Holocene Fire History to Vegetation Change and Human Activities. Quater-
nary Science Reviews 30: 161175.
Kander, A., P. Malanima, and P. Warde. 2013. Power to the People. Energy in Europe
over the Last Five Centuries. Princeton.
Kaniewski, D., E. Paulissen, V. De Laet, K. Dossche, and M. Waelkens. 2007. A high
resolution Late Holocene landscape ecological history inferred from an intra-
montane basin in the Western Taurus Mountains, Turkey. Quaternary Science
Reviews 26: 22012218.
Kaniewski, D., E. Paulissen, E. Van Campo, H. Weiss, T. Otto, J. Bretschneider, and
K. Van Lerberghe. 2010. Late secondearly first millennium BC abrupt climate
changes in coastal Syria and their possible significance for the history of the
Eastern Mediterranean. Quaternary Research 74: 207215.
Kaniewski, D., E. Van Campo, E. Paulissen, H. Weiss, J. Bakker, I. Rossignol, and
K. Van Lerberghe. 2011. The medieval climate anomaly and the little Ice Age in
coastal Syria inferred from pollen-derived palaeoclimatic patterns. Global and
Planetary Change 78: 178187.
Kaplan, J.O., K.M. Krumhardt, and N. Zimmermann. 2009. The Prehistoric and
Preindustrial Deforestation of Europe. Quaternary Science Reviews 28: 3016
3034.
Kaufman, D.S., D.P. Scheider, N.P. McKay, C.M. Ammann, R.S. Bradley, K.R. Briffa,
G.H. Miller, B.L. Otto-Bliesner, J.T. Overpeck, and B.M. Vinther. Artic Lakes 2k
Project Members. 2009. Recent Warming Reverses Long-Term Arctic Cooling.
Science 325: 12361239.
Keenan-Jones, D.C. 2010a. The Aqua Augusta and control of water resources in the
Bay of Naples. ASCS 31 [2010] Proceedings, www.classics.uwa.edu.au/ascs31.
2010b. The Aqua Augusta. Regional water supply in Roman and late antique
Campania. Department of Ancient History. Macquarie University. PhD The-
sis.
Keigwin, L.D. 1996. The Little Ice Age and Medieval Warm Period in the Sargasso
Sea. Science 274: 15041508.
Keppie, L.J.F. 1983. Colonisation and Veteran Settlement in Italy, 4714BC. London.
1984. Colonisation and Veteran Settlement in Italy in the first century AD.
PBSR 52: 77114.
Keys, D. 1999. Catastrophe: An Investigation into the Origins of the Modern World.
London.
Knipping, M., M. Mllenhoff, and H. Brckner. 2008. Human Induced Landscape
Changes around Bafa Gl (Western Turkey). Vegetation History and Archaeo-
botany 17: 365380.
Koepke, N. (without date). Anthropometric Decline of the Roman Empire? Re-
gional Differences and Temporal Development of the Quality of Nutrition in the
Roman Provinces of Germania and Raetia from the First Century to the Fourth
Century AD. http://eh.net/XIIICongress/Papers/Koepke.pdf.
Koepke, N., and J. Baten. 2005. Climate and its Impact on the Biological Standard of
Living in North-East, Centre-West and South Europe during the Last 2000 Years.
History of Meteorology 2: 147159.
Koepke, N., and J. Baten. 2008. Agricultural Specialization and Height in Ancient
and Medieval Europe. Explorations in Economic History 45, 2: 127146.
300 bibliography
Kondrashov D., Y. Feliks, and M. Ghil. 2005. Oscillatory modes of extended Nile
River records (A.D. 6221922). Geophysical Research Letters 32: L10702.
Krber-Grohne, U. 1994. Nutzpflanzen in Deutschland: Kulturgeschichte und Biologie.
Stuttgart.
Kse, N., . Akkemik, H.N. Dalfes, and M.S. zeren. 2011. Tree-ring reconstruc-
tions of MayJune precipitation for western Anatolia. Quaternary Research 75:
438450.
Kouki, P. 2006. Environmental change and human history in the Jabal Harun area,
Jordan. Unpublished licentiates dissertation. Institute for Cultural Research,
Department of Archaeology, University of Helsinki. Available in electronic for-
mat: http://urn.fi/URN:NBN:fi-fe20071209.
2009. Archaeological evidence of land tenure in the Petra region, Jordan:
Nabataean-Early Roman to Late Byzantine. Journal of Mediterranean Archaeo-
logy 22.1: 2956.
Kramer, B. 1995. Arborikultur und Holzwirtschaft im griechischen, rmischen und
byzantinischen gypten. Archiv fr Papyrusforschung 41: 217231.
Kromer, B., S.W. Manning, P.I. Kuniholm, M.W. Newton, M. Spurk, and I. Levin.
2001. Regional 14CO2 offsets in the troposphere: magnitude, mechanisms, and
consequences. Science 294: 25292532.
Kron, G. 2000. Roman ley-farming. JRA 13: 277287.
2002. Archeozoological Evidence for the Productivity of Roman Livestock
Farming. Mnstersche Beitrge zur Antiken Handelsgeschichte 21, 2: 5373.
2004. Roman Livestock Farming in Southern Italy: The Case against Environ-
mental Determinism. In Espaces intgrs et gestion des ressources naturelles dans
lEmpire Romain, edited by M. Clavel-Lveque and E. Hermon, 119134. Besanon.
2005. Anthropometry, Physical Anthropology, and the Reconstruction of
Ancient Health, Nutrition, and Living Standards. Historia 54: 6883.
2008. The Much Maligned Peasant. Comparative Perspectives on the Pro-
ductivity of the Small Farmer in Classical Antiquity. In People, Land, and Politics:
Demographic Developments and the Transformation of Roman Italy 300B.C.A.D.
14, edited by S. Northwood and L. De Ligt, 71119. Leiden.
Kuniholm, P.I. 2002. Dendrochronological Investigations at Herculaneum and
Pompeii. In The Natural History of Pompeii, edited by W.F. Jashemski and F.G.
Meyer, 235239. Cambridge.
Lachniet, M.S. 2009. Climatic and environmental controls on speleothem oxygen-
isotope values. Quaternary Science Reviews 28: 412432.
Lamb, H.H. 1995. Climate, history and the modern world. London.
Lamb, H.F., U. Eicher, and V.R. Switsur. 1989. An 18,000 Year Record of Vegetation,
Lake-level, and Climatic Change from Tigalmamine, Middle Atlas, Morocco.
Journal of Biogeography 16: 6574.
Lamy, F., H.W. Arz, G.C. Bond, A. Bahr, and J. Ptzold. 2006. Multicentennial-scale
hydrological changes in the Black Sea and northern Red Sea during the Holocene
and the Arctic/North Atlantic Oscillation. Paleoceanography 21: PA1008.
Lane, B., and B. Bousquet. 1994. Jordan Petra National Park Management Plan.
Main Report. UNESCO.
Larsen, L.B., B.M. Vinther, K.R. Briffa, T.M. Melvin, H.B. Clausen, P.D. Jones, M.-L.
Siggaard-Andersen, C.U. Hammer, M. Eronen, H. Grudd, B.E. Gunnarson, R.M.
bibliography 301
Hantemirov, M.M. Naurzbaev and K. Nicolussi. 2008. New ice core evidence for a
volcanic cause of the A.D.536 Dust Veil. Geophysical Research Letters 35: L04708.
Lavento, M., M. Huotari, H. Jansson, S. Silvonen, and Z.T. Fiema. 2004. Ancient water
management system in the area of Jabal Haroun, Petra. In Men of Dikes and
Canals, Orient archologie Band 13, edited by H.-D. Bienert and J. Hser, 163172.
Berlin.
Lean, J., and D. Rind. 1998. Climate Forcing by Changing Solar Radiation. Journal
of Climate 11: 30693094.
Le Gall, J. 1953. Recherches sur le culte du Tibre. Paris.
Lehoux, D. 2007. Astronomy, weather, and calendars in the ancient world: parapeg-
mata and related texts in classical and Near Eastern societies. Cambridge.
Leitch, V. 2011. Location, location, location: characterizing coastal and inland pro-
duction and distribution of Roman African cooking wares. In Maritime Archae-
ology and Ancient Trade in the Mediterranean, edited by D. Robinson and A. Wil-
son, 169195. Oxford.
Lemcke, G., and M. Sturm. 1997. 18O and Trace Element Measurements as Proxy
for the Reconstruction of Climate Changes at Lake Van (Turkey): Preliminary
Results. In Third Millennium BC Climate Change and Old World Collapse: 653
678, NATO ASI Series I, vol. 49, edited by N. Dalfes, G. Kukla, and H. Weiss. Berlin.
Leney, L., and R.W. Casteel. 1975. Simplified Procedure for Examining Charcoal
Specimens for Identification. Journal of Archaeological Science 2: 153159.
Leng, M.J., M.D. Jones, M. Frogley, W.J. Eastwood, and C.N. Roberts. 2010. Detri-
tal carbonate influences on bulk oxygen and carbon isotope composition of
lacustrine sediments from the Mediterranean. Global and Planetary Change 71:
175182.
Le Roy Ladurie, E. 1971. Times of Feast, Times of Famine, a History of Climate Since the
Year 1000. Trans. B. Bray. Garden City, NY.
Le Roy Ladurie, E. 2004. Histoire humaine et compare du climat. Paris.
Lettieri, P. 1560 [1803]. Discorso dottissimo del Magnifico Ms. Pierro Antonio de Lec-
thiero cittadino, et Tabulario Napolitano circa lanticha pianta, et ampliatione
della Citt di Nap. et delitinerario del acqua che anticamente flueva, et dentro,
et fora la pred. Citt per aquedocti mjrabili quale secondo per pi raggioni ne
dimostra, era il Sebbetho celebrato dagli antichi auttori. Dizionario geografico-
ragionato del Regno di Napoli. L. Giustiniani, 6: 382411. Naples.
Lev-Yadun, S., D.S. Lucas, and M. Weinstein-Evron. 2010. Modeling the demands
for wood by the inhabitants of Masada and for the Roman siege. Journal of Arid
Environments 74: 777785.
Little, L.K., ed. 2006. Plague and the End of Antiquity: the pandemic of 541750.
Cambridge.
Lo Cascio, E., ed. 2012. Limpatto della peste antonina (Pragmateiai 22). Bari.
Lo Cascio, E., and P. Malanima. 2009. GDP in Pre-Modern Agrarian Economies
(11820ad). A Revision of the Estimates. Rivista di Storia Economica, n.s., 25:
391419.
Lo Cascio, E., and P. Malanima. 2008. Mechanical Energy and Water Power in
Europe. A Long Stability? In Vers une gestion intgre de leau dans lEmpire
Romain, edited by E. Hermon, 201208. Rome.
302 bibliography
Lo Cascio, E., and P. Malanima. Forthcoming. Ancient and Pre-Modern Economies.

GDP in the Roman Empire and Early Modern Europe.
Lockwood, M. 2006. What do cosmogenic isotopes tell us about past solar forcing
of climate? Space Science Reviews 125: 95109.
Lockwood, M., C. Bell, T. Woollings, R.G. Harrison, L.J. Gray, and J.D. Haigh. 2010a.
Top-down solar modulation of climate: evidence for centennial-scale change.
Environmental Research Letters 5 (3): 034008.
Lockwood, M., R.G. Harrison, T. Woollings, and S.K. Solanki. 2010b. Are cold winters
in Europe associated with low solar activity? Environmental Research Letters 5:
024001.
Lockwood, M., and M.J. Owens. 2011. Centennial changes in the heliospheric mag-
netic field and open solar flux: the consensus view from geomagnetic data and
cosmogenic isotopes and its implications. Journal of Geophysical Research 116
(A4): A04109.
Lugli, G. 1953. Fontes ad topographiam veteris urbis Romae pertinentes, vol. 2: Libri V
VII. Rome.
Luterbacher, J., and 48 coauthors. 2006. Mediterranean Climate Variability over
the Last Centuries: A Review. In Mediterranean Climate Variability, edited by
P. Lionello, P. Malanotte-Rizzoli, and R. Boscolo, 27148. Amsterdam.
Luterbacher, J., R. Garca-Herrera, S. Akcer-On, R. Allan, M.C. Alvarez-Castro, G. Ben-
ito, J. Booth, U. Bntgen, N. Cagatay, D. Colombaroli, B. Davis, J. Esper, T. Felis,
D. Fleitmann, D. Frank, D. Gallego, E. Garcia-Bustamante, R. Glaser, J.F. Gonzlez-
Rouco, H. Goosse, T. Kiefer, M.G. Macklin, S.W. Manning, P. Montagna, L. New-
man, M.J. Power, V. Rath, P. Ribera, D. Riemann, N. Roberts, S. Silenzi, W. Tinner,
B. Valero-Garces, G. van der Schrier, C. Tzedakis, B. Vannire, S. Vogt, H. Wan-
ner, J.P. Werner, G. Willett, M.H. Williams, E. Xoplaki, C.S. Zerefos, and E. Zorita.
2012. A Review of 2000 Years of Paleoclimatic Evidence in the Mediterranean.
In The Climate of the Mediterranean Region: From the Past to the Future, edited by
P. Lionello. 87185. Amsterdam.
Luterbacher, J., D. Dietrich, E. Xoplaki, M. Grosjean, and H. Wanner. 2004. European
Seasonal and Annual Temperature Variability, Trends, and Extremes Since 1500.
Science 303: 14991503.
Luterbacher, J., D. McCarroll, D. Fleitmann, F.J. Gonzalez-Rouco, E. Zorita, S. Sal-
cedo, and B. Vinther. 2011. The first Euro-Med2k regional workshop: Review of
current knowledge, available data and plans for multproxy integration. PAGES
news 19: 7576.
Luterbacher, J., R. Rickli, E. Xoplaki, C. Tinguely, C. Beck, C. Pfister, and H.
Wanner. 2001. The Late Maunder Minimum (16751715)A Key Period for
Studying Decadal Scale Climatic Change in Europe. Climatic Change 49: 441
462.
Lyons, G., F. Lunny, and H.P. Pollock. 1985. A Procedure for Estimating the Value of
Forest Fuels. Biomass 8:283300.
MacDonald, B. 2001. Relation between paleoclimate and the settlement of southern
Jordan during Nabataean, Roman and Byzantine periods. Amman: Department
of Antiquities, Studies in the History and Archaeology of Jordan 7: 373378.
Maddison, A. 2007. Contours of the World Economy, 12030A.D. Essays in Macro-
Economic History. Oxford.
bibliography 303
Magny, M. 1993. Solar influences on Holocene climatic changes illustrated by cor-

