Вы находитесь на странице: 1из 13

DETERMINATION OF THE FIRING TEMPERATURE OF

ANCIENT CERAMICS BY MEASUREMENT OF THERMAL


EXPANSION: A REASSESSMENT
BY M. S. TITE*
Ceramics Department, Houldsworth School oi Applied Science,
Utiiversity of Leeds

1. INTRODUCTION
The firing temperatures of ancient ceramics are of interest because they provide
information on the performance of the kilns used in their manufacture and on
the technological capabilities of the ancient potters. Apart from its immediate
archaeological significance, this information is essential for “experiments” in
which full-scale replicas of excavated pottery kilns are built and fired (Mayes,
1961 and 1962). In addition a knowledge of the firing temperature can be of
value in other scientific investigations of ancient ceramics and kilns, such as
thermoluminescent and magnetic dating.
Two different methods have previously been used in the determination of
the firing temperature of ancient ceramics. In the first, a rough indication of
the firing temperature is obtained through studying the minerals present in the
ceramic by means of X-ray diffraction techniques, optical microscopy and
differential thermal analysis (Tobler 1939, Felts 1942, Mariani et al. 1956, Perinet
1960). In the second, the firing temperature is estimated using the data obtained
from thermal expansion measurements on the ceramic.
The thermal expansion method is based on the assumption that when clays are
fired, shrinkage occurs as a result of various sintering processes (i.e. densification
through agglomeration of the fine clay particles). Consequently if a fired clay
ceramic is heated up steadily from room temperature, it exhibits normal reversible
thermal expansion, characteristic of its mineralogical composition, until tempera-
tures comparable with its original firing temperature are reached. With continued
increase in temperature, the ceramic begins to contract since, superimposed on
the reversible expansion, there is an irreversible shrinkage associated with the
resumption of sintering (i.e. the firing of the ceramic is being continued beyond
the point reached during its original firing). The temperature (T,) at which
expansion ceases and contraction begins should therefore provide an estimate
of the original firing temperature (T& the former being defined as the “shrinkage
temperature”.
In principle this method should, with appropriate corrections, provide a more
precise value for the firing temperature than that based on a “mineralogical
temperature scale”. Reasonably satisfactory results have been obtained by Kiefer
(1956). Terrisse (1959) and Roberts (1963). However Matson (1937), working with
low-fired American Indian pottery, observed an irreversible expansion in the
500-800°C range while Cole and Crook (1962), working with Roman brick,
observed a complicated pattern of irreversible expansion and contraction. As
a result of these discrepancies, a detailed investigation into the assumptions, on
which the thermal expansion method is based, was undertaken in order to
establish its validity and accuracy for the determination of firing temperatures
*Present address : Department of Physics, University of Essex.
132 ARCH AEOMETRY

(Section 3). Subsequently the method was used to provide an estimate of the
firing temperature of a wide selection of ancient ceramics (Section 4).

2. APPARATUS
A fused silica extension rod dilatometer. similar to that described by Roberts
(1963), was used for thermal expansion measurements up to 1000°C. For measure-
ments on ceramics fired at temperatures greater than 1000”C, a dilatometer with
polycrystalline alumina components and based on a design by Raudran (1955),
was employed.
In both cases, the specimen and dilatometer assembly were heated in a
platinum-wound tubular furnace (Johnson Matthey and Company Limited: Type
K45B), the heating rate being 200°C per hour, and the change in length of the
specimen was measured using a variable inductance transducer (Boulton Paul
Aircraft Limited: Type F19). The output of the transducer was fed to the Y axis
of a Honeywell Brown X-Y recorder (Model 153 x 32) and the e.m.f. from the
thermocouple, used to measure the temperature of the specimen, was fed to
the X axis, thus providing a plot of the change in length of the specimen versus
temperature.
Changes in specimen length of 0.2 micron (ix. approximately for a
2.5 cm specimen) could readily be measured with both dilatometers and for a
typical specimen the temperature at which shrinkage began could be estimated
with an accuracy of k 5°C. Further details concerning the construction and
performance of the dilatonieters are presented in reports by Tite and Roberts
(1966 and 1969).

