You are on page 1of 17

Geothermometry

GLENN
Department

of Physics.

from 40Ar/39Ar dating experiments


W. BERGER* and
University

DEREK YORK

of Toronto.

Toronto.

Canada

MSS IA7

We hare selected a variety of minerals


from three basic intrusions
within
the high-grade
metamorphic
terrane near Haliburton.
Ontario.
for K Ar dating.
By application
of Ar
4r
Incremental
heating experiments.
we conclude
that the discordant
K--Ar mineral
dates can best hc
explained by a slow-cooling
history.
From the btepwise heating experiments
we calculated
isotopic closure temperatures
using Dodsons
theory for slow-cooling
and from these constructed
a cooling curve. This curve uas then compared
to
Independent
c<>nstraints to test the overall validity
of our procedures and assumptions.
The results d~-c
geologically
reasonable
insofar as: (a) the maximum
calculated
temperatures
(from hornhlcndca)
01
1700 C are consistent with independent.
approximate
metamorphic
grid arguments:
(b) the Ar closure
temperatures for biotites (3X&400 C) are consistent with independent evidence. and (c) mineral\ with
older K ear dates have higher calculated
closure temperatures.
The resultant
geological
cooling curve
implies that a maximum
temperature
of >7OO C was reached before 1000 Ma ago and thereafter
the
temperature
decreased via changes in uplift and cooling rates to below 200 C by 700 Ma .igo.
We have discovered
that plutonic
plagioclases are very sensitive indicators
(comparable
to K-feldspars) of Ar loss and. therefore. can usefully extend the range of application
of the Ar Ar
stepwise
heating method. WC: also describe experiments
in which we habe tested the effects on the Ar syhtcmaka
of acid leaching of plagioclases

Abstract

INTRODLCTION
THY< MAJORIU

of Ar Ar studies up to now have


ignored the potential of this dating method as a rich
source of geothermometry
information.
In practice.
the 40Ar,3QAr incremental heating technique (MI RRIHUE and TCIRNI:R. 1966) permits one to perform both
a dating experiment and, at the same time. a diffusion
experiment. One can tahe advantage of this duality by
calculating paleotemperatures
from the inherent rate
parameters. ConsequentI!.
one may determine simulttrr~rou.\l~ a time and a temperature
and so can construct a geological cooling curve from the analyses of
several minerals.
The importance to earth science of a knowledge of
cooling histories is self-evident since the three variables. time. temperature
and pressure are at the root
of all geological processes. Not only does a complete
cooling cul-\e improve ones understanding
of t~rtic~l
tectonic processes through
inferences about uplift
rates. but also it can be a critical aid in the understanding of /~ori:or~ttr/ tectonic processes via the dating of palcomagnetic
apparent polar wander (APW)
paths (B~KGIK 1.t ~rl.. lY7Y: BI:RCZR and YOKK. 19X1)
with their consequent
inferences about continental
drift rates.
There are other geothermometers
in use. or with
potential for use. such as petrochemical
and stable
isotope systems. The petrochemical
systems such as
the two-feldspar and two-pyroxene
geothermometers

* Present addro\\:
Llni\er<it\.
Burnahy.

Phy\~cs Department.
Slmon
B C.. Canada VSA IS6

Fraser

are pressure sensitive and so can be used us grobarometers (see Vol. 61 of 9r~t~ic~rt ,Ilir~t~trloyi.~,t for
papers on this topic). These indicators can give maximum temperatures or pressures of processes. but cannot give the time variable. The stable isotope geothermometers
such as the oxygen-isotope
two-mineral
systems are pressure insensitive and can yield the temperatures of crystallization
of phases in igneous rocks
but. as emphasized
by Dovso~ (1973) and FAI.RI.
(1977), they often record equilibrium temperatures
in
metamorphic
rocks that are much below the peak
temperature
of the metamorphism.
Like radiogenic
daughter systems, these two-mineral
oxygen isotope
geothermometers
can remain out of equilibrium
because of slow cooling. In such a situation. it is preferable to calculate the isotope equilibrium temprrature from laboratory calibration curves. and then use
this temperature to deduce a cooling rate from Dodsons formula for closure temperatures.
Our approach with the Ar Ar method is the
reverse of this: to assume a cooling
rate and use
Dodsons formula to calculate an isotopic
closure
temperature. The advantage of the Ar .Ar method.
apart from the insensitivit) of Ar loss rates to pressure (GILEITI and T~JLI.IS. 1977). is that the diffusion
parameters permit this closure temperature to he rtlutinely calculated
from the dating experiments.
H)
using a variety of minerals of differing Ar retentlvities,
we can calculate a geological cooling curve which in
turn enables us to verify our assumed cooling rates.
In principle, this process is also possible nith othrlradiogenic daughter or decay-product
dating methods
such as the Rb Sr and fission-tl-acl\ tcchniclue\.