relations between past lake-level fluctuations and the atmospheric 14C record.
Quaternary Research 40: 19.
2004. Holocene climate variability as reflected by mid-European lake-level
fluctuations and its probable impact on prehistoric human settlements. Quater-
nary International 113(1): 6579.
Magny, M., C. Bgeot, J. Guiot, and O. Peyron, O. 2003. Contrasting patterns of
hydrological changes in Europe in response to Holocene climate cooling phases.
Magny, M., J.-L. de Beaulieu, R. Drescher-Schneider, B. Vannire, A.-V. Walter-
Simonnet, Y. Miras, L. Millet, G. Bossuet, O. Peyron, E. Brugiapaglia and A. Ler-
oux. 2007. Holocene climate changes in the central Mediterranean as recorded
by lake-level fluctuations at Lake Accesa (Tuscany, Italy). Quaternary Science
Reviews 26(1314): 17361758.
Magny, M., D. Galop, P. Bellintani, M. Desmet, J. Didier, J.N. Haas, N. Martinelli,
A. Pedrotti, R. Scandolari, A. Stock and B. Vannire. 2009. Late-Holocene cli-
matic variability south of the Alps as recorded by lake-level fluctuations at Lake
Ledro, Trentino, Italy. The Holocene 19(4): 575589.
Magny, M., F. Arnaud, H. Holzhauser, E. Chapron, M. Debret, M. Desmet, A. Leroux,
L. Millet, M. Revel, and B. Vannire. 2010. Solar and proxy-sensitivity imprints
on paleohydrological records for the last millennium in west-central Europe.
Quaternary Research 73: 173179.
Maher, L., E.B. Banning, and M. Chazan. 2011. Oasis or Mirage? Assessing the Role
of Abrupt Climate Change in the Prehistory of the Southern Levant. Cambridge
Archaeological Journal 21: 129.
Mainwaring, A.B., R. Giegengack, and C. Vita-Finzi, eds. 2010. Climate Crises in
Human History. Philadelphia.
aischberger, M. 1999. Tiberis. in E.M. Steinby (ed.), Lexicon Topographicum Urbis
Romae 5, 6973. Rome.
Maiuri, A. 1958. The Phlegraean Fields. Rome.
Makkai L. 1981. Productivit et exploitation des sources dnergie (XIIeXVIIe si-
cle). In Produttivit e tecnologie nei secoli XIIXVII, edited by S. Mariotti, 165181.
Florence.
Malanima, P. 1996. Energia e crescita nellEuropa preindustriale. Roma.
2006. Energy Consumption in Italy in the 19th and 20th Centuries. Naples.
2009. Pre-Modern European Economy. Leiden.
2010. Energy in History. Encyclopaedia of Life Support Systems (Unesco).
2011. The Path Towards the Modern Economy. The Role of Energy. Rivista di
Politica Economica 100: 77106.
forthcoming. Energy. in Oxford Handbook of Economies in the Classical
World, edited by A. Bresson, E. Lo Cascio, and F. Velde. Oxford.
Malouta, M. and A.I. Wilson. 2013. Mechanical irrigation: water-lifting devices in
the archaeological evidence and in the Egyptian papyri. In The Roman agricul-
tural economy: organization, investment, and production (Oxford Studies on the
Roman Economy), edited by A. Bowman and A. Wilson, 273305. Oxford.
Mangini, A., C. Sptl, and P. Verdes. 2005. Reconstruction of temperature in the
304 bibliography
Central Alps during the past 2000 yr from a 18O stalagmite record. Earth and
Planetary Science Letters 235: 741751.
Mango, M. 2011. Byzantine settlement expansion in North Central Syria: the case
of Androna/Andarin. In Le Proche-Orient de Justinien aux Abbassides: Peuple-
ment et dynamiques spatiales (Bibliothque de lAntiquit Tardive 19), edited by
A. Borrut, M. Debi, A. Papaconstantinou, D. Pieri, and J.-P. Sodini, 93122. Turn-
hout.
Manley, G. 1974. Central England Temperatures: monthly means 1659 to 1973.
Quarterly Journal of the Royal Meteorological Society 100: 389405.
Mann, M.E., Z. Zhang, S. Rutherford, R.S. Bradley, M.K. Hughes, D. Shindell, C.
Ammann, G. Faluvegi, and F. Ni. 2009. Global signatures and dynamical origins
of the Little Ice Age and Medieval Climate Anomaly. Science 326: 12561260.
Manning, S.W. 2010. Radiocarbon dating and climate change. In Climate Crises
in Human History, edited by A.B. Mainwaring, R. Giegengack and C. Vita-Finzi,
2559. Philadelphia.
Manning, S.W., C. Bronk Ramsey, W. Kutschera, T. Higham, B. Kromer, P. Steier, and
E.M. Wild. 2006. Chronology for the Aegean Late Bronze Age 17001400B.C.
Science 312: 565569.
Manning, S.W., and B. Kromer. 2011. Radiocarbon dating archaeological samples
in the Eastern Mediterranean, 17301480BC: further exploring the atmospheric
radiocarbon calibration record and the archaeological implications. Archaeom-
etry 53: 413439.
Manning, S.W., and B. Kromer. 2012. Considerations of the Scale of Radiocarbon
Offsets in the East Mediterranean, and Considering a Case for the Latest (Most
Recent) Likely Date for the Santorini Eruption. Radiocarbon 54 (34): 449
474.
Manning, S.W., B. Kromer, C. Bronk Ramsey, C.L. Pearson, S. Talamo, N. Trano, and
J.D. Watkins. 2010. 14C Record and Wiggle-Match Placement for the Anatolian
(Gordion Area) Juniper Tree-Ring Chronology ~1729 to 751 Cal bc, and typi-
cal Aegean/Anatolian (growing season related) regional 14C offset assessment.
Radiocarbon 52: 15711597.
Manning, S.W., C.L. Pearson, C.B. Griggs, and B. Kromer. 2012. Dendro-wiggle-
match placement of an oak tree-ring chronology from mid-first millennium
AD Constantinople. Antiquity 86, Issue 331: http://antiquity.ac.uk/projgall/
manning331/.
Marchetti, D. 1892. Frammento di un antico pilastro per misurare le acque del
Tevere, ed altre notizie topografiche. Bullettino della Commissione archeologica
comunale di Roma 20: 139149.
Marguerie, D. 2011. Short tree ring series: the study materials of the dendro-
anthracologist. In 5th International Meeting of Charcoal Analysis. In The 5th
International Meeting of Charcoal Analysis: The Charcoal as Cultural and Biolog-
ical Heritage, edited by E. Badal, Y. Carrin, E. Grau, M. Macas and M. Ntinou,
1516. Valencia.
Marguerie, D., and J.-Y. Hunot. 2007. Charcoal Analysis and Dendrology: Data from
Archaeological Sites in North-Western France. Journal of Archaeological Science
34: 14171433.
Martn-Chivelet, J., M.B. Muoz-Garca, R.L. Edwards, M.J. Turrero, and A.I. Ortega.
bibliography 305
2011. Land surface temperature changes in Northern Iberia since 4000 yr bp,
based on 13C of speleothems. Global and Planetary Change 77: 112.
Martinez-Cortizas, A., X. Pontevedra-Pombal, J.C. Nvoa-Muoz, and E. Garca-
Rodeja. 1997. Four thousand years of atmospheric Pb, Cd and Zn deposition
recorded by the ombrotrophic peat bog of Penido Vello (northwest Spain).
Water, Air, & Soil Pollution 100: 387403.
Martnez-Cortizas, A., X. Pontevedra-Pombal, E. Garca-Rodeja, J.C. Nvoa-Muoz,
and W. Shotyk. 1999. Mercury in a Spanish Peat Bog: Archive of Climate Change
and Atmospheric Metal Deposition. Science 284: 939942.
Martn-Puertas, C., B.L. Valero-Garcs, A. Brauer, M. Pilar Mata, A. Delgado-Huertas,
and P. Dulski. 2009. The Iberian-Roman Humid Period (26001600 cal yr bp) in
the Zoar Lake varve record (Andaluca, southern Spain). Quaternary Research
71: 108120.
Martn-Puertas, C., B.L. Valero-Garcs, M.P. Mata, P. Gonzlez-Sampriz, R. Bao,
A. Moreno, and V. Stefanova. 2008. Arid and humid phases in the southern Spain
during the last 4000 years: the Zoar Lake record, Crdoba. The Holocene 18:
907921.
Marzano, A. 2013. Harvesting the sea: the exploitation of marine resources in the
Roman Mediterranean (Oxford Studies on the Roman Economy). Oxford.
Masarik, J., and J. Beer. 1999. Simulation of particle fluxes and cosmogenic nuclide
production in the Earths atmosphere. Journal of Geophysical Research 104:
1209912111.
Masse, G., S.J. Rowland, M.A. Sicre, J. Jacob, E. Jansen, and S.T. Belt. 2008. Abrupt
climate changes for Iceland during the last millennium: Evidence from high-
resolution sea ice reconstructions. Earth and Planetary Science Letters 269: 565
569.
Mathers, C., and S. Stoddart, eds. 1994. Development and Decline in the Mediter-
ranean Bronze Age, Sheffield Archaeological Monographs 8. Sheffield.
Mattingly, D.J., ed. 2003. The Archaeology of Fazzan, vol. 1: Synthesis. London.
Mattingly, D.J., and A.I. Wilson. 2003. Farming the Sahara: the Garamantian contri-
bution in Southern Libya. In Arid lands in Roman times (Arid Zone Archaeology,
Monographs), edited by M. Liverani, 3750. Firenze.
Mattingly, D.J., and A.I. Wilson. 2010. Concluding thoughts: Made in Fazzan? In
The Archaeology of Fazzan, vol. 3, edited by D.J. Mattingly, 523530. London.
Mayewski, P.A., E.E. Rohling, J.C. Stager, W. Karlen, K.A. Maascha, L.D. Meekler,
E.A. Meyerson, F. Gasse, S. van Kreveld, K. Holmgren, J. Lee-Thorp, G. Rosqvist,
F. Rack, M. Staubwasser, R.R. Schneider, and E.J. Steig. 2004. Holocene climate
variability. Quaternary Research 62: 243255.
McCarthy, J.J., O.F. Canziani, N.A. Leary, D.J. Dokken, and K.S. White, eds. 2001.
Climate Change 2001: Impacts, Adaptation and Vulnerability. Cambridge.
McCormick, M. 1990. Eternal Victory. Triumphal rulership in late antiquity, Byzan-
tium, and the early medieval West. Cambridge.
2001. Origins of the European economy. Communications and commerce AD
300900. Cambridge.
2012. Movements and markets in the first millennium: information, contain-
ers, and shipwrecks. In Trade and Markets in Byzantium, edited by C. Morrisson,
5198. Washington DC.
306 bibliography
McCormick, M., P.E. Dutton, and P.A. Mayewski. 2007. Volcanoes and the Climate
Forcing of Carolingian Europe, A.D. 750950. Speculum 82: 865895.
McCormick, M., G. Huang, K.L. Gibson, and M. Polk, eds. 2010. Digital Atlas of Roman
and Medieval Civilizations: http://darmc.harvard.edu/.
McCormick, M., U. Bntgen, M. Cane, E. Cook, K. Harper, P. Huybers, T. Litt, S.W.
Manning, P.A. Mayewski, A.M. More, K. Nicolussi, and W. Tegel. 2012. Climate
Change under the Roman Empire and its Successors, 100bc800ad. A First Syn-
thesis Based on Multi-Proxy Natural Scientific and Historical Evidence. Journal
of Interdisciplinary History 43.2: 169220.
McCormick, M., K. Harper, A.M. More and K. Gibson. 2012. Geodatabase of his-
torical evidence on Roman and post-Roman climate. DARMC Scholarly Data
Series, Data Contribution Series # 20121, DARMC, Center for Geographic Analy-
sis, Harvard University, Cambridge MA 01238: http://darmc.harvard.edu/icb/icb
.do?keyword=k40248&pageid=icb.page496495.
McDermott, F. 2004. Palaeo-climate reconstruction from stable isotope variations
in speleothems: a review. Quaternary Science Reviews 23: 901918.
McParland, L.C., M.E. Collinson, A.C. Scott, and G. Campbell. 2009. The use of
reflectance for the interpretation of natural and anthropogenic charcoal assem-
blages. Archaeological and Anthropological Sciences 1: 249261.
McParland, L.C., M.E. Collinson, A.C. Scott, G. Campbell, and R. Veal. 2010. Is
Vitrification in Charcoal a Result of High Temperature Burning of Wood? Journal
of Archaeological Science 37: 26792687.
McParland, L.C., Z. Hazell, G. Campbell, M.E. Collinson, and A.C. Scott. 2009. How
the Romans got Themselves into Hot Water, Temperatures and Fuel Types used
in Firing a Hypocaust. Environmental Archaeology 14: 172179.
Meiggs, R. 1982. Trees and Timber in the Ancient Mediterranean World. Oxford.
Meteorological Department of the Hashemite Kingdom of Jordan. 2004. Tables:
Total rainfall, Mean monthly and yearly temperature, Absolute maximum air
temperature, Absolute minimum air temperature. http://www.jmd.gov.jo/ .pdf
documents accessed 24.10.2004; no longer available in the Internet. Documents
in possession of P. Kouki.
Meyer, W. 1995. Burgen, Pfalzen, Herrensitze. In Mittelalterarchologie in Zen-
traleuropa: Zum Wandel der Aufgaben und Zielsetzungen, edited by G.P. Fehring
and W. Sage, 2736. Cologne.
Mighall, T.M., A. Martinez Cortizas, H. Biester, and S. Turner. 2006. Proxy Climate
and Vegetation Changes during the Last Five Millennia in NW Iberia: Pollen and
Non-Pollen Palynomorph Data from Two Ombrotrophic Peat Bogs in the North
Western Iberian Peninsula. Review of Palaeobotany and Palynology 141: 203223.
Mighall, T.M., S. Timberlake, I.D.L. Foster, E. Krupp, and S. Singh. 2009. Ancient
copper and lead pollution records from a raised bog complex in Central Wales,
UK. Journal of Archaeological Science 36: 15041515.
Migowski, C., M. Stein, S. Prasad, J.F.W. Negendank, and A. Agnon. 2006. Holocene
climate variability and cultural evolution in the Near East from the Dead Sea
sedimentary record. Quaternary Research 66: 421431.
Mikesell, M.W. 1969. The Deforestation of Mount Lebanon. Geographical Review
59: 128.
Miller, G.H., A. Geirsdottir, Y. Zhong, D.J. Larsen, B.L. Otto-Bliesner, M.M. Holland,
bibliography 307
D.A. Bailey, K.A. Refsnider, S.J. Lehman, J.R. Southon, C. Anderson, H. Bjrnsson,
and T. Thordarson. 2012. Abrupt onset of the Little Ice Age triggered by volcan-
ism and sustained by sea-ice/ocean feedbacks. Geophysical Research Letters 39:
L02708.
Mithen, S., and E. Black, eds. 2011. Water, Life and Civilisation: Climate, Environment
and Society in the Jordan Valley. Cambridge.
Moe, D., F.G. Fedele, A.E. Maude, and M. Kvamme. 2007. Vegetational Changes and
Human Presence in the Low-alpine and Subalpine Zone in Val Febbraro, Upper
Valle di Spluga (Italian Central Alps), from the Neolithic to the Roman Period.
Mokyr, J. 1990. The Lever of Riches. Technological Creativity and Economic Progress.
New York; Oxford.
Molle, F., P.P. Mollinga, and P. Wester. 2009. Hydraulic bureaucracies and the hy-
draulic mission: Flows of water, flows of power. Water Alternatives 2(3): 328
349.
Mondin, L. 2003. La data di pubblicazione della Mosella. In Cavarzere 2003, 189
218.
Monteix, N. 2009. Pompi, recherches sur les boulangeries de lItalie romaine. Fasti
Online Documents & Research: The Journal of Fasti Online. Available from: http://
eprints.bice.rm.cnr.it/1129/1/FOLDER-it-2009--168.pdf.
Mook, W.G., and J. van der Plicht. 1999. Reporting 14C Activities and Concentra-
tions. Radiocarbon 41(3): 227239.
Morris, I. 1998. Archaeology and archaic Greek history. In Archaic Greece: New
Approaches and New Evidence, edited by N. Fisher and H. van Wees, 191. London.
2010a. Social Development. Available in www.ianmorris.org.
2010b. Why the West Rulesfor now. The Patterns of History, and what they
reveal about the Future. New York.
Morris, I., R. Saller, and W. Scheidel. 2007. Introduction. In The Cambridge Eco-
nomic History of the Greco-Roman Worlds, edited by W. Scheidel, I. Morris, and
R. Saller, 112. Cambridge.
Motta, R. 1986a. Lacquedotto Felice. In Il trionfo dellacqua: Acque e acquedotti a
Roma IV Sec. a.C.XX Sec., edited by G. Pisani Sartorio and A. Liberati Silverio,
220225. Roma.
1986b. La decadenza degli acquedotti antichi e la conduzione dellacqua
Mariana. In Il trionfo dellacqua: Acque e acquedotti a Roma IV Sec. a.C.XX Sec.,
edited by G. Pisani Sartorio and A. Liberati Silverio, 203205. Roma.
Mller, J., K. Meisen, H. Dittmeier, and M. Zender, eds. 19281971. Rheinisches Wrter-
buch, 9 vols. Bonn.
Munro, J.H. 2003. Industrial Energy from Water-Mills in the European Economy,
5th to 18th Centuries: the Limitations of Power. In Economia e energia Secc.
XIIIXVIII, Istituto Internazionale di Storia economica F. Datini, edited by
S. Cavaciocchi, 223269. Firenze.
Muscheler, R. 2009. 14C and 10Be around 1650 cal BC. In Times Up! Dating the
Minoan eruption of Santorini. Acts of the Minoan Eruption Chronology Workshop,
Sandbjerg November 2007 initiated by Jan Heinemeier & Walter L. Friedrich. Mono-
graphs of the Danish Institute at Athens Volume 10, edited by D.A. Warburton,
275284. Athens.
308 bibliography
Muscheler, R., F. Joos, J. Beer, S. Mller, M. Vonmoos, and I. Snowball. 2007. Solar
activity during the last 1000 yr inferred from radionuclide records. Quaternary
Nakagawa, T., J.-L. de Beaulieu, and H. Kitagawa. 2000. Pollen-Derived History of
Timber Exploitation from the Roman Period Onwards in the Romanche Valley,
Central French Alps. Vegetation History and Archaeobotany 9: 8589.
Naples Water Works Company Ltd. 1885. Acquedotto di Napoli. Padova.
Neff, U., S.J. Burns, A. Mangini, M. Mudelsee, D. Fleitmann, and A. Mater. 2001.
Strong coherence between solar variability and the monsoon in Oman between
9 and 6 kyr ago. Nature 411: 290293.
Nenninger, M. 2001. Die Rmer und der Wald: Untersuchungen zum Umgang mit
einem Naturraum am Beispiel der rmischen Nordwestprovinzen. Stuttgart.
Neumann, J., 1985. Climate change as a topic in the classical Greek and Roman
literature. Climate Change 7: 441454.
Neumann, F., C. Schlzel, T. Litt, A. Hense, and M. Stein. 2007. Holocene Vegetation
and Climate History of the Northern Golan Heights (Near East). Vegetation
History and Archaeobotany 16: 329346.
Newson, P., G. Barker, P. Daly, D. Mattingly, and D. Gilbertson. 2007. The Wadi
Faynan field systems. In Archaeology and Desertification: The Wadi Faynan Land-
scape Survey, Southern Jordan, edited by G. Barker, D. Gilbertson and D. Mattingly,
141174. Oxford.
Nicault, A., Alleaume, S., Brewer, S., Carrer, M., Nola, P. and Guiot, J. 2008. Mediter-
ranean drought fluctuation during the last 500 years based on tree-ring data.
Climate Dynamics 31: 227245.
Nicolussi, K., M. Kaufmann, T.M. Melvin, J. van der Plicht, P. Schieling, and A.
Thurner. 2009. A 9111 year long conifer tree-ring chronology for the European
Alps: a base for environmental and climatic investigations. The Holocene 19:
909920.
Nijboer, A.J., J. van der Plicht, A.M. Bietti Sestieri, and A. De Santis. 1999/2000. A high
chronology for the early Iron Age in central Italy. Palaeohistoria 41/42: 163176.
Nur, A. 2010. Destructive Earthquakes in Alexandria and Aboukir Bay? In Alexan-
dria and the North-Western Delta. Joint Conference Proceedings of Alexandria:
City and Harbour (Oxford 2004) and The Trade and Topography of Egypts North-
Western Delta, 8th Century bc to 8th Century ad (Berlin 2006) (Oxford Centre
for Maritime Archaeology Monograph 5), edited by D. Robinson and A. Wilson.
Oxford.
Nyberg, J., Malmgren, B.A., Kuijpers, A. and Winter, A. 2002. A centennial-scale vari-
ability of tropical North Atlantic surface hydrography during the late Holocene.
Palaeogeography, Palaeoclimatology, Palaeoecology 183: 2541.
OCarroll, E., and F. Mitchell. 2011. Methodological approaches towards quantifica-
tion and identification of charcoal samples retrieved from archaeological sites.
In 5th International Meeting of Charcoal Analysis. The charcoal as cultural and
biological heritage, edited by E. Badal, Y. Carrin, E. Grau, M. Macas and M. Nti-
nou, 2526 Valencia.
Ogley, P., D. Pickles, I. Brocklebank and C. Wood. 2010. Energy efficiency in historic
buildings. Insulating thatched roofs. n. pl.
bibliography 309
Oleson, J.P. 1984. Greek and Roman mechanical water-lifting devices: the history of a
technology, (Phoenix Supplementary Volume). Toronto.
2000. Water-lifting. In Handbook of ancient water technology (Technology
and change in history), edited by . Wikander, 207302. Leiden.
Olson, S.D. 1991. Firewood and Charcoal in Classical Athens. Hesperia 60: 411420.
Orland, I.J., M. Bar-Matthews, N.T. Kita, A. Ayalon, A. Matthews, and J.W. Valley,
2009. Climate deterioration in the eastern Mediterranean as revealed by ion
microprobe analysis of a speleothem that grew from 2.2 to 0.9 ka in Soreq Cave,
Israel. Quaternary Research 71: 2735.
Ortolani, F., S. Pagliuca, and R.M. Toccaceli. 1991. Osservazioni sullevoluzione geo-
morfologica olocenica nella piana costiera di Velia (Cilento, Campania) sulla
base di nuovi rinvenimenti archeologici. Geografia fisica e dinamica quaternaria
14: 163169.
Ostrow, S.E. 1985. Augustales along the Bay of Naples: A Case for Their Early
Growth. Historia 34: 64101.
Paillet, A. 2005. Archologie de lagriculture moderne. Paris.
Parassoglou, G.M. 1987. Return of uninundated land. Chronique dgypte 62: 205
218.
Pareto, R. 1877. On the Works Proper to Prevent the Inundations of the Tiber in the
City of Rome. Minutes of Proceedings of the Institution of Civil Engineers 49(3):
334338.
Parker, D.E., T.P. Legg, and C.K. Folland. 1992. A new daily Central England temper-
ature series, 17721991. International Journal of Climatology 12: 317342.
Parmesan, C., and G. Yohe. 2003. A globally coherent fingerprint of climate change
impacts across natural systems. Nature 421: 3742.
Patterson, W.P., K.A. Dietrich, C. Holmden, and J.T. Andrews. 2010. Two millennia
of North Atlantic seasonality and implications for Norse colonies. Proceedings
of the National Academy of Sciences 107: 53065310.
Pearson, C.L., C.B. Griggs, P.I. Kuniholm, P.W. Brewer, T. Wazny, and L. Canady. 2012.
Dendroarchaeology of the mid-first millennium AD in Constantinople. Journal
of Archaeological Science 39(11): 34023414.
Peltonen-Sainio, P., A. Rajala, H. Knknen and K. Hakala. 2009. Improving farming
systems in Northern European conditions. In Crop Physiology: Applications for
Genetic Improvement and Agronomy, edited by V.O. Sadras and D. Calderini,
7197. Amsterdam.
Perls, C. 1977. Prhistoire du feu. Paris.
Perry, C.A., and K.J. Hsu. 2000. Geophysical, archaeological, and historical evidence
support a solar-output model for climate change. Proceedings of the National
Academy of Sciences 97: 124332438.
Perry, D.A., R. Oren, and S.C. Hart. 2008. Forest Ecosystems, 2nd ed. Baltimore.
Peyron, O., S. Goring, I. Dormoy, U. Kotthoff, J.R. Pross, J.-L. de Beaulieu, R. Drescher-
Schneider, B. Vannire, and M. Magny. 2011. Holocene seasonality changes in the
central Mediterranean region reconstructed from the pollen sequences of Lake
Accesa (Italy) and Tenaghi Philippon (Greece). The Holocene 21(1): 131146.
Piccarreta, M., M. Caldara, D. Capolongo, and F. Boenzi. 2011. Holocene geomorphic
activity related to climatic change and human impact in Basilicata, Southern
Italy. Geomorphology 128(34): 137147.
310 bibliography
Pignatti, S. 1982. Flora dItalia, 3 v. Bologna.