3. PRELIMINARY INVESTIGATIONS
The basic problem, in determining the firing temperature of ancient ceramics
from thermal expansion measurements, is to establish the relationship between
the shrinkage temperature (T,) and the original firing temperature (T,) or
“equivalent firing temperature” (T,J*. This relationship is governed by the
dimensional changes occurring when the clay, used in the manufacture of the
ceramic, is heated and these changes depend critically on the clay minerals
(e.g. kaolinite, montmorillonite and illite) and impurities (e.g. quartz, carbonates,
iron oxides and organic matter) present.
Those mineralogical and dimensional changes, relevant to the determination of
firing temperatures, can be summarised as follows (Kiefer 1952 and 1957, Ford
1967): -
(i) When the clay loses chemically combined hydroxyl water (400-6OO0C), a
contraction occurs in the case of kaolinite and an expansion in the case of
montmorillonite and illite. This contraction or expansion then continues at a
greatly reduced rate until a rapid shrinkage, associated with either vitrification
or the formation of high-temperature phases ( e g spinel, mullite), begins
(figure 1).
* The “equivalent firing temperature” (Teq),estimated from thermal expansion measurements
in the case of ancient ceramics, is essentially that constant temperature which would have
brought about the same amount of sintering as was achieyed during the original firing. The
term, firing temperature (Te), has no precise meaning since it is probably that, with many
ancient ceramics, the technique of heating to a peak temperature, followed by immediate
cooling, would have been employed and therefore a major part of the shrinkage, or
sintering, would have taken place at temperatures below the peak temperature.
ARCHAEOM ETRY 133

"0 500 m -0
T'C T"C
FIG. 1. Thermal expansion curves for unfired and prefired specimens of (a) Farnley Fireclay,
(b) Dorset Ball Clay: percentage change in length (AL) versus temperature (TOC).

(ii) Vitrification, by definition, involves the appearance of a viscous liquid


phase in the clay body, which on cooling solidifies to form a glass phase, and
commences somewhere in the 700-950°C range depending on the concentration
of fluxing impurities (e.g. Na,O, K,O, Fe,O,, CaO) present in the clay. In the
vitrification range, the shrinkage increases systematically with firing temperature
and time; the kinetics are those of sintering in the presence of a liquid phase
(Kingery 1959).
(iii) Finally large scale distortion or "slumping" of the clay body occurs in
the 1100-1600°C range due to an increasing amount and a decreasing viscosity
of the liquid phase.
The determination of the shrinkage temperature (TR)of clay specimens which
had been prefired at various temperatures (T,) for 1 hour in air showed that
the difference (T, - T,) ranged from - 210°C to +41"C (Table I). Consequently
equating T,, observed for ancient ceramics, to the "equivalent firing temperature"
(T,J could lead to large errors.

TABLE I
Mirterologicul Vitrification Firing
Cloy composition temp. (T."C) temp. (T,"C) (T. - Td"C
Q K M
London Clay s s a 710 768 - 22
948 + 38
Etruria Marl a a s 850 750 - 26
Dorset Ball Clay m a s 850 806 - 66
946 + 19
Farnley Fireclay a s a 870 654 - 210
785 - 80
960 + 41
China Clay - a - >950 960 - 55
Q-quartz a = abundant > 30%
K-kaolinite s= subsidiary > 15% <30%
M-mica or illite m=minor > 5% < 15%
134 ARCHAEOMETRY