GLENNW. BERGERand DEREKYORK

796

Our reasons for using Dodsons theory of closure


temperatures have been stated earlier (BUCHANrt al..
1977). Since then we have shown how such Ar closure
temperatures can be used to constrain the interpretation of paleomagnetic poles from Precambrian terranes (BERGERet ul., 1979; BERGERand YORK 1979:
BERGERand YORK 1981) in situations where episodic
thermal overprinting
and/or slow cooling have
occurred. Other attempts
at using the *0Ar39Ar
method for geothermolnetry have recently been Lade
on meteorites. TURNER rt al., (1978) used an approximatidn to Dodsons formula in studies on ordinary
chondrites but because of the tight clustering of the
apparent ages they were unable to make strong statements about slow cooling. In his examination of episodic reheating in meteorites caused by impacts on
parent bodies TURNER(1979) introduced the concept
of effective outgassing temperature in order to draw
inferences from the observed age histograms.
In this paper, we will report the details of our procedure and discuss the advantages and limitations of
this method of geothermometry (Ar isotope thermochronometry~. Our rationale has been to adopt
reasonably conservative criteria for data selection and
rejection, to calculate closure temperatures from
internally derived rate parameters (knowing, however.
that experiments in LUCIIO do not necessarily mimic
nature), and to construct a cooling curve. If by proceeding in this fashion with different specimens of the
same mineral type and different minerals from the
same geological body we obtain temperatures that are
geologically reasonable and relatively consistent internally, then we accept them until independent evidence
can be produced to show them to be erroneous.

PROCEDURES
Thr sarnplrs
Our selection of samples was motivated by a wish to
resolve the so-called Grenville problem of Precambrian
paleomagnetics (e.g. BKHAN and DUNLOP, 1976). To this
end, we chose samples suitable for dating from the extensive collection made by Buchan and Dunlop. They had
observed that many of the rocks of the Haliburton mafic
intrusions (the Dudmon and Bark Lake dioritic and the
Glamorgan gabbroic-anorthositic
bodies) carried multicomponent magnetizations and that at bSt two of the
three different components could be thermoremanent magnetizations {TRMs). We did not know 4~priori whether an
episodic or slow-cooling (or a combined) thermal history
could have caused the reset TRMs. There were some pegmatitic events in the general area. but a high grade of
regional metamorphic facies (upper amphibolite to granulite) also implied great depths of burial and a consequent
slow-cooling history (MACIXTYREef al., 1967). The actual
times of intrusion of the three mafic bodies were not
known but they preceded the regional metamorphism (e.g.
BUCHANand DUNLOP.1976).
We selected hornblende and biotite because of their
known relative differences in sensitivity to Ar loss (e.g.
DALRYMPLEand LANPHERE.1969), K-feldspar and biotite
because of their potential usefulness as indicators of episodic reheating (BERGER,197% and plagioclase and nepheline because of our relative ignorance of their 4Ar,jAr

characteristics. For sample identification, we use the notation of BUCHANand DUNLOP(1976). For example, 82-149
specifies that sample 149 of the Haliburton coliection is
from site 2 of the Bark Lake diorite. The site locations are
given in BUCHANand DUNLOP(1976). We include general
descriptions of the thin sections of the rocks we &OS? for
dating in an appendix.

For
Ar extractions, the mineral separates i>VQ! pure
generally) were placed in a high purity quartz boat within
a quartz tube that was heated externally by ;f Lindberg
Hevi-Duty furnace capable of reaching 1SOVC with an accuracy of 1C at the thermocouple. We estimate the actual
temperature at the sample to be uncertain bq SC (1~)
below 900C and by 3C (In) above 900C. To avod rapid
quartz sag and large diffusion rates of atmospheric Ar into
the vacuum system, we never heated samples above 1200C
in this system. This was sufficient to outgas 9X- loo,, of the
Ar from biotites and hornbiendes but may have left
1& 157:of the Ar remaining in the feldspars (BEXER and
YORK, 1979). However, this loss is of no consequence for
our purposes since, as is shown later, the most important
information from feldspars is obtained at lower temperatures. The length of the interval at each temperature setting
was about 45min. and in calculating the diffusion toe%cients we made allowances for the small delays in reaching
new temperature settings by increasing the uncertainty in
each step interval [generally 20-60 set (In)].
During the project a number of blank stepwise runs were
made. The volume of Ar (blanks bad an atmospheric
composition) typically varied from a low of 2 x ltt cm
STP at 600C to a high of I x 10-~ncm SIP at 12uoC
with an assigned uncertainty of about 2Ob. For the data
presented below we have not separated system-blank Ar
from other components except through the usual 6Ar correction. Making this separation would not significantly
alter the information contained in the age spectra because
the non-radiogenic component of 40hr was generally less
than 109,
of the total OAr.