1997. I Boschi dItalia. Sinecologia e biodiversit. Torino.
Pireddu, G. 1990. Lenergia nellanalisi economica. Milan.
Pitts, M., and R. Griffin. 2012. Exploring Health and Social Well-being in Late Roman
Britain. AJA 116: 253276.
Planchais, N. 1982. Palynologie lagunaire de ltang de Mauguio: Paloenviron-
nement vgtal et volution anthropique. Pollen et Spores 24: 93118.
Plunket, P., and G. Uruuela. 2006. Social and cultural consequences of a late
Holocene eruption of Popocatpetl in central Mexico. Quaternary International
151: 1928.
Plunkett, G., and G.T. Swindles. 2008. Determining the Suns influence on Late-
glacial and Holocene climates: a focus on climate response to centennial-scale
solar forcing at 2800 cal. BP. Quaternary Science Reviews 27: 175184.
Poff, N.L., J.D. Allan, M.B. Bain, J.R. Karr, K.L. Prestegaard, B.D. Richter, R.E. Sparks
and J.C. Stromberg. 1997. The Natural Flow Regime. BioScience 47(11): 769784.
Polzer, M.E. 2011. Early Shipbuilding in the Eastern Mediterranean. in The Oxford
Handbook of Maritime Archaeology, edited by A. Catsambis, B. Ford and D.L.
Hamilton, 349378. Oxford.
Pop-Lazic, S.P. 2002. Nekropole rimskog Singidunuma. Singidunum 3: 7100.
Pope, K.O., and T.H. Van Andel. 1984. Late Quaternary alluviation and soil forma-
tion in the southern Argolid: its history, causes and archaeological implications.
Journal of Archaeological Science 11: 281306.
Portillo, M., and R.M. Albert. 2011. Husbandry practices and livestock dung at the
Numidian site of Althiburos (el Mdina, Kef Governorate, northern Tunisia):
the phytolith and spherulite evidence. Journal of Archaeological Science 38.12:
32243233.
Poska, A., and I.A. Pidek. 2010. Pollen Dispersal and Deposition Characteristics of
Abies alba, Fagus sylvatica and Pinus sylvestris, Roztocze Region (SE Poland).
Pothecary, S. 2002. Strabo, the Tiberian Author: Past, Present and Silence in Strabos
Geography. Mnemosyne 55: 387438.
Praux, C. 1963. Dclaration dinondation dficitaire du Brooklyn Museum (P.
Brooklyn gr. 5). Chronique dgypte 38: 117133.
Preiser-Kapeller, J. 2010. Complex historical dynamics of crisis: the case of Byzantium:
Working Paper. Historical Dynamics of Byzantium 4 (Version November 2010):
http://oeaw.academia.edu/JohannesPreiserKapeller/Papers/506625/Complex_
historical_dynamics_of_crisis_the_case_of_Byzantium.
Provincia di Avellino. 2009. 5. Inquadramento fisico-geografico. Piano di Emergen-
ze Provinciale. Piano stralcio rischio meteorologico: crisi idriche, 3262 http://
www.provincia.avellino.it:8180/documents/11528/14908/Relazione_Crisi+
Idriche.pdf.
Purcell, N. 1996. Rome and the management of water: environment, culture and
power. In Human landscapes in classical antiquity: environment and culture,
edited by G. Shipley and J.B. Salmon, 180212. London; New York.
2005. The Ancient Mediterranean: the View from the Customs House. In
Harris 2005a, 200232.
bibliography 311
Py, V. 2006. Mine charcoal deposits: methods and strategies. The Mediaeval Fournel
silver mines in the Hautes-Alpes (France). In Charcoal Analysis: New Analytical
Tools and Methods for Archaeology. Papers from the Table-Ronde held in Basel
2004, BAR International Series 1493, edited by A. Dufraisse, 3546. Oxford.
Qin, X., J. Liu, H. Jia, H. Lu, X. Xia, L. Zhou, G. Mu, Q. Xu, and Y. Jiao. 2012. New
evidence of agricultural activity and environmental change associated with the
ancient Loulan kingdom, China, around 1500 years ago. The Holocene 22: 5361.
Rackham, O. 1982. Land-use and the Native Vegetation of Greece. In Symposia of
the Association for Environmental Archaeology No. 2, 46 September, 1981 (BAR
International Series No. 146), edited by M. Bell and S. Limbrey, 177198. Bristol.
1996. Ecology and Pseudo-ecology: the Example of Ancient Greece. In
Human Landscapes in Classical Antiquity: Environment and Culture, edited by
G. Shipley and J. Salmon, 1643. London; New York.
Rackham, O., and J. Moody. 1996. The making of the Cretan landscape. Manchester.
Railsback, L.B., F. Liang, J.R. Vidal Roman, A. Grandal-dAnglade, M. Vaqueiro Rodr-
guez, L. Santos Fidalgo, D. Fernndez Mosquera, H. Cheng, and R.L. Edwards.
2011. Petrographic and isotopic evidence for Holocene long-term climate change
and shorter-term environmental shifts from a stalagmite from the Serra do Cou-
rel of northwestern Spain, and implications for climatic history across Europe
and the Mediterranean. Palaeogeography, Palaeoclimatology, Palaeoecology 305:
172184.
Rambeau, C., and S. Black. 2011. Palaeoenvironments of the southern Levant 5,000
bp tp present: linking the geological and archaeological records. In Water, Life
and Civilisation: Climate, Environment and Society in the Jordan Valley, edited by
S. Mithen and E. Black, 94104. Cambridge.
Ramrath, A., L. Sadori and J.R.F.W. Negendank. 2000. Sediments from Lago di
Mezzano, central Italy: a record of Lateglacial/Holocene climatic variations and
anthropogenic impact. The Holocene 10(1): 8795.
Rasmusson, E.M., and J.M. Wallace. 1983. Meteorological aspects of the El Nio/
Southern Oscillation. Science 222: 11951202.
Rasulo, M. 2002. Bolla aqueduct: a two-thousand-year lasting service. In Hydraulic
Information Management (Ecology and the Environment 52), edited by C.A. Breb-
bia and W.R. Blain, 369378. Southampton.
Rauh, N.K., R.F. Townsend, M.C. Hoff, M. Dillon, M.W. Doyle, C.A. Ward, R.M.
Rothaus, H. Caner, . Akkemik, L. Wandsnider, F.S. Ozener, and C.D. Dorner.
2009. Life in the Truck Lane: Urban Development in Western Rough Cilicia.
Jahreshefte des sterreichischen Archologischen Institutes in Wien 78: 253312.
Rayner, N.A., D.E. Parker, E.B. Horton, C.K. Folland, L.V. Alexander, D.P. Rowell,
E.C. Kent, and A. Kaplan. 2003. Global analyses of sea surface temperature, sea
ice, and night marine air temperature since the late nineteenth century. Journal
of Geophysical Research 108(D14), 4407.
Redman, C.L. 1999. Human impact on ancient environments. Tucson.
Reger, G. 1994. Regionalism and Change in the Economy of Independent Delos, 314167
BC. Berkeley.
Regione Campania A.G.C. Sviluppo Attivit Settore Primario Settore Foreste Caccia
e Pesca, G. de Filippo, M. Fraissinet and M.F. Caliendo. 2011. Rapporto Ambien-
tale sui possibili impatti ambientali significativi derivanti dallattuazione
312 bibliography
del Piano Faunistico Venatorio, Regione Campania. http://agricoltura.regione