Since the relationship between T, (or T,,) and T, depends on whether T, is


greater or less than the temperature (TY)at which vitrification begins, these two
possibilities are discussed separately in secti0r.s 3.I and 3.2 respectively.
3.1 Firing within the vitrification range (T, > T,)
Measurements on specimens of London Clay, prefired at either 740°C or 920°C,
suggested that if a ceramic (firing temperature = T,, shrinkage temperature = TA
is refired at a temperature, T,’ (>Te), and the new shrinkage temperature, Ti,
is determined then for one value of the refiring temperature :
T, - T, T,’ - T,’
The experimental results further suggested that, for a wide range of firing
temperatures (700-950°C). the criterion for selecting the refiring temperature is
that (T,’-Te) should equal 2O-3O0C, the firing and refiring times being 1 hour.
An analysis of the sintering characteristics of London Clay provided a theoretical
basis for this result and indicated that it could be extended to other clays in
which the shrinkage is associated with sintering in the presence of a liquid phase
(Tite and Roberts 1969).
Consequently if refiring data is obtained for the ancient ceramics then the
“equivalent firing temperature” can be estimated using the relationship :
T,, = T, + ( T l - T i )................, .. ...... . .......(1)
providing a refiring temperature can be found such that (Te’- Tes) equals 20-30°C.
However this estimate is subject to further errors associated with firing times,
moisture absorption and bloating.
3.1.1. Firing time. The difference (T, - T,) observed for prefired clay specimens
also varied depending on the time for which the firing temperature was maintained:
typically, increasing the firing time from hour to 3 hours increased T, by
approximately 30°C. Consequently if a refiring time of 1 hour is chosen for
ancient ceramics, the “equivalent firing temperature” must be presented in the
form: T,, for 1 hour. However, in practice, a lower temperature for a longer
time or a higher temperature for a shorter time may have been employed.
3.1.2. Moisture absorption. After firing, water is absorbed from the environment
by the glass and amorphous phases present in clay ceramics and this absorption
gives rise to the phenomenon of moisture expansion. When the ceramic is subse-
quently heated the water is desorbed and contraction occurs. Most of this
absorbed water is removed by heating to 500°C but desorption continues at least
up to 1000°C (Vaughan and Dinsdale 1962) and as a result the shrinkage
temperature observed during thermal expansion measurements can be reduced
(Cole 1959).
Absorption of moisture during burial can therefore lead lo a low estimate of
the ‘‘equivalent firing temperature” of ancient ceramics. An indication of the
effect of absorbed moisture can be obtained by subjecting specimens to steam
treatment after refiring. Steam treatment provides greatly accelerated moisture
absorption: 100 hours of steam treatment is equivalent to four years of natural
moisture absorption and thus achieves a practically saturated moisture content
for the ceramic (Freeman and Smith 1967).
3.1.3. Bloating. When a clay is fired a large irreversible expansion, referred
to as bloating, can occur at temperatures greater than that necessary for the
commencement of vitrification. Bloating is caused by the expansion of gases,
ARCHAEOMETRY 135

produced by the decomposition of organic matter, sulphides or carbonates and


trapped in the liquid phase formed during vitrification.
If these gas-producing compounds were not removed during the original firing,
then bloating can occur during thermal expansion measurements on ancient
ceramics, thus masking the shrinkage associated with the resumption of sintering
and leading to a high estimate of the “equivalent firing temperature”. However
the existence of bloating can easily be detected since when it occurs the expansion
during heating is greater than the contraction during cooling in the 700-900°C
range.
3.2 Firing in the pre-vitrification range (T, <T,)
A precise estimate of the “equivalent firing temperature” of ancient ceramics
is not possible since, even when shrinkage occurs during firing (figure l), it does
not obey thc sintering kinetics associated with vitrification and equation (1) is not
therefore applicable.
In addition since clays with a high illitic* content do not contract until
Vitrification begins. it must be emphasized that even though the shrinkage tempera-
ture observed for ancient ceramics is greater than 8OO”C, this temperature was
not necessarily reached during the original firing. Typically for Farnley Fireclay
the shrinkage temperature was approximately 870°C for unfired clay specimens
and for specimens prefired at temperatures 1654°C and 785°C) below T, (Table I).
However examination of the expansion of the ceramic in the 450-550°C range
can provide information on this problem since the expansion differs depending
on whether the clay minerals have retained their chemically combined water (a1).
have been fired to temperatures less than T, ( a 2 ) or have been fired in the
vitrification range (a3). For the clays illustrated in figure 1, the ratios of the
expansion in the 450-550°C range ( a l / a 3 and aZ/aJ were both greater than unity;
the values for the Farnley Fireclay being 3.7 and 1.6 respectively.

4. FIRING TEMPERATURE OF ANCIENT CERAMICS


Firing temperature determinations were undertaken on a wide selection of
ancient ceramics, ranging in age from 5000 B.C. to 1650 A.D. and in provenance
from China, the Near East, Africa and Europe.
4.1. Procedure
The specimens were cut from the pottery sherds to the appropriate size and
shape, typically parallelopipeds 2.5 X 1.0 x 1.0 cm, and any glaze or paint
present was removed from the surface. They were then dried at 130°C for
5 hours.
For the majority of the ceramics, thermal expansion measurements were made
on two specimens: the fused silica extension rod or “Baudran type” dilatometer
being used depending on the shrinkage temperature anticipated. With the first
specimen the shrinkage temperature (Ta), defined as that temperature at which
a contraction of 0.2 micron from the maximum length had occurred, was
determined. The thermal expansion measurements were continued up to at least
850°C.irrespective of the shrinkage temperature, in order to obtain an indication
* The clay pineral, illite, is sometimes referred to as hydrous mica since it is structurally
similar to mica but contains less potash and more combined water.
136 ARCHAEOMETRY