~rrudiafiuf~.s
und rejkmce

~i~er~~.~

To monitor the horizontal gradient in the fast neutron


Uuence. we included within each can (holding 13 samples
including standards), 7 high purity Ni wires. Samples were
generally given about 150 MWhr of irradiation in hole SC
of the MeMaster research reactor pool (39Ar and Ar
concentrations were later corrected for any interrupiions rn
the irradiation). After a suitable delay time to allow safe1
handling the relative activities of the wire Ruence monitors
were measured by integrating the counts in the 81 I keV ;
peak of the Fe produced via the fast neutron reaction
*Ni(n. p) 5sCo 7

/f +

Fe

Integrated counts of about 150,uoOper wire were obtained


with a Ge (Li) detector and a Tracer Northern TN-11
system when deadtimes were less than lo?;;. Combined
with measurements of wire mass (typically 5 mg) accurate
to v 5 pg (lo), the relative counting statistics enabled us to
estimate the lo error in a relative fiuence correction at
0.3-o.4:;.
To verify the validity of this procedure. we Included
three samples of a reference mineral (standard) within
each can. In the five irradiations where this com~drison
was possible, the measured ratios 4oAr*/39Ar, (hereafter
called 40*/39K) for the individual standards. when corrected according to the Ni wires, agreed with each other
within analytical errors. The fluence gradient corrections to
individual samples were as large as 3;, in some ir..
radiations.
At the beginning of this project, we used a htotite
from the Obedjiwan nepheline syenite in Quebec for ;*

Geothermometry

from 40Ar, 3qAr dating

of (3.603 + 0.03 1)
standard,
having a Ar* concentration
x 10m4 cm3 STP/g (n = 6. la S.E.) and a K concentration
of 7.343 + O.OlS~, (n = 13. la S.E. or standard error of the
mean. This standard was used for hornblendes
B2-149 and
B5-139 (run t). biotites B12-415, and B2-149. K-feldspar
B2-149 and plagioclases
B12-45 and B2-149 (unetchedl
The remaining
samples were analyzed
with hornblende
G14-4 as a standard (see below).
During this project, we prepared a new laboratory
standard with an apparent age also near 1.0 Ga. We selected a
hornblende
from nepheline bearing rocks adjacent to the
Glamorgan
gabbro complex because of their high hornblende content and good optical appearance.
A detailed
thin-section
description
of this hornblende
(Gl4-3) and
the other major constitutents
of the rock is presented
in
the Appendix. Three bottles of this hornblende
were prepared by coning and quartering
after careful and repeated
purification.
The results reported here pertain to battle I.
The isotope dilution
measurements
of Ar*
and flame
photometric
determinations
of K are as follows: Ar* =
(6.279 & 0.048) x lo- cm3 STP:g (II = 8, la S.E.. subsamples 245 mg) and K = 1.2200 + 0.0074,, (PI = 6. 1~
S.E., subsamples
2 100 mg). The resulting average apparent
age, when these values are combined
with the results of
two incremental
heating experiments,
is 995 & 8 (la) Ma.
using the revised decay and isotopic constants
for K
(STEIC;EKand J,&K.
1977). We should point out that we
used only the aforementioned
Ar* and K values in the
age equation (BFRc;~R. 1975) for the data presented below,
giving an age of 993 Ma.
In a comparison
of this hornblende
with a reference
hornblende,
hb3gr. supplied by Zartman
G14-4 gave an
Ar,Ar
age of 1002 Ma (BOTTOMLEY. personal
communication).
agreeing with the conventional
age of Y93 Ma
calculated from the above Ar and K concentrations.
Similarlj. simultaneous
irradiations
of the Obedjiwan
biotite
and hornblende
hb3gr showed excellent agreement (HAhES
and YORK, 1979). The principal
advantage
of hornblende
G14-4 over the biotite is that its lower K content and finer
mesh size (- 70 + 120)ensure smaller subsampling
errors
because 60 mg of the hornblende
provides about the same
volume of Ar* as IOmg of the biotite. Additionally,
the
Ar* and K concentrations
for hornblende
G14-4 art: vcr}
reproducible.
Also. the irradiated
hornblende
is easier to
handle because it develops less static charge than the biotite, reducing mechanical
loss during transfers.
A relatively detailed age spectrum
for this hornblende
(Fig. I) shows it to have an irregular release pattern. Houof the fofu/
ever, as is well known. the reproducibility
joArt and K concentrations
is more important
for a reference mlneral than the presence or absence of a plateau, if
one extracts the Ar in one step.
The nephclinc in thib rock was analyzed to test the concept that this mineral might be a sensitive indicator of 10~
temperature
events (ioRK tf d., 1970) but we can draw no
conclusions
from the irregular age spectrum because of the
possible
presence
of plagiolcasc
inclusions
within
the
sample (see Appendis)

In the calculation
of ages. we used the correction
factors
for interfcrencc
isotopes listed in BERNR (1975). We also
used the decay
and isotopic constants
of STEI(;ER and
JAC;FR (1977) (i,, = 4.962 x 10 yr->
EL, = 0.5X1 r
IO- I yr . K/K = 0.01 167 at. ,). Our analytical errors
in isotopic ratios are quoted as 10 S.E. and have been
propagated
quadratically
from isotopic peak heights extrapolated
to zero-time
conditions
in the digitized
and
automated
mass spec~rometric
analyses. Errors in absolute
volumes in the incremental
heating experiments
should be
valid for within-lab comparisons
but do not include possible hctween-lab
error\.
One point we wish to make about errors in stepwise

797

experiments

heating experiments
is that the analytical error in the mtegrated or total 40*/3YK ratio, and hence total age, is often
smaller than the smallest error for an individual step. This
is especially true for samples that yield plateaus and for
which large numbers of steps with approximately
equal
volumes of Ar are obtained. An example is given later. This
error contraction
effect is potentially
useful for high precision comparison
of relative ages of samples irradiated
together. We mention this effect because it has gone unremarked in the Ar, Ar iiterature.