.campania.it/VAS/rapporto_ambientale_2011.pdf.
Reimer P.J., M.G.L. Baillie, E. Bard, A. Bayliss, J.W. Beck, C.J.H. Bertrand, P.G. Black-
well, C.E. Buck, G.S. Burr, K.B. Cutler, P.E. Damon, R.L. Edwards, R.G. Fairbanks,
M. Friedrich, T.P. Guilderson, A.G. Hogg, K.A. Hughen, B. Kromer, G. McCormac,
S. Manning, C. Bronk Ramsey, R.W. Reimer, S. Remmele, J.R. Southon, M. Stu-
iver, S. Talamo, F.W. Taylor, J. van der Plicht, and C.E. Weyhenmeyer. 2004. Int-
Cal04 terrestrial radiocarbon age calibration, 026 cal kyr BP. Radiocarbon 46:
10291058.
Reimer, P.J., M.G.L. Baillie, E. Bard, A. Bayliss, J.W. Beck, P.G. Blackwell, C. Bronk
Ramsey, C.E. Buck, G.S. Burr, R.L. Edwards, M. Friedrich, P.M. Grootes, T.P.
Guilderson, I. Hajdas, T.J. Heaton, A.G. Hogg, K.A. Hughen, K.F. Kaiser, B. Kromer,
G. McCormac, S.W. Manning, R.W. Reimer, D.A. Richards, J.A. Southon, S. Talamo,
C.S.M. Turney, J. van der Plicht and C.E. Weyhenmeyer. 2009. IntCal09 and
Marine09 radiocarbon age calibration curves, 050,000 years cal BP. Radiocar-
bon 59(4): 11111150.
Renberg, I., M.-L. Brnnvall, R. Bindler, and O. Emteryd. 2000. Atmospheric Lead
Pollution History during Four Millennia (2000BC to 2000AD) in Sweden. Ambio
29 (3): 150156.
Renfrew, C. 1972. The Emergence of Civilization. The Cyclades and the Aegean in the
third millennium B.C. London.
Renfrew, J. 2004 [1985]. Roman Cookery. Recipes and History. London.
Revelle, R., and H.E. Suess. 1957. Carbon dioxide exchange between atmosphere
and ocean and the question of an increase of atmospheric CO2 during the past
decades. Tellus 9: 1827.
Reynolds, T.S. 1983. Stronger than a Hundred Men. A History of the Vertical Water
Wheel. Baltimore; London.
Riccio, A. 2002. Lantico acquedotto della Bolla. Lacqua e larchitettura. In Acque-
dotti e fontane del regno di Napoli, edited by F. Starace, 115180. Lecce.
Richter, H.G., D. Grosser, I. Heinz, and P.E. Gasson. 2004. IAWA list of microscopic
features for softwood identification. Leiden.
Richter, B.D., R. Mathews, D.L. Harrison, and R. Wigington. 2003. Ecologically Sus-
tainable Water Management: Managing River Flows for Ecological Integrity. Eco-
logical Applications 13 (1): 206224.
Riera-Mora, S., and A. Esteban-Amat, A. 1994. Vegetation History and Human
Activity during the Last 6000 Years on the Central Catalan Coast (Northeastern
Iberian Peninsula). Vegetation History and Archaeobotany 3: 723.
Roberts, N. 1990. Human-induced Landscape Change in South and Southwest
Turkey during the Later Holocene. In Mans Role in the Shaping of the Eastern
Mediterranean Landscape, edited by S. Bottema, G. Entjes-Nieborg and W. van
Zeist, 5367. Rotterdam and Brookfield, VT.
1998. The Holocene: An Environmental History. Oxford.
Roberts, N., D. Brayshaw, C. Kuzucuolu, R. Perez, and L. Sadori. 2011. The mid-
Holocene climatic transition in the Mediterranean: causes and consequences.
The Holocene 21: 313.
Roberts, N., W.J. Eastwood, C. Kuzucuolu, G. Fiorentino, and V. Caracuta. 2011.
bibliography 313
Climatic, vegetation and cultural change in the eastern Mediterranean during

the mid-Holocene environmental transition. The Holocene 21: 147162.
Roberts, N., M.D. Jones, A. Benkaddour, W.J. Eastwood, M.L. Filippi, M.R. Frog-
ley, H.F. Lamb, M.J. Leng, J.M. Reed, M. Stein, L. Stevens, B. Valero-Garcs, and
G. Zanchetta. 2008. Stable isotope records of Late Quaternary climate and
hydrology from Mediterranean lakes: The ISOMED synthesis. Quaternary Sci-
ence Reviews 27: 24262441.
Roberts, N., A. Moreno, B.L. Valero-Garcs, J.P. Corella, M. Jones, S. Allcock, J. Wood-
bridge, M. Morelln, J. Luterbacher, E. Xoplaki, and M. Trkes. 2012. Palaeolim-
nological evidence for an east-west climate see-saw in the Mediterranean since
AD900. Global and Planetary Change 84: 2334.
Roberts, N., T. Stevenson, B. Davis, R. Cheddadi, S. Brewster, and A. Rosen. 2004.
Holocene Climate, Environment and Cultural Change in the Circum-Medi-
terranean Region. In Past Climate Variability through Europe and Africa, vol-
ume 6, edited by R.W. Battarbee, F. Gasse, and C.E. Stickley, 343362. Dor-
drecht.
Robinson, S.A., S. Black, B.W. Sellwood, and P.J. Valdes. 2006. A review of palaeocli-
mates and palaeoenvironments in the Levant and Eastern Mediterranean from
25,000 to 5000 years bp: setting the environmental background for the evolution
of human civilization. Quaternary Science Reviews 25: 15171541.
Rodgers, R.H., ed. 1975. Palladii Rutilii Tauri Aemiliani viri inlustris Opus agriculturae
de veterinaria medicina de insitione. Leipzig.
Rogers, R.S. 1931. Lucius Arruntius. Classical Philology 26: 3145.
Rohling, E.J., P.A. Mayewski, R.H. Abu Zied, J.S.L. Casford, and A. Hayes. 2002.
Holocene atmosphere-ocean interactions: Records from Greenland and the
Aegean Sea. Climate Dynamics 18: 587594.
Rohling, E.J., and H. Plike. 2005. Centennial-scale climate cooling with a sudden
cold event around 8,200 years ago. Nature 434: 975979.
Ropelewski, C.F., and M.S. Halpert. 1987. Global and regional scale precipitation
patterns associated with the El Nio/Southern Oscillation. Monthly Weather
Review 115: 16061626.
Rosen, A.M. 2007. Civilizing Climate: Social Responses to Climate Change in the
Ancient Near East. Lanham, MD.
Rosen, A.M., and S.A. Rosen. 2001. Determinist or Not Determinist? Climate, Envi-
ronment, and Archaeological Explanation in the Levant. In Studies in the
Archaeology of Israel and Neighboring Lands in Memory of Douglas L. Esse. Stud-
ies in Ancient Oriental Civilization 59, edited by S.R. Wolff, 535549. Chicago.
Roos-Barraclough, F., W.O. van der Knapp, J.F.N. van Leeuwen, and W. Shotyk.
2004. A Late-glacial and Holocene record of climatic change from a Swiss peat
humification profile. The Holocene 13: 719.
Rosman, K.J.R., W. Chisholm, S. Hong, J.P. Candelone, and C.F. Boutron. 1997. Lead
from Carthaginian and Roman Spanish mines isotopically identified in Green-
land ice dated from 600B.C. to 300A.D. Environment, Science and Technology 31:
34133416.
Rossignol, B. 2012. Le climat, les famines et la guerre: lments du contexte de la
Peste Antonine. In Limpatto della peste antonina, edited by E. Lo Cascio, 87122.
Bari.
314 bibliography
Rossignol, B., and S. Durost. 2007. Volcanisme global et variations climatiques de

courte dure dans lhistoire romaine (Ier s. av. J.-C.IVme s. ap. J.-C.): leons dune
archive glaciaire (GISP2). Jahrbuch des rmisch-germanischen Zentralmuseums
Mainz 54: 395438.
Roth, R. 2012/3. Scientific History and Experimental History. Journal of Interdisci-
plinary History 43: 443458.
Rouvillois-Brigol, M. 1985. La steppisation en Tunisie depuis lpoque punique:
dterminisme humain ou climatique? Bulletin archologique du Comit des tra-
vaux historiques et scientifiques 19.B: 215224.
1986. Quelques remarques sur les variations de loccupation du sol dans le
Sud-Est algrien. Histoire et archologie de lAfrique du Nord. Actes du 3e colloque
international, Montpellier, 1985, 3553. Paris.
Ruddiman, W.F. 2003. The Anthropogenic Greenhouse Era Began Thousands of
Years Ago. Climatic Change 61: 261293.
Ruddiman, W.F., M.C. Crucifix, and F.A. Oldfield, eds. 2011. Holocene Special Issue:
The early-Anthropocene hypothesis. The Holocene 21(5): 713879.
Ruddiman, W.F., and E.C. Ellis. 2009. Effects of per-capita Land Use Changes on
Holocene Forest Clearance. Quaternary Science Reviews 28: 30113015.
Rudolf, B., and U. Schneider. 2005. Calculation of gridded precipitation data for
the global land-surface using in-situ gauge observations. Proceedings of the 2nd
Workshop of the International Precipitation Working Group IPWG, Monterey Octo-
ber 2004, EUMETSAT, 231247. Monterey, CA.
Ruello, M.R. 2008. Geoarcheologia in aree costiere della Campania: i siti di Neapolis
ed Elea-Velia. Tesi di dottorato, XX ciclo, Universit di Napoli Federico II. www
.fedoa.it.
Russell, B. 2009. The dynamics of stone transport between the Roman Mediter-
ranean and its hinterland. Facta 2: 107126.
Russo Ermolli, E., and G. di Pasquale. 2002. Vegetation dynamics of South western
Italy in the last 28 kyr inferred from pollen analysis of a Tyrrhenian Sea core.
Sabatier, P., L. Dezileau, C. Colin, L. Briqueu, F. Bouchette, P. Martinez, G. Siani,
O. Raynal, and U. Von Grafenstein. 2012. 7000 years of paleostorm activity in
the NW Mediterranean Sea in response to Holocene climate events. Quaternary
Research 77: 111.
Sadori, L., and M. Giardini. 2008. Environmental History in the Mediterranean
Basin: Microcharcoal as a Tool to Disentangle Human Impact and Climate
Change. In Charcoals from the Past: Cultural and Palaeoenvironmental Implica-
tions (BARIS 1807), edited by G. Fiorentino and D. Magri, 229236. Oxford.
Sallares, R. 1991. The Ecology of the Ancient Greek World. London.
2007. Ecology. In The Cambridge Economic History of the Greco-Roman
World, edited by W. Scheidel, I. Morris and R. Saller, 1537. Cambridge.
Salzer, M.W., and M.K. Hughes. 2007. Bristlecone pine tree rings and volcanic
eruptions over the last 5000 yr. Quaternary Research 67: 5768.
Sangiorni, F., L. Capotondi, N. Combourieu Nebout, L. Vigliotti, H. Brinkhuis, S.
Giunta, A.F. Lotter, C. Morigi, A. Negri, and G.J. Reichart. 2003. Holocene seasur-
face temperature variations in the southern Adriatic Sea inferred from a multi-
proxy approach. Journal of Quaternary Science 18: 723732.
bibliography 315
Santarelli, A. 1882. Forl. NSA 1882: 41.