of the sintering characteristics of the sherd and measurements were also taken
whilst the specimen was cooling down to room temperature.
With the second specimen, the shrinkage temperature (Ta)was again determined
and the specimen was then refired, while still mounted in the dilatometer, at
temperature, T l , for 1 hour: the refiring temperature was chosen on the basis
of the sintering characteristics observed for the first specimen. After cooling to a
temperature approximately 300°C below the refiring temperature, the specimen
was reheated and the new shrinkage temperature, T:, was determined. The
“equivalent firing temperature” was estimated using equation (1).
In order to obtain some indication of the effect of absorbed moisture, a
selection of the specimens were, after refiring and measurement of shrinkage
temperature, T l , subjected to steam treatment at atmospheric pressure for
100 hours. After drying at 130°C, the thermal expansion measurements were
repeated and a further shrinkage temperature, T,”, was determined.

FIG.2. Thermal expansion curves during heating and cooling of ceramic


specimens: percentage change in length (AL) versus temperature
(TC).
ARCHAEOMETRY 137

Finally a selection of the ceramics were examined using X-ray diffraction


techniques and differential thermal analysis (d.t.a.) in order to establish which
clay or subsidiary minerals were present.
4.2. Results
In Table I1 the results for the ceramics are presented: the values obtained
for T,. Tl,TL, T,, and T,". the ratios of the expansion (ah) observed during
heating to the contraction (a,) observed during cooling, in the 450-550°C and
700-900°C ranges, and the mineralogical data obtained from X-ray diffraction
measurements are listed for each sherd.
Figure 2 shows a selection of the thermal expansion/contraction curves
observed during the heating and cooling of the first specimen. Typically (figure 2a)
the contraction starting at approximately 100°C was due to loss of absorbed
water while the accelerated expansion occurring at approximately 570°C was
caused by the - p phase inversion of the quartz present in the clay.
(Y