We have calculated
an isotopic
from Dodsons iterative formula:

closure

temperature

7;

where the cooling time constant


r = --RT (/?I). E is an
experimental
activation
energy, R the gas constant.
.A
mainly a geometrical
parameter.
Do i? IS a prc-exponential
frequency factor, ~1is a characteristic
dimension
of the
mineral grain. and ? is the cooling rate at ;I particular
temperature
7. The advantage
of the use of this formula
for Precambrian
erogenic belts is that the closure temperature is insensitive to reasonable
changes in the assumed
cooling rate. For example. a change in ? by a factor of ten
changes r. by about IO,,. Errors in T, were calculated
with
a formula similar to Dnr>so~s ( IY73).
The diffusion parameters
E and U, o were calculated
from least-squares
fitting of diffusion
coefficients
to a
straight line in an Arrhemus plot. .4n example will be discussed below. The rate parameters
D LI wrre calculated
from the rclatlve volumes of Ar* released during the
step-heating
experiments
using the formulae
of FEUITI(;
and KALRITZLR (1966). For biotites and hornhlcndes.
we
used the formulae corresponding
to cylindrical
geometry
while for the other minerals. WC assumed spherical
geometry. The assumption
of cylmdrical
geometry
for hornblendes IS reasonable
from a knowledge of the structure of
hornblende.
while for mica. GILI:I.~~I(1974) has shown this
geometry
best explains the diffusion data from hydrathermal experiments.
The structure
of the feldspar? cuggcsts the use of spherIcal geometry.
an assumption
upported b> the esperimcnts
of FOLA~II (1974).
Other workers (e.g. TCRMK cr a!.. 197X; TIXWR 1979;
OZIMA and TAKIGAMI. 1980) have calculated
internal rate
parameters
from the neutron-induced
3Ar on the supposition that recoil effects are negligible and that Ar initially has a uniform distribution.
unlike .lAr* in certain
cases. In this stud) we have compared ths rate parameters
from Art and Ar and ha\se found no significant differences in the derived closure temperatures
for biotite and
hornblende
in almost all cases. This is not surprising
since
most of these minerals ha\,c cxccllent plateaus in their agespectrum
plots (see bclou) :~nd hcncc these two isotopes
are highly correlated.
For the fcldspnrs we preferred .4r* for two reasons.
Firstly, because plutomc fcldspars arc generally tno-phase
systems. the phases distingulshable
by K, Ca ratios over
distances of less than one micron. one might cspect Ar
recoil effects to be significant. SecondI?. Fe have examined
both 34Ar and Ar* rate parameters
for the fcldspars and,
excepting plagioclase
G71-1 13 mentioned
in Table I. observed no differences in Ar closure temperatures
that could
be consldcred
significant in view of the overall limitations
of our assumptions
and procedures.
Because the dimensions of K-rich (Ca-poor) and K-poor
(Ca-rich) phases will generally differ. the Ar Isotopes will
likely he released at different rates from each phase. This
would complicate
the interpretation
of rate parameters
unless some attempt at isolating these components
is made.
An adbnntage of using 3qAr Instead of Ar* is that potenrelease components
tially one can partition
the Ar
and C,t-poolphase5
by us;lng the
between
Co-rich

<FLYNN W. BURGER and DEREK YORK

Ar .-Ar ratio (proportional


to K.Ca in the absence of
recoil). In general this has appeared to he successful when
interpreting
age-spectrum
plot\
from
extraterrestrial
samples. Nevertheless. there is at least one example of a
terrestrial feldspar where the Ar -Ar ratio is relativcl?

nent method
components.

uniform in spite of a known micropcrthitic


structure and a
lack of correlation of OAr* and Ar releases (BERGEK and
YORK. 1979). For such samples relative recoil of Ar and
Ar could reduce the usefulness of the ratio AI- -Ar.

The detailed isotopic data from the mcremental


heating experiments.
including the derived diffusion
parameters D/u. are available elsewhere+.
Table I summarizes the diffusion parameters E am.1
&ju.
statistical
parameters.
closure temprrature~
and corresponding
apparent ages. In this table. thi,
goodness-of-fit
of all linear correlations
ma> k~
assessed
by comparing
the statistical
Il;lramcter
SUMS (a minimized weighted sum of residuals called
S in YOKK. 1969) with II - 2 where II is the ntnnbcr of
points fitted. SUMS approximately
obqs
:L ch:2 degl-ees t!f freedom.
squared distribution
with II
We show in Table I the 95<, confidence limits f1.11
SUMS, that is, the range which should hc cxcct~lcd
with a probability of 7YC, and fallen short of with the
same probability. A SUMS less than the lower 11mn
implies the errors have been overestimated.
whereas H
SUMS over the limit means there is more scatter III
the data than can be explained with the .issigrled
errors.