Scagel, R.F., R.J. Bandoni, G.E. Rouse, W.B. Schofield, J. Stein, and T.M.C. Taylor. 1969.
Plant Diversity: An Evolutionary Approach (The Wadsworth Botany Series), edited
by W.A. Jensen and L.G. Kavaljian. Belmont, CA.
Scheidel, W. 2002. A model of demographic and economic change in Roman Egypt
after the Antonine plague. JRA 15: 97114.
2007. Demography. In The Cambridge Economic History of the Greco-Roman
World, edited by W. Scheidel, I. Morris, R. Saller, 3886. Cambridge.
2009. In search of Roman economic growth. JRA 22: 4670.
Scheidel, W., and S.J. Friesen. 2009. The Size of the Economy and the Distribution
of Income in the Roman Empire. JRS 99: 6191.
Scheidel, W., I. Morris, and R.P. Saller. 2007. The Cambridge Economic History of the
Greco-Roman World. Cambridge.
Schenkel, Y., P. Bertaux, S. Vanwijnbserghe, and J. Carre. 1998. An Evaluation of the
Mound Kiln Carbonization Technique. Biomass and Bioenergy 14: 505516.
Scherillo, G. 1844. Dellaria di Baja a tempo dei Romani e di una meravigliosa spelonca
scoperta nelle vicinanze di Cuma. Naples.
Schiebold, H. 2010. Heizung und Wassererwrmung in rmischen Thermen: histori-
sche EntwicklungNachfolgesystemeneuzeitliche Betrachtungen und Untersu-
chungen. Siegburg.
Schilman, B., M. Bar-Mathews, A. Almogi-Labin, and B. Luz. 2001. Global climate
instability reflected by eastern Mediterranean marine records during the
late Holocene. Palaeogeography, Palaeoclimatology, Palaeoecology 176: 157
176.
Schilman, B., A. Ayalon, M. Bar-Matthews, E.J. Kagan, and A. Almogi-Labin., 2002.
Sealand paleoclimate correlation in the Eastern Mediterranean region during
the late Holocene. Israel Journal of Earth Sciences 51: 181190.
Schmidt, B. 2010. Der Jahrhundert Sommer im Jahr 49 n. Chr. im Rheinland. Klner
Jahrbuch 43: 695699.
Schmidt, B., and W. Gruhle, W. 2003a. Klimaextreme in Rmischer Zeit. Eine Struk-
turanalyse dendrochronologischer Daten. Archologisches Korrespondenzblatt
33: 421426.
2003b. Niederschlagsschwankungen in Westeuropa whrend der letzten
8000 Jahre. Versuch einer Rekonstruktion mit Hilfe eines neuen dendrochro-
nologischen Verfahrens (Grad der Wuchshomogenitt). Archologisches Korre-
spondenzblatt 33: 281300.
Schmidt, B., W. Gruhle, A. Zimmermann, and T. Fischer. 2005. Mgliche Schwan-
kungen von GetreideertrgenBefunde zur rheinischen Linienbandkeramik
und rmischen Kaiserzeit. Archologisches Korrespondenzblatt 35: 301316.
Schmiedt, G., ed. 1972. Il livello antico del Mar Tirreno. Testimonianze dei resti arche-
ologici. Firenze.
Schneider, H. 2007. Technology. In Cambridge Economic History of the Greco-
Roman World, edited by W. Scheidel, I. Morris, R. Saller, 144171. Cambridge.
Schoch, W., I. Heller, F.H. Schweingrber, and F. Kienast. 2004. Wood Anatomy of
Central European Species. In Swiss Federal Research Institute WSL. http://www
.woodanatomy.ch/ (accessed 2004 onwards).
316 bibliography
Schweingrber, F.H. 1990. Anatomie europischer Hlzer: ein Atlas zur Bestimmung
europischer Baum-, Strauch-, und Zwergstrauchhlzer. Bern.
Scott, A.C. 2005. Charcoal Reflectance as a Proxy for the Emplacement Tempera-
ture of Pyroclastic Flow Deposits. Geology 33(July): 589592.
Scott, A.C., and R. Veal. 2010. New Methods and Applications in Roman Charcoal
Studies. In Scienze Naturali e Archeologia, Il paesaggio antico: uomo/ambiente
ed eventi catastrofoci. Proceedings of the Conference, Naples 1416 October, 2010,
edited by M.R. Senatore and A.M. Ciarallo, 203206. Rome.
Seager, R., N. Harnik, W.A. Robinson, Y. Kushnir, M. Ting, H.P. Huang, and J. Vlez.
2005. Mechanisms of ENSO-forcing of hemispherically symmetric precipita-
tion variability. Quarterly Journal of the Royal Meteorological Society 131: 1501
1527.
Seeck, O. 1920. Geschichte des Untergangs der antiken Welt. Stuttgart.
Seidenkrantz, M.-S., S. Aagaard-Srensen, H. Sulsbrck, A. Kuijpers, K.G. Jensen, and
H. Kunzendorf. 2007. Hydrography and climate of the last 4400 years in a SW
Greenland fjord: implication for Labrador Sea palaeoceanography. The Holocene
17: 387401.
Seim, A., K. Treydte, U. Bntgen, J. Esper, P. Fonti, H. Haska, F. Herzig, W. Tegel, and
D. Faust. 2010. Exploring the potential of Pinus heldreichii Christ for long-term
climate reconstruction in Albania. Tree Rings in Archaeology, Climatology and
Ecology 8: 7582.
Seyfarth, W., ed. 1978. Ammiani Marcellini Rerum gestarum libri qui supersunt. Leip-
zig.
Serre, F. 1978. The dendroclimatological value of the European larch (Larix decidua
Mill.) in the French Maritime Alps. Tree-Ring Bulletin 38: 2534.
Sgobbo, I. 1938. Lacquedotto romano della Campania: Fontis Augustei Aquaeduc-
tus. NSA 14: 7597.
Shanzer, D. 1998a. The date and literary context of Ausonius Mosella: Ausonius,
Symmachus, and the Mosella. In Style and Tradition: studies in honor of Wendell
Clausen, edited by P. Knox and C. Foss, 284305. Stuttgart.
1998b. The date and literary context of Ausoniuss Mosella. Historia 47:
204233.
Shao, X., Y. Xu, Z.-Y. Yin, E. Liang, H. Zhu, and S. Wang. 2010. Climatic implications of
a 3585-year tree-ring width chronology from the northeastern Qinghai-Tibetan
Plateau. Quaternary Science Reviews 29: 21112122.
Shaw, B.D. 1981. Climate, environment and history: the case of Roman North Africa.
In Climate and history: studies in past climates and their impact on man, edited by
T.M.L. Wigley, M.J. Ingram, and G. Farmer, 379403. Cambridge.
Sheppard, P., P. Tarasov, L. Graumlich, K. Heussner, M. Wagner, H. Osterle, and
L. Thompson. 2004. Annual precipitation since 515bc reconstructed from living
and fossil juniper growth of Northeast Qinghai Province, China. Climate Dynam-
ics 23: 869881.
Sherratt, A., and S. Sherratt. 1991. From Luxuries to Commodities: The Nature of
Mediterranean Bronze Age Trading Systems. In Bronze Age Trade in the Mediter-
ranean (Studies in Mediterranean Archaeology 90), edited by N.H. Gale, 351386.
Gteborg.
bibliography 317
Shindell, D.T., G.A. Schmidt, M.E. Mann, D. Rind, and A. Waple. 2001. Solar Forcing
of Regional Climate During the Maunder Minimum. Science 294: 21492152.
Shindell, D.T., G.A. Schmidt, R.L. Miller, and M.E. Mann. 2003. Volcanic and solar
forcing of climate change during the preindustrial era. Journal of Climate 16:
4094107.
Shipley, F.W. 1931. Chronology of the Building Operations in Rome from the Death
of Caesar to the Death of Augustus. Memoirs of the American Academy in Rome
9: 760.
Shotyk, W., D. Weiss, P.G. Appleby, A.K. Cheburkin, R. Frei, M. Gloor, J.D. Kramers,
S. Reese, and W.O. Van Der Knapp. 1998. History of Atmospheric Lead Deposi-
tion Since 12,370 14C yr bp from a Peat Bog, Jura Mountains, Switzerland. Science
281: 16351640.
Siegenthaler, U., M. Heimann, and H. Oeschger. 1980. 14C variations caused by
changes in the global carbon cycle. Radiocarbon 22: 177191.
Sigurdsson, H., and S. Carey. 2002. The Eruption of Vesuvius in A.D. 79. In The
Natural History of Pompeii, edited by W.F. Jashemski and F.G. Meyer, 3764.
Cambridge.
Sigurdsson, H., S. Carey, W. Cornell, and T. Pescatore. 1985. The eruption of Vesuvius
in A.D.79. National Geographic Research 1: 332387.
Sim, D. 1998. Beyond the Bloom. Bloom Refining and Iron Artifact Production in the
Roman World, BAR International Series 725. Edited by Isabel Ridge. Oxford.
Sirot, E. 2011. Allumer le feu: Chemine et pole dans la maison noble et le chteau du
XII e au XVI e sicle. Paris.
Sivan, H.S. 1990. Redating Ausonius Moselle. AJPh 111: 383394.
Slim, H., P. Trousset, R. Paskoff, and A. Oueslati. 2004. Le littoral de la Tunisie. tude
goarchologique et historique (tudes dAntiquits africaines). Paris.
Smil, V. 1994. Energy in World History. Boulder, CO; San Francisco; Oxford.
2010. Why America is not a New Rome. Cambridge, MA.
Smith, A.H.V. 1997. Provenance of Coals from Roman Sites in England and Wales.
Britannia 28: 297324.
Smith, A.T. 2003. The Political Landscape: Constellations of Authority in Early Com-
plex Polities. Berkeley.
Smith, W. 2001. Environmental Sampling. In Leptiminus (Lamta) Report No. 2.
The East Baths, Cemeteries, Kilns, Venus Mosaic, Site Museum, and other Studies,
edited by L.M. Stirling, D.J. Mattingly, and N.B. Lazreg. Portsmouth, RI.
Snodgrass, A.M. 1977. Archaeology and the rise of the Greek state. Cambridge.
1980. Archaic Greece; the age of experiment. London.
Sohi, S., E. Lopez-Capel, E. Krull, and R. Bol. 2009. Biochar, climate change and soil:
A review to guide future research. In CSIRO Land and Water Science Report 05/09.
Sydney.
Solanki, S.K., I.G. Usoskin, B. Kromer, M. Schssler, and J. Beer. 2004. Unusual
activity of the Sun during recent decades compared to the previous 11,000 years.
Nature 431: 10841087.
Sommella, P. 1978. Forma e urbanistica di Pozzuoli romana. Naples.
Southon, J.R. 2002. A first step to resolving the GISP2 and GRIP ice-core chronolo-
gies, 014,500 yr BP. Quaternary Research 57: 3237.
318 bibliography
2004. A Radiocarbon Perspective on Greenland Ice-Core Chronologies: can

we use ice cores for 14C calibration? Radiocarbon 46: 12391259.
Spadea, R., ed. 1998. La Campania antica dal pleistocene alleta romana ritrovamenti
archeologici lungo il gasdotto transmediterraneo. Naples.
Speranza, A., B. van Geel, and J. van der Plicht. 2002. Evidence for Solar Forcing
of Climate Change at ca. 850 cal. BC from a Czech Peat Sequence. Global and
Planetary Change 35: 5165.
Sperber, D. 1978. Roman Palestine, 200400the Land: Crisis and Change in Agrarian
Society as Reflected in Rabbinic Sources. Ramat-Gan.
Spurr, M.S. 1986. Arable Cultivation in Roman Italy c. 200B.C.c. A.D.100. London.
Squatriti, P. 2002. Water and Society in Early Medieval Italy, 4001000. Cambridge.
Starr, C.G. 1941. The Roman imperial navy: 31B.C.A.D.324. Ithaca, NY.
Stathakopoulos, D.C. 2000. The Justinianic Plague Revisited. Byzantine and Mod-
ern Greek Studies 24: 256276.
2004. Famine and pestilence in the late Roman and early Byzantine empire:
a systematic survey of subsistence crises and epidemics. Aldershot; Burlington,
VT.
Staubwasser, M., and H. Weiss. 2006. Holocene climate and cultural evolution in
late prehistoricearly historic West Asia. Quaternary Research 66: 372387.
Steckel, R.H. 2009. Height and Human Welfare: Recent Developments and New
Directions. Explorations in Economic History 46: 123.
Steinhilber, F., J. Beer, and C. Frhlich. 2009. Total solar irradiance during the
Holocene. Geophysical Research Letters 36: L19704. doi: 10.1029/2009GL040142.
Stevens, L.R., E. Ito, A. Schwalb, and H.E. Wright. 2006. Timing of atmospheric
precipitation in the Zagros Mountains inferred from a multi-proxy record from
Lake Mirabad, Iran. Quaternary Research 66: 494500.
Stiros, S.C. 2001. The AD 365 Crete earthquake and possible seismic clustering
during the fourth to sixth centuries ad in the Eastern Mediterranean: a review of
historical and archaeological data. Journal of Structural Geology 23.23: 545562.
Stuiver, M., and T.F. Braziunas. 1988. The Solar Component of the Atmospheric
14C Record. In Secular Solar and Geomagnetic Variations in the Last 10,000 Years,
edited by F.R. Stephenson and A.W. Wolfendale, 245266. Dordrecht.

1993. Sun, ocean, climate and atmospheric 14CO2: an evaluation of causal and
spectral relationships. The Holocene 3: 289305.
1998. Anthropogenic and solar components of hemispheric 14C. Geophysical
Research Letters 25: 329332.
Stuiver, M., T.F. Braziunas, B. Becker, and B. Kromer. 1991. Climatic, solar, oceanic,
and geomagnetic influences on late-glacial and Holocene atmospheric 14C/12C
change. Quaternary Research 35: 124.
Stuiver, M., and P.D. Quay. 1980. Changes in atmospheric carbon-14 attributed to a
variable sun. Science 207: 1119.
Stuiver, M., P.J. Reimer, and T.F. Braziunas. 1998. High-precision radiocarbon age
calibration for terrestrial and marine samples. Radiocarbon 40: 11271151.
Suess, H.E. 1955. Radiocarbon Concentration in Modern Wood. Science 122: 415
417.
Swan, P.M. 1987. Cassius Dio on Augustus: A Poverty of Annalistic Sources? Phoenix
41: 272291.
bibliography 319
Swindles, G., G. Plunkett, and H.M. Roe. 2007. A delayed climatic response to solar
forcing at 2800 cal. bp: multiproxy evidence from three Irish peatlands. The
Holocene 17: 177182.
Syme, R. 1981. Vibius Rufus and Vibius Rufinus. ZPE 43: 365376.
Tandy, D.W. 1997. Warriors into Traders: The Power of the Market in Early Greece.
Berkeley; Los Angeles.
Taylor, R.M. 2000. Public needs and private pleasures: water distribution, the Tiber
river and the urban development of ancient Rome. Rome.
Teleles, I.G. 2004. Meterologika phainomena kai klima sto Vyzantio. Athens.
Temin, P. 2006. The economy of the early Roman empire. Journal of Economic
Perspectives 20.1: 133151.
Thry-Parisot, I., L. Chabal, and J. Chrzavzez. 2010. Anthracology and taphonomy,
from wood gathering to charcoal analysis. A review of the taphonomic processes
modifying charcoal assemblages, in archaeological contexts. Palaeogeography,
Palaeoclimatology, Palaeoecology 291: 142153.
Thomas, R.G., and A.I. Wilson. 1994. Water supply for Roman farms in Latium and
South Etruria. PBSR 62: 139196.
Thommen, L. 2012. An Environmental History of Ancient Greece and Rome. Translated
by P. Hill. Cambridge.
Thompson, D.J.W., and J.M. Wallace. 1998. The Arctic Oscillation signature in win-
tertime geopotential height and temperature fields. Geophysical Research Letters
25: 12971300.
Thompson, L.G., E. Mosley-Thompson, M.E. Davis, A. Henderson, H.H. Brecher,
V.S. Zagorodnov, T.A. Mashiotta, P.-N. Lin, V.N. Mikhalenko, D.R. Hardy, and
J. Beer. 2002. Kilimanjaro Ice Core Records: Evidence of Holocene Climate
Change in Tropical Africa. Science 298: 589593.
Thompson, L.G., E. Mosley-Thompson, H. Brecher, M. Davis, B. Len, D. Les, P.-N.
Lin, T. Mashiotta, and K. Mountain. 2006. Abrupt tropical climate change: Past
and present. Proceedings of the National Academy of Sciences 103: 10536543.
Thonemann, P. 2011. The Maeander Valley: a Historical Geography from Antiquity to
Byzantium. Cambridge.
Thornton, M.K. 1986. Julio-Claudian Building Programs: Eat, Drink, and Be Merry.
Historia 35: 2844.
Thornton, M.K., and R.L. Thornton. 1990. The Financial Crisis of A.D. 33: A Keyne-
sian Depression? Journal of Economic History 50(3): 655662.
Tingley, M.P., P.F. Craigmile, M. Haran, B. Li, E. Mannshardt, and B. Rajaratnam.
2012. Piecing together the past: statistical insights into paleoclimatic reconstruc-
tions. Quaternary Science Reviews 35: 122.
Torelli, M. 2006. The Topography and Archaeology of Republican Rome. In A Com-
panion to the Roman Republic, edited by N. Rosenstein and R. Morstein-Marx,
81101. Malden, MA; Oxford.
Touchan, R., and M.K. Hughes. 1999. Dendrochronology in Jordan. Journal of Arid
Environments 42: 291303.
Touchan, R., D.M. Meko, and M.K. Hughes. 1999. A 396-year reconstruction of pre-
cipitation in Southern Jordan. Journal of the American Water Resources Associa-
tion 35: 4555.
Touchan, R., G.M. Garfin, D.M. Meko, G. Funkhouser, N. Erkan, M.K. Hughes, and
320 bibliography
B.S. Wallin. 2003. Preliminary reconstructions of spring precipitation in south-