Since the estimation of the "equivalent firing temperature" from thermal


expansion data depends on whether the vitrification range of the clay was reached
during the original firing and since vitrification can start at temperatures as
low as 70O0C, the sherds were initially divided into two groups:-
Group A: Ta>700"C
Group B: T,<70OoC
In addition a third group (Group C), consisting of sherds which contained
calcite as an impurity, was distinguished.
4.2.1. Group A sherds. For the sherds in this group it is probable that the
vitrification range for the clays was reached during the original firing since T, was
greater than 700°C and in addition the expansion in the 450-550°C range was
unchanged after heating to 850°C or above ( a h l a c z 1). (Section 3.2.)
For sherd A2, the shrinkage temperature (T):' observed after steam treatment
was approximately 70°C lower than that observed before steam treatment (T,,').
Consequently it is probable that moisture absorbed during burial had reduced
the initial shrinkage temperature (T,) by a similar amount and therefore the value
of T. used in estimating T,, should be 790-820°C rather than 720-750°C (Section
3.1.2.). However the reduction in shrinkage temperature Or,' - T,"), observed for
the other sherds (Al, 8, 12 and 13) in this group subjected to steam
treatment was only 516°C. The dramatic effect of absorbed moisture for
sherd A2 arose because of its exceptionally small thermal expansion coefficient
(a = 0.6 X per "C) in the 600-800°C range. Since no other sherd in this
group had such a low value for the thermal expansion coefficient (typically
a = 3 X lo-" per "C) then, with the exception of sherd A2. the effect of
moisture absorbed during burial could be adequately taken into account by adding
10°Cto the values obtained for T,.
Since the ratio of the expansion during heating to the contraction during
cooling (ah/ac) in the 700-900°C range was significantly greater than unity for
sherds A6, 10, 11, 14, 15, 16, 18 and 24 (figure 2c). it is probable that bloating
occurred during thermal expansion measurements (Section 3.1.3.). Consequently
the values obtained for T, are probably too high in these cases.
In the case of ceramics fired at very high temperatures an extensive glasspiquid
phase may have been formed. Therefore when the viscosity of this liquid phase
decreases during high-temperature thermal expansion measurements (figure 2b). a
OSL L- S6L LS L 6SL P3
LOL L- Z9L P IL OOL €3
- - - - oror< 23
- - LLL 13
(juasad a j p 1 ~ 3
) dnoq)
PE - (ELS) LS - €99 (OE9) t0€9) s0
- 829 o€- 85L 859 LE9 P0
- €95 OL - L69 EE9 op9 €0
LZ + a9 SI - LIL SW 9z9 z0
9v+ oz9 51 - 989 SE9
01 - EZO I €9- SLII 9801 - 8ZV
01 - EEOI op- 5611 EL01 - LZV
1’1 - tZ1 I z9 - LOZI 5811 622I 9zpT
I’I - SIII sz - 2611 opI I - SZV
P’I - €01I El- SEII 9111 6Wl PZV
01 - €21I 9z+ LPI I L601 - EZV
01 - LIII op+ LZI I ILOI - zzv
01 6‘0
- 5901 IE + 0601 PEOl pool
-
IZV
1’1
-
€801 S+ 9111 8LOI ozv
2’I - 9SI I L+ LLlI 6P1 I PPI I
- 6 IV
E’ I 086 IZ- z901 I001 81V
Z’I 01 6LOl E+ SIII 9LOI 1001 LIV
E’1 6‘0 SWI 0 I601 SPOI - 91V
p’1 1’1 oz0I El - 8901 EEOI p66 s IV
L‘1 01 986 EE - LSOI 6101 686 PIV
01 0 1 8% L+ I001 1% 8% EIV
01 8’0 E% t- I001 996 oP6 ZlV
61 01 256 LZ + EL6 526 9E6 I IV
9E L‘O 9C6 It+ 8% 526 €26 01v
60 01 PZ6 8+ LS6 916 LO6 6V
01 0’I 026 P+ E% 9t6 206
~ ~~ 8V
01 L’O 6% IZ + 986 826 168 LV
s*I 01
-
PE6 SI+ 056 616 888 9v
01
- 0 1 6 I6 6E + 856 088 8L8 sv
1‘0 - 16L L€ - p88 828 Zt8 PV
- 01 - 8S9 6P - ELL LOL ZSL €v
1‘1 OL + EEL 81 -
- 818 ISL
I
01 S+ L99 68 - Z9L 9sL
3.006-OOL 3,oss-os9 30,’L 3.O.L
/
Jri Val
ARCHAEOMETRY 139