Alternatively
one could partition either the Ar* or
Ar components using a combined graphical and numcri..
cal bootstrap technique (e.g. PEPIN ct rrl.. 1964) if the presence of two or more components IS suggested in a plot of
cc&me release vs temperature, as for the Haliburton
biotites. However. this approach is inextricably bound up with
assumptions about the behaviour of real samples irl VNC~,C,.
Consequently. as a tirst approximation.
we have used for
plagioclases a relative11 simple method (the multi-compo-

+ See NAPS document No. 03801 for 31 pages of supplementar) material. Order from ASIS/NAPS
c/o Microfiche
Publications. P.O. Box 3515. Grand Central Station, New
York. NY 10017. U.S.A.. remitting $7.00 for microfiche or
$5.(K) for photocopies. Cheques to he made payable to
Microliche Publicutionc.
Table I. Summary

Sums/n

of argon closure temperatures

S'FEPS=

E*

below) for partitionmp

AI- rclcacc

RESl L,lX

________.._
Sample a

described

(T,) and age-spectrum

dates

_
Do/a e

x
::0lUWXlt

DATE(M)

Tc("C) f

__
Hornblende
G14-4
(-7o+120)

4.8/4(.05-7.4)
131./10(2.2-18.)

750-960

97+5

472?45

860-1000

83+10

443t107

GZl-113

9.6/9(1.7-16.)

920-1060

(-7C+120)

2.8/6(.48-11.)

1015-1090

15/8(1.2-14.)

980-1090

G16-25
(-70+120)

B12-415
(-12ot150)

82-149

lOM3

548:32

957.617.1

14026

647+48

9/1.-18.

75i-4

442+43

24/6(.48-11.)

940-1030

112t-12

565f105

3.4/4(.05-7.4)

920-980

162+15

659*108

23./9(1.7-16.)

930-1075

14./7(.83-U.)

930-104s

4.6/6(.48-U)

1000-107s

4.9/7(.82-13.)

930-1080

85-139

11.5/7(.83-13.)

930-1080

(-7W120)

3.5/6(.48-11.)

1030-1080

3.8/4(.05-74)

995.6.

598.'7.

705-t56

989.817.5

LUL*I

692t68

988.29.

'W%J-1075(Pi

965.8e7.6

r ci t. 3 i

96it-1200

747t71w

623*36(R)

(-7o+120)

154t6

664*44(R)

995.t8.

178*1?

705*85(R)

98.5.?8.

1000-1150

76+6

449*62

500-700

75f-3

355e32>

1000-1100

57f.6

355t7

90+3

391t27>

Biotite
-_
Dl-190
c-35+70)

BlZ-415
c-14+35)

1.6/5(.22-9.3)

.03/3(.001-5.0)
.2/3(.001-5.0)

600-700

375138
(R)

357+9

Y09.3f7.0

rota1

916.W7.2

t0CaL

Geothermometry

from 40Ar/3Ar
Table

Sampll!

SUMS/n

STEPS '

l--

dating

experiments

(continwdl

Ed

Do/a e

82-149

17./4(.05-7.4)

950-1100

37i4

+105.
(24._2D )

(-14+35)

.1/4(.05-7.4)

550-700

74+1

(.27':;;)xlo11

G4-2.'

-/10(2.2-18.)

see text

216i63 352122
(N
355110

898.0t7.1

905.hih.l

2.4/4(.05-7.4)

1020-1150

.1/3(.001-5.0)

600-700

96.2

Glh-'5

5.8/Y(1.7-16.)

990-1160

63i3

(.27+:;;)x106

39336

c-14+35)

23./4(.05-7.4)

1000-1150

6919

(.26y$x107

419+100

1.4/4(.05-7.4)

550-700

62i3

(.35

322131

3.9/5(.22-9.3)

580-850

45il

1.2/6(.48-Il.)

650-900

47'1

(1.6;:;)x104

22558

2.5/4(.05-7.4)

800-920

5514

(.44'$106

268145
151ilO (R)

(-35+7(l)

Comment R

DATE(W)

Tc(CC) f

CR)
955.Dih.9

K-feldspar
B?-141

?30?18

(R)

8117.3'6.5

(-7o+l?o)
Plagioclase
BZ-14'3

El!-415
(-7O+l.!o)

RS-139
(-7rl+lio)

G?l-113
(-7ci+lJo)

Glh-li

(R)

.86/6(.48-11)

600-850

34-1

+3.2
(8.5
)
-2.3

1.5/6(.05-7.4)

800-920

4213

(.9+2.2)x103
-.6

201'36

3.4/6(.48-11.)

750-1050

3872

+50.
(53.
-26.)

179+21 (R)

6.3/5(.22-9.3)

800-1000

26-2

(.12+.19)
-.07

5.2/6(.48-11.)

720-960

49i2

(.34':;;)xlo5

.28/1(.OrI-5.)

760-840

36<2

+26.
(18._11.)

6.5/8(1.2-14.)

720-1050

3853

+2a3.
(9".-71. )

88136

802.'lrI.

iroo-9511(P)

787.+10.

hhn-1wo(P)

hi?.tl(i.