western Turkey from tree-ring width. International Journal of Climatology 23:
157171.
Touchan, R., E. Xoplaki, G. Funkhouser, J. Luterbacher, M.K. Hughes, N. Erkan,
. Akkemik, and J. Stephan. 2005. Reconstructions of spring/summer precipi-
tation for the Eastern Mediterranean from tree-ring widths and its connection
to large-scale atmospheric circulation. Climate Dynamics 25: 7598.
Touchan, R., . Akkemik, M.K. Hughes, and N. Erkan. 2007. MayJune precipitation
reconstruction of southwestern Anatolia, Turkey during the last 900 years from
tree rings. Quaternary Research 68: 196202.
Touchan, R., K.J. Anchukaitis, D.M. Meko, S. Attalah, C. Baisan, A. Aloui. 2008. Long
term context for recent drought in northwestern Africa. Geophysical Research
Letters 35: L13705.
Touchan, R., D.M. Meko, and A. Aloui. 2008b. Precipitation reconstruction for
Northwestern Tunisia from tree rings. Journal of Arid Environments 72: 1887
1896.
Touchan, R., K.J. Anchukaitis, D.M. Meko, M. Sabir, S. Attalah, and A. Aloui. 2010.
Spatiotemporal drought variability in northwestern Africa over the last nine
centuries. Climate Dynamics 37: 237252.
Toynbee, J.M.C. 1973. Animals in Roman life and art. London.
Trenberth, K.E. 1997. The definition of El Nio. Bulletin of the American Meteoro-
logical Society 78: 27712777.
Trigger, B.G. 2003. Understanding early civilizations: a comparative study. Cam-
bridge.
Trigo, I.F., T.D. Davies, and G.R. Bigg. 2000. Decline in Mediterranean rainfall caused
by weakening of Mediterranean cyclones. Geophysical Research Letters 27: 2913
2916.
Trouet, V., J. Esper, N.E. Graham, A. Baker, J.D. Scourse, and D.C. Frank. 2009. Persis-
tent Positive North Atlantic Oscillation Mode Dominated the Medieval Climate
Anomaly. Science 324: 7880.
Tucci, P.L. 2004. Eight Fragments of the Marble Plan of Rome Shedding New Light
on the Transtiberim. PBSR 72: 185202.
Turchin, P., and W. Scheidel. 2009. Coin hoards speak of population declines in
ancient Rome. Proceedings of the National Academy of Sciences 106: 17276279.
Trke, M., and E. Erlat. 2003. Precipitation Changes and Variability in Turkey
Linked to the North Atlantic Oscillation During the Period 19302000. Interna-
tional Journal of Climatology 23: 17711796.
Ulrich, R.B. 2007. Roman Woodworking. New Haven, CT.
Unal, Y.S., A. Deniz, H. Toros, and S. Incecik. 2012. Temporal and spatial patterns of
precipitation variability for annual, wet, and dry seasons in Turkey. International
Journal of Climatology 32: 392405.
United States Department of Agriculture, Foreign Agricultural Service. Monthly
Crop Growth Stage and Harvest Calendars. http://www.fas.usda.gov/pecad/
weather/Crop_calendar/crop_cal.pdf, consulted January 4, 2012.
University of Florida, the Food and Agricultural Organization of the United Nations,
and the National Museum of Natural History of the Smithsonian Institution.
bibliography 321
Ecoport, Secale cereale, Description: Physiology, http://ecoport.org/ep?Plant

=1929&entityType=PL&entityDisplayCategory=full.
Usoskin, I.G., and B. Kromer. 2005. Reconstruction of the 14C Production Rate from
Measured Relative Abundance. Radiocarbon 47: 3137.
Usoskin, I.G., S.K. Solanki, and G.A. Kovaltsov. 2007. Grand minima and maxima
of solar activity: new observational constraints. Astronomy & Astrophysics 471:
301309.
van Andel, T.H., E. Zangger, and A. Demitrack. 1990. Land use and soil erosion in
prehistoric and historical Greece. JFA 17: 379396.
van de Mieroop, M. 2004. A History of the Ancient Near East, ca. 3000323bc. Oxford.
van der Leeuw, S.E., and the Archaeomedes Research Team. 2005. Climate, Hydrol-
ogy, Land Use, and Environmental Degradation in the Lower Rhone Valley during
the Roman Period. Comptes Rendus. Geoscience 337 (12): 927.
van der Plicht, J., H.J. Bruins, and A.J. Nijboer. 2009. The Iron Age around the
Mediterranean: a High Chronology perspective from the Groningen radiocarbon
database. Radiocarbon 51: 213242.
van Geel, B., N.A. Bokovenko, N.D. Burova, K.V. Chugunov, V.A. Dergachev, V.G. Dirk-
sen, M. Kulkova, A. Nagler, H. Parzinger, J. van der Plicht, S.S. Vasiliev, and
G.I. Zaitseva. 2004. Climate change and the expansion of the Scythian culture
after 850BC: a hypothesis. Journal of Archaeological Science 31: 17351742.
van Geel, B., J. Buurman, and H.T. Waterbolk. 1996. Archaeological and palaeo-
ecological indications for an abrupt climate change in The Netherlands and evi-
dence for climatological teleconnections around 2650bp. Journal of Quaternary
Science 11: 451460.
van Geel, B., C.J. Heusser, H. Renssen, and C.J.E. Schuurmans. 2000. Climatic change
in Chile at around 2700bp and global evidence for solar forcing: a hypothesis.
The Holocene 10: 659664.
van Geel, B., O.M. Raspopov, H. Renssen, J. van der Plicht, V.A. Dergachev, and
H.A. Meijer. 1999. The role of solar forcing upon climate change. Quateranry
van Geel, B., and H. Renssen. 1998. Abrupt climate change around 2,650bp in North-
West Europe: evidence for climatic teleconnections and a tentative explanation.
In Water, Environment and Society in Times of Climatic Change, edited by A.S. Issar
and N. Brown, 2141. Amsterdam.
van Geel, B., J. van der Plicht, M.R. Kilian, E.R. Klaver, J.H.M. Kouwenberg, H.
Renssen, I. Reynaud-Farrera, and H.T. Waterbolk. 1998. The sharp rise of 14C
ca. 800 cal bc: possible causes, related climatic teleconnections and the impact
on human environments. Radiocarbon 40: 535550.
Van Ossel, P. 1992. tablissements ruraux de lAntiquit tardive dans le nord de la
Gaule. Paris.
Veal, R. 2009. The Wood Fuel Supply to Pompeii, 3rd c. BC to AD79: an environmental,
historical and economic study based on charcoal analysis. Doctoral thesis, Depart-
ment of Archaeology, University of Sydney, Sydney.
2012. From Context to Economy: Charcoal as an Archaeological Interpreta-
tive Tool. A Case Study from Pompeii (3rd c. B.C. to A.D. 79). In More than just
numbers? The role of science in Roman archaeology, Journal of Roman Archaeol-
ogy Supplement 91, edited by I. Schrufer-Kolb, 1952. Portsmouth, RI.
322 bibliography
forthcoming. Fuelling Pompeii. London.

Veal, R., L. ODonnell, and L. McParland. 2011. Measuring burn temperatures from
charcoal using the reflectance method, first results from an Irish Bronze Age
cremation site. In 5th International Meeting of Charcoal Analysis. The charcoal as
cultural and biological heritage, edited by E. Badal, Y. Carrin, E. Grau, M. Macas
and M. Ntinou. 2122. Valencia.
Veal, R., and G. Thompson. 2008. Fuel Supplies for Pompeii. Pre-Roman and Roman
charcoals of the Casa delle Vestali. In Charcoals From the Past: Cultural and
Palaeoenvironmental Implications. Proceedings of the Third International Meet-
ing of Anthracology, CavallinoLecce (Italy) June 28thJuly 1st 2004, BAR Intern-
taional Series 1807, edited by Girolamo Fiorentino and Donatella Magri, 287298.
Oxford.
Verheyden, S., F.H. Nader, H.J. Cheng, L.R. Edwards, R. Swennen. 2008. Paleoclimate
reconstruction in the Levant region from the geochemistry of a Holocene stalag-
mite from the Jeita Cave, Lebanon. Quaternary Research 70: 368381.
Vermoere, M., S. Six, J. Poblome, P. Degryse, E. Paulissen, M. Waelkens, and E. Smets.
2003. Pollen Sequences from the City of Sagalassos (Pisidia, Southwest Turkey).
AS 53: 161173.
Vernant, J-P. 1957. Remarques sur les formes et les limites de la pense technique
chez les Grecs. Revue dhistoire des sciences et de leurs application 10, 3: 205225.
Vetters, W., and H. Zabehlicky. 2001. Eine Klimakatastrophe um 200 n. Chr. und ihre
archologisch-historische Nachweisbarkeit. In Archologie, Naturwissenschaf-
ten, Umwelt: Beitrge der Arbeitsgemeinschaft Rmische Archologie auf dem 3.
Deutschen Archologenkongress in Heidelberg, 25.530.5.1999, edited by M. Frey
and N. Hanel, 912. Oxford.
Vicente-Serrano, S.M., J.I. Lpez-Moreno, L. Gimeno, R. Nieto, E. Morn-Tejeda,
J. Lorenzo-Lacruz, S. Beguera, and C. Azorin-Molina. 2011. A multiscalar global
evaluation of the impact of ENSO on droughts. Journal of Geophysical Research
116, D20109.
Vieira, L.E.A., S.K. Solanki, N.A. Krovova, and I. Usoskin. 2011. Evolution of the solar
irradiance during the Holocene. Astronomy & Astrophysics 531 (A6).
Vigneron, P. 1968. Le cheval dans lAntiquit greco-romaine. Nancy.
Vinther, B.M., H.B. Clausen, S.J. Johnsen, S.O. Rasmussen, J.P. Steffensen, K.K. Ander-
sen, S.L. Buchardt, D. Dahl-Jensen, I.K. Seierstad, A.M. Svensson, M.-L. Siggaard-
Andersen, J. Olsen, and J. Heinemeier. 2008. Reply to comment by J.S. Denton and
N.J.G. Pearce on A synchronized dating of three Greenland ice cores throughout
the Holocene. Journal of Geophysical Research 113: D12306.
Visser, R.M. 2010. Growing and Felling? Theory and Evidence Related to the Appli-
cation of Silvicultural Systems in the Roman Period. In TRAC 2009: Proceedings
of the Nineteenth Annual Theoretical Roman Archaeology Conference, edited by
A. Moore, G. Taylor, P. Girdwood, E. Harris, and L. Shipley, 1122. Ann Arbor, MI;
Southampton.
Vita-Finzi, C. 2008. Fluvial solar signals. In Landscape Evolution: Denudation, Cli-
mate and Tectonics over Different Time and Space Scales, Geological Society, Lon-
don, Special Publications 296, edited by K. Gallagher, S.J. Jones, and J. Wain-
wright, 105115. London.
Vogt, S., R. Glaser, J. Luterbacher, D. Riemann, G.H. Al Dyab, J. Schoenbein, and
bibliography 323
E. Garcia-Bustamante. 2011. Assessing the Medieval Climate Anomaly in the