shrinkage due to the pressure exerted by the dilatometer components, rather than
to the resumption cf sintering, can occur. This effect was observed with a specimen
of electrical porcelain (T, = 1230°C) for which T, was increased by as much
as 100°C by reducing the pressure of the dilatometer rods on the specimen.
Consequently it is probable that the values obtained for T, were too low in the
case of some sherds fired at high temperatures (e.g. A25, 26, 27 and 28).
In principle therefore the “equivalent firing temperature” of Group A sherds
can be estimated using the refiring data in conjunction with equation (1). Errors
of 2 5 ° C in the measurement of TI, and T:, together with the approximate
nature of the correction for the effect of absorbed moisture, suggest that for a
single specimen the error in T,, is *2O0C, when the criterion that (Te’-T,J
equals 20-30°C is adequately satisfied. However when this criterion is not satisfied
(sherds A l , 3, 4, 6, 1.1, 18, 22, 25, 26, 27 and 28) or when the values of T, for
specimens cut from a single sherd differ by more than lO”C, the error in T,,
is greater. Furthermore no correctirn for bloating or the presence of an extensive
glass/liquid phase is possible.
4.2.2. Group B sherds. For the sherds in this group, it is unlikely that the
vitrification range of the clays was reached during the original firing since T, was
less than 650°C and in addition, with the exception of sherd B5, the expansion in
the 450-550°C range was reduced after heating to 850°C (i.e. CY,,/CY, > l ) . Conse-
quently determination of T,, using equation (1) is not valid and the thermal
expansion data must be interpreted in conjunction with information concerning
the mineralogical composition of the ceramics.
Sherd BZ. X-ray diffraction and d.t.a. measurements established that the
hydrated clay minerals, kaolinite and illite, were present and consequently
it is probable that the firing temperature was less than 500°C. However the
possibility that a higher firing temperature was employed and that rehydration
of the clay minerals occurred as a result of moisture absorption during burial
(Grim and Bradley 1948, Hill 1953), cannot be neglected. The value obtained
for T, provides no useful information since it was determined by the com-
pletion of expansion associated with dehydration of illitic clay minerals and
with the a - p quartz inversion.
Sheds B2, 3 and 4 (figure 2d). X-ray diffraction and d.t.a. measurements
established that no hydrated clay minerals were present and therefore it is
probable that the firing temperature was greater than 500°C. Since the value
obtained for T., provides no precise information, the results merely suggest
that the firing temperatures were in the 500-700°C range.
Sherd B5 (figure 2e). Shrinkage started below 100°C and was continuous
except for a very slight expansion at approximately 570°C due to the a - p
quartz inversion. The shrinkage was presumably associated with desorption of
moisture which had been absorbed during burial; similar results were observed
by Riley (1965), working with Egyptian ceramics. It is therefore only possible
to suggest a firing temperature in the 500-700°C range on the basis of the
absence of hydrated clay minerals and the fact that shrinkage became more
rapid above 650°C.
4.2.3. Group C sherds. Thermal expansion data does not provide a basis for
estimating T,, since with these sherds either the expansion prior to shrinkage was
extremely irregular or the shrinkage was followed by a large expansion (figure 2f).
X-ray diffraction and d.t.a. measurements established that calcite is present in
TABLE 111
Sherd Archaeological data Equivalent p i n g
no. Provenance Period 1Age Type temp. (T,* C)
Near East and Mediterranean Ceramics
A4 Mersin, Turkey Neolithic (c. - 750- 820
c1 Byblos, Lebanon Neolithic (c. So00 B1C.j - <800
A15 Arpachiyak, Iraq Chalcolithic (c. 4500 B.C.) Halaf ware 970- lO5Ot
c2 Chagar B a r , Syria Chalcolithic (c. 4500 B.C.) Halaf ware <800
A19 A1 'Ubaid, Iraq Bronze Age (c. 3500 B.C.', 'Ubaid ware ii40-ii80
A20 Ur, Iraq Bronze Age (c. 3500 B.C.j 'Ubaid ware 1070- 1110
B5 Erimi, Cyprus Chalcolithic (c. 3000 B.C.) - 500- 700
A14 Hala Sultan Tekke, Cyprus c. 1300 B.C. Mycenaean ware 940- 10007
A16 Myrtou, Cyprus c. 1300 B.C. Mycenaean copy 1030-10707
A17 A1 Mina, Turkey c. 500 B.C. Greek Attic ware 1OOo-1100
A18 A1 Mina, Turkey c. 500 B.C. Greek Attic ware 940- lOlOt
Roman Ceramics
c3 Dragonby, Lincs. Iron Age-Roman - <800
B3 Catterick, Yorks. Roman Spanish amphora 500- 700
B4 Castor, Hunts. Roman BIack burnished ware 500- 700
c4 Bainbridge, Yorks. Roman Calcite gritted ware <800
A6 Stibbington, Hunts. Roman Grey ware 900- 9607
A1 Hartshill, Wa+cks. Roman Mortarium : kiln 14a 600- 700
A8 Hafitshill, Wamcks. Roman Mortarium : kiln 5b 900- 950
A12 Hartshill, Warwicks. Roman Mortarium: kiln 15 930- 990
A13 Hafitshill, Warwicks. Roman Mortarium: kiln 2 950- 990
A7 Catterick, Yorks. Roman Colour coated ware 910- 980
A21 Catterick. Yorks. Roman Central Gaulish Samian 1020- 1090
A22 Catterick, Yorks. Roman Central Gaulish Samian 1 100- 1150
A23 Catterick, Yorks. Roman South Gaulish Samian 1100- 1150
A24 Lezoux, France Roman Central Oaulish Samian 1030-11307
Post-Roman Ceramics
B2 Baston, Lincs. Saxon - 500- 700
A10 Ipswich, Suffolk Saxon Thetford ware 920- 9607
A3 Olney, Bucks. Mediaeval - 620- 740
A2 Laverstock, Wilts. Mediaeval Kiln 2 750- 820
A1 1 West Cowick, Yorks. Mediaeval Kiln 1, level 4 940- 990t
A5 Pottersbury, Northants. c. 1650 A.D. Kiln 1B 910- 950
A9 Halifax, Yorks. c. 1650 A.D. Pule Hill kiln 910- 950
Miscellaneo~m
B1 Ilpox, Nigeria - Clay figurine <500
A25 China Sung dvnastv Porcelain i070-1140*
A26 China Sung (Chek:ang) dynasty Celadon 1070-1190'
A27 China - Porcelain : provincial 970 - 1060'
A28 China - .
Ching~.(K'ang- Hsi). dynasty Porcelain 960- 1050+
?Bloating: T,,possibly too high 'Extensive glassiliquid phase: T., possibly too low
ARCHAEOMETRY 141