80~1-1~lI~0(P)

(R)

2Jh'?3 (R) 779.t12.


-__--_-____-1fl5+27

174141 (R)

7,n-Yhil(P)

540.'?11.

7:0-loin(p)

x511-9iUlP)

liilh

'5./9(1.7-16.)

720-1050

2112

(.:!8y;)x10-1

.96/3(.001-S.)

850-1000

62'6

(.41;+a7

314+72

x59:25.

4.4/6(.48-11.)

650-950

39+1

3
(.49+.39)x10
-.22

19h'lfi

881.

4.?/8(1.2-14.)

650-1000

22'Ll

(.59+::~)xln-'

(-7O+l?o)

614-4
(-7n+l:o)

h5(1-95('

62ill

The mesh size is given in parentheses.


h SUMS is a statistical parameter (see text) and n is the number of points fitted to a straight line rn an Arrhcnrus plot.
In parentheses
are the 95,, confidence limits to SUMS.
This column gives the temperature
( C) of the first and last steps of the range of those fractrons either- fitted to ri
straight lure or (for the multi-component
method. footnote f) used to calculate diffusion coefficients.
This experimental
activation energy has units of kcal,mol. All errors in this table are 1~.
Thus is the pre-exponential
frequency factor (set- r ) derived from an Arrhenius plot.
R denotes those temperatures considered reliable (see text). For the plagioclases, the dashed line separates temperatures calculated by the multi-component
(top. preferred) and one-component
(bottom) methods (see text). For plsgioclase
BZ-149 two results are shown for each method. The second result of each pair and the second date correspond
to the
acid-etched
sample. For plagioclase
G16-25, only the multi-component
result is shown since the other method yields a
temperature
near zero. For plagioclase G21-113. the data representing
the one-component
result were derived from Ar,
diffusion coefficients since those from 4oAr* coefficients scatter widely in an Arrhenius plot. All non-feldspar
temperatures
use r = 5 C/Ma. The remaining temperatures
(excepting plagioclase G14-4) use p = 0.5 C/Ma.
e Total means integrated date while the numbers give the temperatures
( C) of the first and last steps of the range of
fractions used to calculate the mean dates (weighted according to the fraction of j9Ar,). The errors in all non-total dates
are simple averages and include, as for total. the errors in the standards.
P indicates a plateau according to our & /I~X
detinition (see text).
Unlike the other dates. this is an arithmetic
mean with standard
error of the mean (SE.), taking no ;1ccount of the
volume of Ark in each step.

800

GWNP~W. BERGERand DEREKYORK


I

DlSCUSSlON
1020

In the following major subsections we discuss first


our data selection criteria, then the interpretatioil
of
the age-spectrum
plots, and finally the itlterpretation
of the Arrhenius diagrams.
criteria for Ar dosurc

Reliability

rrmpuxtrrtw

To minimize subjectivity in the calcuiution of closure tem~ratures


we adopted two ud hoc and conservative criteria For evaluating the reliability of a closure temperature.
Firstly, at least five points should
correspond to a reliable plateau in an age-spectrum
plot. Secondly. the same five points should lie on ;k
statistically well-defined straight line in an Arrhenitts
diagram.
The problem of establishing
isotopic criteria fog
distinguishing reliable from unacceptable plateaus has
been discussed by DALRYMPLF and LANWERI. (1974)
and FLECK et (11.(1977). This is not an abstract problem because
disregarding
statistically
significant

-r---~

BARK LAKE 82-149


t

-i

980

940
2
z
z

YW
a
-.,

& 860

PLAGIOCLASE

i
:

j:
!

820

780

i
0

05

s___i___d

IO

FRACTION OF 3gArk
Fig. 2. Age spectra for four coexisting minerals irotn tk
Bark Lake dioritc that were irradiated togolhcr
1020 -

960-

HORNBLENDE

900 3
B

I
y

NEPHELINE

840-

780-

720 I

Fig. 1. An age-spectrum plot for our laboratory reference


hornblende G14-4 and a coexisting nepheline separate.
Almost all of the Ar is released before IOOPC (the second
last step here). fn this and ail succeeding age-spectrum diagrams, the temperatures of some steps are shown in celcius
degrees. Likewise, the vertical heights of the bars represent
+ lo S.E. errors, suitable for comparing within-spectra
ages or between-spectra
ages of samples irradiated
together. The errors in the standards are nul included in
any of the age spectra errors. These two sampies were irradiated together.

anomalies can lead to incorrect or miskadmg interpretations. C)ur understa~lding is limited by the dit&
culty in isolating the effects on age-spectra of sample
histories or sample parameters
from the effects of
laboratory
variables such as sample pre-treatment,
heating schedules and lattice changes ire VUUH!
With this limitation in mind WC USC the [crm rcliable plateau for any sequence of five or mctrc heating
steps where all dates or all but one agree ~~iirhin 20
error limits. Obviously these errors must exclude contributions from the reference mineral. It is these within-spectra errors that are plotted in our age-spectrum
thagrams. A sequence of dates defines 3 reliable plateau if the single discordant
value represents d relatively small fraction of the total Ar.
Reliable plateaus are defined by the 7M>-YSO.I
interval for plagioclase BZ-14Y. the 710 1050 ( segment for plagioclase G21-113. hiotitcs B12-315 and
G16-25, and the first seven steps shown for hornblende B2-149 (e.g. Fig. 4). Hornblende
K-139 is
a special case to be discussed below. Other reliable
plateaus are those of biotite G4-21, L)l .-I40 and
B2-149, and those of hornblendes
B12-415. G4-X
(excluding step 960 C), G16-25 and the last five steps
of G21-113. K-feldspar B2-149 is considered reliable
for reasons discussed below. Unacceptable
plateaus
are shown by hornblende
G14-4 (Fig. I I and the
650-950C interval for plagioclase Ct4-4.