Middle East: the potential of Arabic documentary sources. PAGES News 19: 28
29.
Vonmoos, M., J. Beer, and R. Muscheler. 2006. Large variations in Holocene solar
activity: Constraints from 10Be in the Greenland Ice Core Project ice core. Journal
of Geophysical Research 111: A10105. doi: 10.1029/2005JA011500.
Walker, G.T., and E.W. Bliss. 1932. World Weather V. Memoirs of the Royal Meteoro-
logical Society 4(36): 5384.
Walling, D.E. 2006. Human impact on land-ocean sediment transfer by the worlds
rivers. Geomorphology 79(34): 192216.
Wang, T., D. Surge, and S. Mithen. 2012. Seasonal temperature variability of the
Neoglacial (33002500bp) and Roman Warm Period (25001600bp) recon-
structed from oxygen isotope ratios of limpet shells (Patella vulgata), Northwest
Scotland. Palaeogeography, Palaeoclimatology, Palaeoecology 317318: 104113.
Wanner, H., J. Beer, J. Btikofer, T.J. Crowley, U. Cubasch, J. Flckiger, H. Goosse,
M. Grosjean, F. Joos, J.O. Kaplan, M. Kttel, S.A. Mller, I.C. Prentice, O. Solom-
ina, T.F. Stocker, P. Tarasov, M. Wagner, and M. Widmann. 2008. Mid- to Late
Holocene climate change: an overview. Quaternary Science Reviews 27: 1791
1828.
Wanner, H., O. Solomina, M. Grosjean, S.P. Ritz, and M. Jetel. 2011. Structure and
origin of Holocene cold events. Quaternary Science Reviews 30: 31093123.
Ward-Perkins, B. 2005. The Fall of Rome and the End of Civilization. Oxford.
Warde, P. 2006. Fear of Wood Shortage and the Reality of the Woodland in Europe,
c. 14501850. History Workshop Journal 62: 2857.
Wassenburg, J., J. Fietzke, D. Richter, and A. Immenhauser. 2010. A Holocene speleo-
them record from Morocco, NW Africa. EGU General Assembly 2010, held 27
May, 2010 in Vienna, Austria. [Katlenburg-Lindau].
Watson, A. 1999. Aurelian and the Third Century. London.
Watson, R.T., M.C. Zinyowera, and R.H. Moss, eds. 1998. The Regional Impacts of
Climate Change: An Assessment of Vulnerability. A special report of IPCC Working
Group II. Project administrator: David J. Dokken. Cambridge.
Weinstein, L. 2009. Limitations on Anthropogenic Global Warming. In
noconsensus.wordpress.com/2009/05/27/the-truth-about-arctic-and-greenland
-ice/.
Weiss, H. 2000. Beyond the Younger Dryas: Collapse as Adaptation to Abrupt Cli-
mate Change in Ancient West Asia and the Eastern Mediterranean. In Environ-
mental Disaster and the Archaeology of Human Response, edited by G. Bawden
and R.M. Reycraft, 7598. Albuquerque.
Weiss, H., and R. Bradley. 2001. What Drives Societal Collapse. Science 291: 609610.
Weiss, H., M.-A. Courty, W. Wetterstrom, F. Guichard, L. Senior, R. Meadow, and
A. Curnow. 1993. The Genesis and Collapse of Third Millennium North Mesopo-
tamian Civilization. Science 261: 9951004.
Weiss, E., M.E. Kislev, and A. Hartmann. 2006. Autonomous Cultivation before
Domestication. Science 312 (16 June): 16081610.
Weninger, B., L. Clare, E.L. Rohling, O. Bar-Yosef, U. Bhner, M. Budja, M. Bundschuh,
A. Feurdean, H.G. Gebel, O. Jris, J. Linstadter, P. Mayewski, T. Mhlenbruch,
A. Reingruber, G. Rollefson, D. Schyle, L. Thissen, H. Todorova, and C. Zielhofer.
324 bibliography
2009. The impact of rapid climate change on prehistoric societies during the
Holocene in the eastern Mediterranean. Documenta Praehistorica 36: 759.
Werner, J.P., J. Luterbacher, and J.E. Smerdon. 2013. A Pseudoproxy Evaluation of
Bayesian Hierarchical Modelling and Canonical Correlation Analysis for Climate
Field Reconstructions over Europe. Journal of Climate 26(3): 851867.
Wertime, T.A. 1983. The Furnace versus the Goat: The Pyrotechnologic Industries
and Mediterranean Deforestation in Antiquity. JFA 10: 445452.
Wheeler, E.A., P. Baas, and P.E. Gasson. 1989. IAWA list of microscopic features for
hardwood identification. Leiden.
Wheeler, S.M. 2012. Nutritional and Disease Stress of Juveniles from the Dakhleh
Oasis, Egypt. International Journal of Osteoarchaeology 22: 219234.
Whitby, M. 2000. The Balkans and Greece, 420602. Cambridge Ancient History
XIV: 701730.
White, K.D. 1970. Roman Farming. Ithaca, NY.
Wick, L., G. Lemcke, and M. Sturm. 2003. Evidence of Lateglacial and Holocene
climatic change and human impact in eastern Anatolia: high-resolution pollen,
charcoal, isotopic and geochemical records from the laminated sediments of
Lake Van, Turkey. The Holocene 13: 665675.
Wightman, E.M. 1970. Roman Trier and the Treveri. London.
Wikander, . 1979. Water-mills in Ancient Rome. ORom: 1336.
2000. Handbook of Ancient Water Technology. Leiden; Boston.
2008. Sources of Energy and Exploitation of Power. In The Oxford Hand-
book of Engineering and Technology in the Classical World, edited by J.P. Oleson,
136157. Oxford.
Williams, M. 2003. Deforesting the Earth. Chicago.
Wilson, A. 1997. Water management and usage in Roman North Africa: a social and
technological study. D. Phil. Thesis, University of Oxford.
2002a. Machines, Power and the Ancient Economy. JRS 92: 132.
2002b. Marine resource exploitation in the cities of coastal Tripolitania. In
LAfrica Romana XIV, vol. 1, edited by M. Khanoussi, P. Ruggeri, and C. Vismara,
429436. Roma.
2003. Une cit grecque de Libye: fouilles dEuhesprids (Benghazi).
Comptes rendus des sances de lAcadmie des inscriptions et belles-lettres,147e
anne, N. 4, 2003: 16481675.
2006. The economic impact of technological advances in the Roman con-
struction industry. In Innovazione tecnica e progresso economico nel mondo
romano, edited by E. Lo Cascio, 225236. Bari.
2007. The metal supply of the Roman Empire. In Supplying Rome and
the empire: the proceedings of an international seminar held at Siena-Certosa di
Pontignano on May 24, 2004, on Rome, the provinces, production and distribu-
tion, (JRA Supplement), edited by E. Papi and B. Scardigli, 109125. Portsmouth,
RI.
2008a. Hydraulic Engineering. In Handbook of Engineering and Technology
in the Classical World, edited by J.P. Oleson, 285318. Oxford.
2008b. Machines in Greek and Roman Technology. In The Oxford Hand-
book of Engineering and Technology in the Classical World, edited by J.P. Oleson,
337366. Oxford.
bibliography 325
2009. Indicators for Roman economic growth: a response to Walter Scheidel.

JRA 22.1: 7182.
2011. City sizes and urbanization in the Roman Empire. In Settlement, Urban-
ization, and Population, (Oxford Studies on the Roman Economy), edited by
A. Bowman and A. Wilson, 160195. Oxford.
2012a. Saharan trade in the Roman period: short-, medium- and long-
distance trade networks. Azania: Archaeological Research in Africa 47.4: 409
449.
2012b. Raw materials and energy generation. In The Cambridge Companion
to the Roman Economy, (Cambridge Companions to the Ancient World), edited
by W. Scheidel, 133155. Cambridge.
Wilson, A.I., and D.J. Mattingly. 2003. Irrigation technologies: foggaras, wells and
field systems. In The Archaeology of Fazzan, vol. 1., edited by D.J. Mattingly,
235278. London.
Woodbridge, J., and N. Roberts. 2011. Late Holocene climate of the Eastern Mediter-
ranean inferred from diatom analysis of annually-laminated lake sediments.
Wossink, A. 2009. Challenging Climate Change: Competition and cooperation among
pastoralists and agriculturalists in northern Mesopotamia (c. 30001600BC). Lei-
den.
Wrigley, E.A. 2010. Energy and the English Industrial Revolution. Cambridge.
2013. Energy and the English Industrial Revolution. Philosophical Transac-
tions of the Royal Society 371: 110.
Wynn, G. 2011. Analysis: wood fuel poised to be next global commodity. http://uk
.reuters.com/article/2011/05/19/us-energy-biomass-commodity-idUSTRE
74I3NK20110519
Xoplaki, E. 2002. Climate variability over the Mediterranean. Ph.D. thesis, Univer-
sity of Bern. Available from: http://sinus.unibe.ch/klimet/docs/phd_xoplaki.pdf
[acccessed January 2012].
Xoplaki, E., J. Luterbacher, H. Paeth, D. Dietrich, N. Steiner, M. Grosjean, and H. Wan-
ner. 2005. European spring and autumn temperature variability and change of
extremes over the last half millennium. Geophysical Research Letters 32: L15713.
Yair, A., and A. Kossovsky. 2002. Climate and surface properties: hydrological re-
sponse of small arid and semiarid watersheds. Geomorphology 42: 4357.
Yakir, D., A. Issar, J. Gat, E. Adar, P. Trimborn, and J. Lipp. 1994. 13C and 18O of wood
from the Roman siege rampart in Masada, Israel (AD7073): Evidence for a less
arid climate for the region. Geochimica et Cosmochimica Acta 58: 35353539.
Yin, H., and C. Li. 2001. Human impact on floods and flood disasters on the Yangtze
River. Geomorphology 41(23): 105109.
Zabehlicky, H. 1994. Kriegs- oder Klimafolgen in archologischen Befunden?
Markomannenkriege: Ursachen und Wirkungen, edited by H. Friesinger, J. Tejral,
and A. Stuppner, 463469. Vienna.
Zhang, Q.B., G.D. Cheng, T.D. Yao, X.C. Kang, and J.G. Huang. 2003. A 2326-year tree-
ring record of climate variability on the northeastern Qinghai-Tibetan Plateau.
Geophysical Research Letters 30(14), 1739.
Zielinski, G.A., P.A. Mayewski, L.D. Meeker, S. Whitlow, and M.S. Twickler. 1996. A
110,000-yr record of explosive volcanism from the GISP2 (Greenland) ice core.
Quaternary Research 45:109118.
INDEX
Ancient and modern place names are listed indiscriminately, according to the usage of the
various contributors. All lake and river names are listed under Lake and River.
Abate, Pietrantonio, 255 Iulia, 239

Abella, 254 Traiana, 255
Abellinum, 245 Virgo, 255
Acarnania, 184 and see Acquedotto
Acer spp.: see maples Arabia, 199
Acquaro spring, 242246 arboreal pollen, 183185
and see Serino and see palynological evidence
Acquedotto: Arcadia, 180
della Bolla, 254 aridification, 45, 41, 203, 208210
Campano, 254 Aristotle, 267268
di Serino, 255 Arpinum, 240
Aegean, 160 Arruntius, L. (curator aquarum), 248,
Islands, 180 252
Africa, north, 178, 188, 263, 273 Asia, central, 45, 89102 passim
Proconsularis, 81 Asia (Minor), 179
agriculture, beginnings of, 105 Ateius Capito, C. (curator aquarum), 248,
alder, 53 252
Alexandria, 178, 180, 184 Athens, 115, 177, 179
Alps, 151, 185 Attica, 177, 184, 194
Althiburos (Numidia), 273 Augustus, 242, 248, 252253, 269
Amisus, 179 Aurelian, 271
Ammerman, A., 260 Ausonius, 4, 6368, 87
amphorae, 272273 Avars, 5, 8993, 101
Anatolia, 161163
Ancient Climatic Optimum, 24 Baiae, 241
animals, working, 32 Basilicata, 237
Anio Novus, 254 beech (Fagus sylvatica), 3, 53, 56, 216
anthropogenic factors: see human impact Behre, K.-E., 4
on landscape and rivers, pollution Bereket Basin (Turkey), 161
Antigonus the One-Eyed, 180 Beryllium-10 records, 121123, 126127, 130,
Antioch, 178, 186 132, 134
Antonine Plague, 27, 264265 Bologna, 255
Antonius, M., 181 Bonneau, D., 4, 71, 7682
Apicius, 55 Boswijk, G., 93, 102
Appennines, 182, 216, 233, 235 bradyseism, 242
aqueducts, 910, 233, 254255, 268 Bresson, A., 3, 22
Acqua Felice, 255 Britain, Roman, 23, 259260
Marrana Mariana, 254 Bronze Age, Early, 132
Paola, 255 Bucca della Renella (cave in the Apuan
Aqua Alexandrina, 254 Alps), 147148, 149, 150
Augusta, 9, 240246, 252, 254255 Bntgen, U., 89, 103, 105, 116, 117, 139145, 157,
Claudia, 254 262, 266
Crabra, 239
328 index
Caesar, C. Iulius, 248, 251252, 269271 coastline changes, 218221, 225228,

Calpe (Bithynia), 179 n. 229231, 275
Campania, 9, 233246, 253256, 268 colonization, Greek, 114115, 219
Campi Flegrei, 214215, 218 Columella, 32, 190191
Carbon-14: see Radiocarbon Comte, Auguste xiii
Carpinus spp.: see hornbeams Constantinople, 72, 240, 268
Carthage, 268 construction, timber for, 177, 179183
Catalonia, 184 consular honors, consulate, 6567
Cato, 32, 190191 Cook, E., 2, 4, 262
cedars, 40, 180 coppicing, 44, 190, 191192
cereals, 8384 Corcyra, 179
chaff, 39 Corfinium, 255
Chakrabarty, D., 2 Corinth, 179
charcoal, 3, 3758, 190 Corsica, 182
varieties of, 5455 cremation, 179 n.
China, climate in, 45, 89101, 161, 275 Crete, 137
use of machines in ancient, 22 Cumae, 215
Cicero, 239240, 250 curatores alvei Tiberis et riparum, 270
Cilicia, 179 Curius Dentatus, M., 250
Claudian, 251 cypresses, 136137, 180, 190191
Claudius, 270271 Cyprus, 180
Cleopatra, 181
climate, climate change, 110, 21, 2429, Dalton Minimum, 124
61170 passim, 185, 192193, 200211, Danube provinces of Roman Empire, 186
261266, 276 Dead Sea, 7072, 157, 160, 203204
central-European, 4, 24, 2629, 67, 85, 89, declarations of unflooded land (apographai
136143, 145, 150, 152153, 157, 163165, abrochias), 8182
167, 169170, 188189, 213 deforestation, 8, 2728, 42, 116, 173194, 236,
defectiveness of concepts of favourable, 273
unfavourable, 211 Delos, 40
indifference of rural settlement to, 13C, 146, 148, 150, 167
197211 18O, 146, 147148, 235
methods for studying, 6163, 7476, 82, demography: see population
168170 dendochronology, 73
regional differences within the Mediter- and see tree rings
ranean world, 109112, 113170 passim Dio Cassius, 251252
see-saw relationship between eastern Diocletians Price Edict, 40 n.
and western Mediterranean, 116, 162 Dionysius of Halicarnassus, 181
short- and long-term, 82 Douglas fir (Pseudotsuga menziesii), 93,
stability during Roman period, 26, 70, 96100
104, 133, 149, 151, 154, 157, 161, 169 Drake, B.L., 115
timescales in, 108109, 197; droughts, 89101 passim, 114, 243244
and see Ancient Climatic Optimum, Dulan-Wulan (north-central China), 9091,
Dalton Minimum, Greenland temper- 100101
ature proxy record, instability, Iron Dumbarton Oaks, 4, 69, 102, 103 n.
Age Cold Period, Little Ice Age (LIA), Dumnissus (Kirchberg), 6364, 67
Maunder Minimum, Medieval Cli-
mate Anomaly, microclimates, North earthquakes, 274275
Atlantic Oscillation, precipitation, Egypt, 4, 31, 71, 82, 174, 178180, 265
Roman Climate Optimum, Roman and see River Nile
Warm Period Elba, 178
coal, 2223, 28 Elea: see Velia
index 329
elm, 191 Greek Renaissance, 6, 112115, 132