the sherds. Since calcite decomposes in the 750-850°C range, it is probable that the
firing temperature was less than 800°C. However the possibility that a higher
firing temperature was employed and that hydration and subsequent recarbonation
of dissociated calcite occurred during burial, cannot be neglected. This uncertainty
could be resolved by optical examination of the sherds in thin section since, by
this technique, it is possible to distinguish the fine granular mass of the recarbonated
form from the unaltered calcite (Tobler 1939, Felts 1942).
4.3. Discussion
In Table 111, the values for the “equivalent firing temperatures” are presented
in the form of a temperature range which was determined on the basis of the
above considerations (Section 4.2). The archaeological data is included and
the sherds are grouped according to age and provenance.
From these results, i t can be seen that firing temperatures ranging from
500°C to 1200°C were used in the manufacture of ancient ceramics and that
low or high firing temperatures were not confined to particular periods or parts of
the world. In addition the following tentative comments can be made with respect
to specific groups of sherds : -
(i) The high firing temperatures (l05O-l20O0C)used in the manufacture of
Ubaid pottery (A19 and 20) confirms the results of Tobler (1939)who suggested
that some Ubaid pottery, found at Tepe Gawra in Iraq, had been fired to
1200°C since “grains of orthoclase in the sand filler had melted and had been
coated with a glass”.
(ii) Similar firing temperatures (950-1050°C) were employed in the manufacture
of pottery from Mycenae, itself (A14). and locally-produced copies found in
Cyprus (A16).
(iii) The firing temperatures employed during the Roman period ranged from
500-700°C for the coarse pottery, such as amphora (B3). to 1000-1200°Cfor
the high class Samian ware (A21,22,23and 24).
(iv) During the post-Roman period, a similar range of firing temperatures
to that for the Roman period was employed except that there were no ceramics,
in the group selected for this work, comparable in quality with the Samian ware.
(v) Although the results suggest that the firing temperatures for Chinese
porcelain (A25-28)were in the 950-1200°Crange, it is possible that because of
the presence of an extensive glass/liquid phase these values are too low.
Ancillary data for the interpretation of thermal expansion results was obtained
from the temperature measurements made during the firing of full-scale replicas
of Romano-British pottery kilns (Mayes 1961 and 1962). These kiln experiments
indicated that satisfactory ceramics could be produced if temperatures greater
than 850°C were maintained for approximately 1 hour and hence the values
for T,,, presented in Table 111. were calculated assuming an original firing time
of 1 hour.
In addition the kiln experiments showed that at the peak temperature (950°C),
the temperature at different points on the kiln floor varied by at least 100°C.
This observation, together with the fact that different values of T, were observed
for specimens cut from a single sherd. indicates that it would be dangerous
to aiiach too mu& arciiaeoiogical significance to a precise vaiue of T,, obtained
for a single sherd. This warning is especially important when the sherd is found
on the kiln site itself (eg. A l , 2, 5, 8, 9, 11, 12, 13 and 24) since in this case
142 ARCHAEOMETRY

the sherd has a greater probability of being a “waster”: that is a sherd which
has been over- or under-fired.