In Fig. 2 are the age-spectra


for four coexisting
minerals from a hand sample of the Bark Lake diorite
(possibly granod~oritic near site 2) and the presence of
three broad plateaus
in hornblende.
hiotitc and

Geothermometry

X01

from 40Ar, j9Ar dating experiments

K-feldspar is remarkable. While one can always elaborate a multi-stage


episodic thermal history that
could generate successive plateaus in different minerals, it is our contention that such a scenario is more
complicated
than necessary
to explain our data.
Specifically, this scheme becomes less likely as the
number of minerals
having different plateaus
increases.
In the following minor subsections,
we discuss
briefly the significance of plateaus in each of the minerals, mindful that the most compelling evidence for
slow cooling is not so much the detailed structure of
the age spectra but rather the existence of a smooth.
decreasing curve passing through the calculated closure temperatures--the
derived cooling curve.

There is evidence that undeformed biotites which


release their Ar in two pulses and have suffered partial Ar loss through episodic reheating will show diagnostic anomalies in the last half of their age spectra.
Most of the samples of BERGER (1975). HANSON rt al.
(1975) and some of those of DALLMEYER(1975) show
such anomalies.
MALUSKI (1978) has shown that
deformed biotites. on the other hand, exhibit more
complex and irregular release patterns.
The presence of a reliable plateau in the biotite of
Fig. 2 indicates that no thermal event later than
900 Ma could have reached a temperature
much
above 2: 3ooC, its calculated
closure temperature
(Table I). Other biotites from the Haliburton
intrusives also yield reliable plateaus and have significantly
different apparent ages between 900 and 950 Ma. This
fact rules out any thermal pulses in the 900-950Ma
interval.
We note in passing the between-pulse
deviation
shown in biotite G16-25 (Fig. 3) which is reminiscent
of results reported by TETLEYand MCDOUGALL (1978)
for biotites known to have experienced no significant
secondary thermomechanical
disturbance.
This type
of deviation representing a few percent of the total Ar
is not understood but we feel it does not detract from
the usefulness of our so-called diagnostic anomaly.
For a given rock body, the existence of biotite pairs
having significantly different plateau ages also supports slow cooling. In particular. from the Bark Lake
intrusives. samples 149 and 415 (Fig. 3) have integrated ages of 898.0 f 1.3 and 916.0 k 1.6 (la) Ma
respectively.
Because these two biotites were irradiated together, errors from the standard are omitted
in this comparison.
Incidently, the error contraction
effect (mentioned above) is shown clearly here since
the integrated errors are smaller than the smallest
errors in the individual steps.
The different integrated plateau dates for these two
biotites reflect different Ar retentivities.
Since both
samples have the same mesh size and approximate
natural grain size (Appendix), the distinct retentivities
may be related to differences in chemical composition.
Such a variation in micas in the phlogopite-annite

980

940 t

E/2-4/5

II

II

0.5
FRACTION

I,

IO

OF 3gArk

Fig. 3. Age spectra for the Haliburton


morgan,
identified
cal scales.
were 190

biotites. The GlaDudmon


and Bark Lake basic intrusions
are
by the letters G. D and B. Notice the offset vertlSamples 415 and 149 were irradiated
together as
and 25. The temperature
increments
were generally 4c60 C.

series has been suggested as a factor affecting Ar


retentivities (NoRw~~~J. 1974: GILETTI, 1974). but as
yet has not been firmly demonstrated.
Another biotite pair having significantly different
integrated and plateau dates is that of G16-25 and
G4-22 from the Glamorgan
complex (Fig. 3). The
maximum grain size of sample 25 is _ 1500 pm while
that of 22 is 5 500 pm. It may be that this distinction
in mesh size (reflecting differences in natural grain
size, Appendix) affected the Ar retentivities and therefore led to different plateaus.