El Nio/Southern Oscillation (ENSO), 5, Greenland ice cores, 2425, 110, 127, 154, 155,
7071, 89, 92101 159, 274, 275
energy, 1236, 259261 Grissino-Mayer, H., 93, 102
energy-converters, efficiency of, 19, 21 Grove, A.T., 43
Ermolli, E. Russo, 8 growth, economic, 13, 2829, 264265
erosion, 185187, 237
Euboea, 180 Hadrian, 182183
Eupolis, 177 Hamaxia (in Cilicia), 181
Euro-Climhist, 61 n. Harris, W.V., 57, 273
harvest yields in Egypt, 78
Finnish Jabal Harun Project (FJHP), 198 heating systems, domestic, 8586
firewood, 14, 1516, 19, 3032, 34, 36, 272273 Hero of Alexandria, 22
and see fuel, wood history in vs. history of, 12, 3, 37
fish, fishing, 7, 64, 240, 246, 275 Histria, 182
flood gauge in Rome, 270 Hitchner, B., 264
flooding, 9, 74, 157, 162, 165, 216, 228229, Horden, P., 8, 120
235239, 246253, 268272 hornbeams (Carpinus spp.), 56
and see River Nile Hughes, J.D., 175
Florentia, 253 human impact on landscape and rivers,
fodder, 14, 16, 32 173256 passim
food consumption in antiquity, 15, 30, 34 Huns, 45, 7071, 8993, 101
foraminifera, use in climate history, 147148 Huntington, E., 89
forest, 64 hydrological cycle, ancient understanding
husbandry: see management of wood- of, 266272
land hypocausts, 85
Forino aqueduct tunnel (Campania), 244
Fowler, A., 93, 102 Iberia, 108, 157, 184
Frontinus, 239 and see Spain
fruit trees, wood of, 56 Ida, Mount, 179
fuel, 1236, 3740 Ilopango volcano (El Salvador), 275
and see charcoal, coal, firewood, olive instability, climatic, 5, 7071, 203, 261
lees and pits, wood indexes, 88 n.
and see climate
Garamantes, 266 Interamna, 250, 253
Gaul, 63, 193 International Tree-Ring Data Bank (ITRDB),
GDP, 1920, 3536, 3840 93
Geoponica, 83 Iran, 161
Germany, ancient, 4, 26 Iron Age Cold Period, 148
Gilliam, J.F., 265 iron production, Roman, 3132, 176, 178
Giraudi, C., 234235, 238239 irrigation, 260
glaciers, 70, 112, 135, 235 in Italy, 239240, 245
global warming, 106 in Nabataean territory, 201202, 211
goats, 177 in North Africa, 266
Golan Heights, 184185 Israel, 26, 203
government regulation of trees, forest, 180, and see Soreq Cave
182183 Istanbul, timber excavated at, 137, 139140
grafting, 190 Italy, climate history of, 118
Gratian, 64 ecological science in, 73
La Graufesenque pottery, 272 environmental case-studies, 213256
Greece, environmental conditions in woodlands of, 181
ancient, 180, 194 Iulius: see Caesar
330 index
Jabal Harun (Jordan), 201202, 263 maples (Acer spp.), 56

Jabal ash-Shara (Jordan), 198, 204209 Masada, 157
Jebel al-Aqra (Turkey), 186187 materia, 179
Jongman, W., 30 Mauguio, tang de (Provence), 184 n.
Jordan, 174, 198211, 273 Maunder Minimum, 107, 111, 113, 121, 144
juniper (Juniperus przewalskii), 90 McCormick, M., 2, 4, 10, 89, 192193, 261262,
264265, 274
Kaplan, J.O., 187189 McParland, L., 53
Kasserine Survey, 263 medieval climate, 165166
kauri (Agathis australis), 93100 Medieval Climate Anomaly (MCA),
Keenan-Jones, D., 9, 268, 274 106110, 119120, 149, 151, 154, 158,
Kocain Cave (southern Turkey), 146, 149 n. 162
Kouki, P., 6, 8, 263 and see late antiquity
Krumhardt, K.M., 187189 Meiggs, R., 40
metal-smelting, 194
Lake Accesa, 235, 238 and see iron production
Bafa, 187 microclimates, 3
Fucino, 235, 238, 256 micro-regions, 8
Ledro, 238239 migration, 56, 27, 7071
Martignano, 238 Migration Period, 143, 213
Mezzano, 238, 256 Miletus, 184
Monticchio (Lago Grande di Montic- Misenum, 241
chio), 236 Morocco, 108, 109, 144
Nar (Nar Gl), 161, 162 Morris, I., 3436
Praver, 186
Van, 158, 160 Nabataean kingdom, 198201
Velinus, 248, 250, 270 Naples, 89, 214225, 229, 237, 240, 244245,
Zoar (southern Spain), 162 268
lake levels, 235, 237238 natura, 250251
land clearance, 173176, 236 navicularii lignarii, 178
land reclamation, 256 navies, 177, 179181
land requirements per capita, 18, 30 Neapolis: see Naples
landscapes, ruined, 175, 180 Negev, 201202
typology of, 175 Nero, 251
larch (Larix decidua), 139 New Mexico, tree rings from, 92, 93, 9598
late antiquity, 89101, 158162, 163166, New Zealand, tree rings from, 92, 9398
169170, 185186, 194, 203, 209, 242, North Atlantic Oscillation, 106111
266 nut trees, wood of, 56
Laurion silver mines, 177
Lebanon, Mount, 180, 182183 oaks (Quercus spp.), 50, 53, 56, 68, 137138,
Levant, southern, 161163 139, 185
lignum, 177179 olive lees and pits, 39, 260
Little Ice Age (LIA), 106110, 119120, 144, trees, 185
149, 151, 154, 162, 235 Olympus, Mount, 179
OSL (optically stimulated luminescence)
Macedon, 179180 dating, 202203
marginal land, 43, 200, 205, 263264 Ostia, 178, 182, 270271
Malanima, P., 24, 30, 57, 259261, 263
Malthusian constraints, 264 palaeoclimate research, 120121
management of woodland, 4244, 52, 58, Palladius, 83
178, 185186, 189192 Palmer, J., 93, 102
Manning, S., 3, 56, 2527, 192, 262 Palmer Drought Series Index, 109111, 145
index 331
palynological evidence, 42, 52, 183185, 193, Ombrone, 237

203, 216218, 222225, 236 Po, 235, 237, 239
Parthenope, 214215 Rhone, 237
Paulinus of Nola, 234, 254 Sabato, 9, 244246
Pausanias, 267 Tiber, 9, 235237, 239, 246253, 256,
Pelosi spring, 242246 268272
and see Serino rivers, 233256, 260
Petra, 6, 8, 198211 Roman Climate Optimum, 133135, 157
pines: stone pine (Pinus cembra), 139 Roman Empire, total cultivated area of, 18
Pisa, 181 Roman Warm Period, 23, 24, 70, 134135,
place-names, 64 148, 158 n., 162, 235, 262
Pliny the Elder, 182, 191192, 249, 251252, Romans and hydrology, 233256
267 Rome, 178, 268272
Pliny the Younger, 2 n. Baths of Constantine and Diocletian, 178
pollarding, 44 Campus Martius, 247248, 268271
pollen: see palynological evidence Cloaca Maxima, 269
pollution, ancient, 273274 Forum Romanum, 248, 269270
Pompeii, 3, 5557, 273 Forum Boarium, 247
and see Vesuvian cities fuel needs of, 31
Pompey (Cn. Pompeius Magnus), 269 Horologium, 271
population growth, 24, 27, 114, 176, 187189, Pantheon, 271
192 Porticus inter lignarios, 178
pottery manufacture at La Graufesenque, Pons Aurelius, 270
272 Pons Sulpicius, 269
in the Petra region, 204 Porta Trigemina, 178
precipitation in Roman times, 2627, 68, 70, Velabrum, 269270
7172, 159161, 167, 234, 235 Rossignol, B., 25
progress, 253254 runoff cultivation, 202, 204, 276
Provence, 184 rye, 4, 8384
public-works projects, 252253
Puglia, 184185 Sagalassos, 185
Purcell, N., 12, 8, 120 Sahara, 136, 266
Puteoli, 240241, 255 Salerno, 254
Bay of, 216
Quercus spp.: see oaks Sannio, Universit degli Studi del, 243
S. Lucia di Serino: see Serino
Rackham, O., 43 Santorini (Thera) eruption, 126 n.
radiocarbon dating, 75, 121122, 123125, Saserna (Roman agronomist), 32
126127, 129132, 134, 138, 151, 167, Scheidel, W., 264
168 sedimentation, 186187, 213
Raetia, 182 Seneca, 267
Reate, 250253 Serino springs, 243246, 268
reflectance, 47, 5354 ships, sailing-, 21, 33, 260
religio, 251252 ship-building, timber for, 42, 177, 179, 180, 181,
River Arno, 9, 237, 248249 194
Clanis (Chiana), 9, 248250, 256, 270, Sicily, 180
272 Sicyon, 180
Frittolo, 215216, 225229 silt, 237
Guadalquivir, 272 Sinope, 179
Moselle, 6364 Smil, V., 36
Nar, 248249, 270 Sofular Cave (Turkey), 146147, 149, 150,158,
Nile, 7071, 7681, 82, 157, 264266 161
332 index
solar activity, 103 Trajan, 271

influence of on climate, 120135, 139, 144, transport, cost of, 57
151, 159, 168169 water-borne, 186
solar minimum about 2860 bc, 132 tree rings, 5152, 6768, 89, 90100, 119120,
solar minimum about 765 bc, 112, 132 136140, 152153, 154, 156
solar radiation, veiling of about 536/7 ad, Trier, 64
72, 139, 165, 275 Trigger, B.G., 104
and see Maunder Minimum, Sun Spot Troodos Mountains (Cyprus), 186
Number, Total Solar Irradiance
Soreq Cave (Israel), 147149, 150, 158, 160 Ulrich, R.B., 182
Spain, 148, 150, 151, 152, 167, 185 UNESCO Libyan Valleys Study, 263
and see Iberia uranium-thorium dating, 75
Spannagel Cave (Austria), 151153, 163 Urciuoli spring, 243246
speleothems, use of for climate history, 73,
120, 146153 Val Febbraro (Sondrio), 186
state, Roman, and the environment, 9, 250, Valentinian I, 6467
253, 269 Varro, 32, 191
states, formation of in third millen- Veal, R., 3, 273
nium bc, 132 Velia (Elea), 810, 215217, 225231, 237
and see government regulation Vernant, J.-P., 22
stature, variations in, 24, 259260 Vesuvian cities, 182
steam power, 22, 28 and see Pompeii
Stoicism, 250, 267 vines, 191
Strabo, 181, 250 Vitruvius, 266267
sun: see solar volcanic eruptions, 70, 71, 88, 103, 120, 154,
Sun Spot Number, 131 155156, 214215, 274275
Switzerland, 26
Syracuse, 179 Wadi Araba (Jordan), 198, 204205
Syria, 179, 182183 Wadi Faynan (Jordan), 201, 273
water mills, 1617, 21, 33, 64
Tabernae (Roman Germany), 6364 Wilson, E.O., 6869
Tacitus, 246, 250, 253 wind power, 1617
Talmud, 203 wood: atlases, 49
tax flight, 265 calorific value of various types of, 5253
Tebtunis, 178 demand for, 174, 176177
Terminio aquifer, 242243 as fuel, 38, 45, 47, 176179
thatched roofs, 8687 supply, 4144
Theophrastus, 179180, 190, 266267 and see charcoal, firewood, fuel
Thrace, 179 woodland, over-exploitation of, 52
Tiberius, 246248, 268270 and see deforestation, management
Tierra Blanca Joven event, 275 Wrigley, E.A., 13, 28
Torone, 177
Total Solar Irradiance, 122, 126, 128129, 130, Zimmerman, N., 187189
167

(Columbia Studies in The Classical Tradition 39) Harris, William Vernon-The Ancient Mediterranean Environment Between Science and History-Brill Academic (2013)

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

(Columbia Studies in The Classical Tradition 39) Harris, William Vernon-The Ancient Mediterranean Environment Between Science and History-Brill Academic (2013)

Загружено:

Авторское право:

Доступные форматы

The Ancient Mediterranean Environment

between Science and History

The titles published in this series are listed at brill.com/csct

Library of Congress Cataloging-in-Publication Data

This book is printed on acid-free paper.

List of Tables and Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

What Kind of Environmental History for Antiquity? . . . . . . . . . . . . . . . . . . . 1

Energy Consumption in the Roman World . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Defining and Detecting Mediterranean Deforestation, 800 bce to

Problems of Relating Environmental History and Human Settlement

The Mediterranean Environment in Ancient History: Perspectives

8 The Douglas firKauri average interhemispheric ENSO index

7 Top: detail of the Total Solar Irradiance (dTSI) reconstruction

16 Three 18O plots from Mediterranean locations. A: Soreq Cave

3 Environmental zones in the Petra region . . . . . . . . . . . . . . . . . . . . . . . . . . . 205

Edward R. Cook is Ewing Research Professor at Lamont-Doherty Earth

Elda Russo Ermolli is a Researcher in Physical Geography and Geomor-

Duncan Keenan-Jones is currently a Mellon Postdoctoral Research Asso-

technology and environment. He has recently published on the lead con-

Paolo Malanima is the Director of the Institute of Studies on Mediter-

Sturt Manning is the Goldwin Smith Professor of Classical Archaeology

Michael McCormick is Francis Goelet Professor of Medieval History at

Paola Romano is Associate Professor at the Earth Science Department of

Maria Rosaria Ruello is a post-doctoral researcher in geomorphology and

Andrew Wilson, Professor of the Archaeology of the Roman Empire, Uni-

agreed to contribute a paper about water management, a subject that the

1 Clarified in Horden and Purcell 2005, 357.

papers included here.

McCormicks paper complements Malanimas. Having amusingly settled

valley wasas seems to me quite likelylargely unlocated in time.

or the European occupation of the Americas: when we know about their

Renaissance (Manning, this volume, p. 132).

29 On the latter see especially Purcell 2005.

These case studies, together with those incorporated in the papers by

Empire. It is possible, however, to present plausible data and plausible con-

1. The Input of Energy

energy carriers of ancient Mediterranean civilization. The discovery of fire

7 I use here the word macroinvention following Mokyr 1990.

in temperature and industrial development. A range between 1 and 2 kg.,

Ch. VII and Fig. 7.3.

Table 1 presents a likely consumption range for the ancient Mediterranean

21 Technical change in maritime technology was continuous and certainly contributed to

3. Efficiency and Energy Intensity

27 See the useful Herman 2007.

This ratio depends on the efficiency of the converters, but, contrary to

4. The Energy Constraints

Vegetable energy carriers, such as those exploited in past pre-modern civ-

28International 1990 Geary-Khamis dollars Purchasing Parity Power.

Although this 80 percent was not devoted completely to providing energy,

ingenious concentration of power.

The progress of technology in the ancient Mediterranean world did not

34 Greene 2000; Schneider 2007.

trend of the British economy described by Ward-Perkins 2005, Ch.V.

It is still hard to quantify the rise in population during the millennium

41 Scheidel 2007, 47.

47 Rossignol 2012, 96. See also Manning this volume, Fig. 8.

unusual status, reducing some of the typical variability, uncertainty and

50 Manning this volume.

5th or 6th centuries ad seem to be relatively arid in several areas of the

54 Manning this volume.

Fig. 4. Estimates of forest clearance in Central Europe (Germany, North-Eastern

The energy system of ancient Mediterranean civilizations was the same

62 The topic is discussed in Malanima 2010 and 2011.