5. CONCLUSIONS
From the results presented in sections 3 and 4, it is apparent that the validity
and accuracy of the theimal expansion method for thc determination of the
firing temperature of ancient ceramics depends on the firing temperature employed
and the mineralogical composition of the clay.
When the shrinkage temperature is less than 700°C (i.e. no vitrification during
firing) or when the ceramic contains calcite, the thermal expansion data can only
provide a very approximate value for T,, and must be considered in conjunction
with the mineralogical data.
When the shrinkage temperature is greater than 700°C (i.e. vitrification during
firing) then a value for Teq, accurate to -t-20°C for a single specimen, can be
obtained from the thermal expansion data using equation (1); a firing time
of 1 hour being assumed. However even in this case, each result must be carefully
considered in order to establish the importance of further inaccuracies associated
with: -
(i) the possibility that the high value for T, is due to the fact that clays
with high illitic content do not contract until vitrification begins (Section 3.2).
(ii) failure to satisfy the criterion that (Te’- Trq)must equal 20-30°C (Section
3.1).
(iii) contraction, during thermal expansion measurements, due to desorption
of moisture absorbed during burial (Section 3.1.2).
(iv) bloating during thermal expansion measurements (Section 3.1.3).
(v) contraction, due to pressure exerted by the dilatometer components, when
the viscosity of the liquid phase decreases at high temperatures.
In conclusion therefore it can be stated that, with careful assessment of
the results, thermal expansion measurements provide a valuable method for
determining the firing temperature of ancient ceramics. Further measurements on
ceramics would add to the information on the performance of ancient kilns and
the technological capabilities of the potters, provided by the above results. In
addition measurements could be made on specimens taken from the kiln fabric
itself although assumptions concerning the firing time would be more hazardous
since each kiln has been used for a large, but unknown, number of firings.

ACKNOWLEDGEMENTS
In am extremely grateful to Professor J. P. Roberts for his active help and
encouragement in this work. I am indebted to Mr. R. Bluett. Dr. H. W. Catling,
Dr. E. T. Hall, Mr. and Mrs. B. R. Hartley, Mr. J . May, Mr. P. Mayes,
Dr. P. R. S . Moorey, Dr. J. Musty and Mr. F. Willett for providing the pottery
sherds and also to Woodside Brickworks Limited and Pike Bros., Fayle and
Company Limited for providing the clay samples.
Financial support for the project from the Science Research Council and an
I.C.I. Research Fellowship are gratefully acknowledged.
ARCHAEOMETRY 143

REFERENCES
Baudran, A., 1955, Bull. SOC.Franc. Ceram., 27, 13-24.
Cole, W. F., 1959. Austral. J . Appl. Sci.. 10, 346-363.
Cole, W. F. and Crook, D. N., 1962, Trans. Brit. Ceranl. Soc., 61, 299-315.
Felts, W. M., 1942, Amer. 1. Archaeology, 46, 237-244.
Ford, W. F., 1967, “The Effect of Heat on Ceramics”, Institute of Ceramics Textbook Series.
No. 4, MacLaren, London.
Freeman, I. L. and Smith. R. G.. 1967. Trans. Brit. Cerani. SOC.. 66. 13-35,
Grim, R. E. and Bradley, W. F., 1948;Amer. Mineral., 33, 50-59.
Hill, R. D., 1953, Trons. Brit. Ccram. Soc., 52, 589-61 3.
Kiefer, Ch., 1952, C.R. du 3‘ Congres Ceram. Intern., Paris, 55-69.
Kider, Ch., 1956, Bull. SOC.Franc. Cerani., 30, 3-24 and 31, 17-34.
Kiefer, Ch., 1957, Bull. SOC.Franc. Ccram., 35, 95-1 14.
Kingery, W. D., 1959,J. Appl. Phys., 30, 301-310.
Mariani, E., Peco, G. and Storti, C., 1956, Sibrium, 3, 143-157.
Matson, F. R., 1937, in E.F. Greenman, “The Young Site”, Occas. Contrib. Mus. Anthropology,
Univ. of Michigan, 6, 99-124.
Mayes, P., 1961, Archacometry, 4, 4-30.
Mayes, P., 1962, Archaeometry, 5, 80-92.
Perinet, G . , 1960, Trans. 7th Intern. Ceram. Congrcss. London, 371-376.
Riley, B., 1965, Private communication.
Roberts, I. P., 1963, Archaeonietry, 6, 21-25.
Terrisse. J. R.. 1959. Rei Cretariae Romanae Fautorum. Ac:a 11. 63-67.
Tite, M: S. and Roberts, J. P., 1966 and 1969, Progrcss Reports, Nos. 1 and 2, Ceramics
Department, University of Leeds.
Tobler, A. J., 1939, Excavations at Tepe Gawra, Vol. 11, Appendix H, 159-162. Univ. Mus.
Philadelphia.
Vaughan, F. and Dinsdale, A., 1962, Trans. Brit. Cerarn. Soc., 61, 1-19.

Вам также может понравиться