All of the analyzed hornblendes except the standard


G14-4 are represented
in Fig. 4. Four of these
(samples 22. 25. 149 and 415) show reliable plateaus
over 90. of the Ar release. Sample 113 has a reliable
plateau over 60, of the Ar release while the ambiguous 139 is discussed below.
Two hornblendes (25 and 113) from the same body
have real differences between their plateau ages.
Because chemical composition
could affect the Ar
retentivity of amphiboles (ONIONS rt ~1.. 1969). the
data in Fig. 5 could be consistent with a slow cooling
interpretation
if this pair had different compositions.
On the other hand, because hornblendes
may give
false plateaus under conditions
of episodic thermal
overprinting
(BERGER. 1975) these age spectra by
themselves cannot be used to support an argument
for slow cooling.

f---l---l-

II00

OII0

HORNBLENDE_

--7--T---7--

--r--Y

HORNBLENDE B5 - 139
--

106O-

102o-

98 O-

I(

I(100

0
r
S
a

96 0

2
b

9 60

94 0

I(100

1 -.._i-

.J.-.-L

._.*

FRACTION OF 3sAr

_... __L___
II

t ig. 5. Age spectra for duplicate expcrimsnt~ iru Iloinhlrndc 139 from the Bark Lake diorite. These ++cieirradlated separately using different standards. lhc Isolated
vertical har indicates the appropriate between-spectra .i lcr
erl-ors for each step. The errors in the total or ultograted
apes include the errors in the standards. The lemperaturc
increments for run I were 20 C while those for ruri 1 \vzrc
10 C

s160
/

we lacked sufti&nt control of the system blanks 10


make precise corrections to the small PII- signals ir!
these experiments.

IOCIO-

96 IO*

G21-//3

9210I

0.5

IO

FRACTION OF 3gArk
hornblendea
F1g. 4 Age spectra fur the fiahburton
(excludmg G14-4). With the exception of sample 27. the
temperature increments were generally ahout 70 C. Notice
the offset certical scales and the hortrontal tirro~~ that
point to the appropriate scalv\.
Hornblende
B5-129 (Fig. 4) IS unusual in that it
shows a staircase pattern. We duplicated this run by
using IOC increments
instead of 70 C increments
over the bulk of the release pattern. We hoped to
thereby distinguish isotopically the staircase pattern
from a acceptable plateau. We also wanted to see the
effect of changing the heating schedule on the correlation of rate parameters
in an Arrhenius diagram
(discussed below).
The results of the duplicate are compared with the
original in Fig. 5. Because different reference minerals
were used, the errors appropriate
for between-sample
comparison
are represented
by the vertical bar and
the errors in the integrated ages. From the smaller
within-spectra
errors it is clear that this staircase pattern is reproducible.
Unfortunately
we could not use
isotopic correlation plots in this comparison because

We shall classify the age spectrum for K-feldspar


82-149 (Fig. 2) as a reliable plateau to a first approuimation (however. see comments below on an apparent staircase) because normallq this type of feldspar ip
so sensitive to thermal disturbances that one observes
a prominent minimum in the first half of the ~gc spectrum (BEKGEK, 1975). The presence of a plateau *II
the K-feldspar suggests that no thermal cvcnt greatcl
than _ 150 C (e.g. BEKGEK, 1975) afectcd this sample
since . X00 Ma.
It is interesting that ALBAK~.IX:or ul. ( 19%) have
observed modest plateaus in K-feldspars that from
independent
evidence are believed to have: romaincd
undisturbed since Hercynian time. This contntsts with
the presence of prominent minima in such age spectra
when there is independent evidence of thermomechanical overprinting
(BERGER. 1975: Ar,nARbl,t tr tri..
1078; MAINSKI. 1978) or when there is .rtr-onp tit,>picion of the same (BERGEKand YORK. 1979).
It is worth noting here that the fine structure in the
K-feldspar age spectrum (Fig. 7) may be a real consequence of very slow cooling. Specifically, the age spectrum appears to start low and rise to a well-defined
plateau (excluding the second last step) above the
X50 C step. The maximum range of dates here IS
-- 20~ 30 Ma. TURNER (1969) has speculated that as a
consequence
of slow cooling an ideal, sphericalgrained K Ar clock would yield an age spectrum mdistinguishable
from that corresponding
to ;I sphcri-

Geothermometry

from

Ari39Ar

dating

X0

experiments

model which has lost Ar by episodic


reheating. Furthermore. he suggested that the maxi-

cal-grained

mum spread in dates in such an age spectrum should


correspond to passing through a temperature interval
of l@?oC. Our derived cooling curve (below) imphcs
that this K-feldspar cooled at _ fC/Ma at _ 800 Ma.
For this cooling rate. the predicted range of dates
would be XHOMa
years. just what seems to be observed in Fig. 17.Clearly, this subtle effect needs confirmation by more precise stepwise heating experiments on this or similar feldspars.

An important result in Fig. 2 is the presence of an


approximate
plateau segment in the temperature
interval
_ 7GO-950 C for the plutonic plagioclase.
Because this plagioclase segment is concordant
with
that of the coexisting
K-feldspar,
we feel it has
chronological
significance. This implies that piutonic
plagioclases in basic rocks could be dated as sensitive
indicators of late-stage cooling or of episodic reheating. It also calls into question the validity of convcntional K-Ar dates from such plagioclases and from
gabbroic whole rocks.
This has been recognized by most workers using
the conventional K Ar method. Some have attempted
to overcome this limitation bJ- using correlation plots
of JAr.hAr vs JK:3Ar (e.g. HAYATSUand PALMER
1975; PALMER (jr u/., 1979). We were unable to treat
our data in an equivalent
fashion (40Ar.i3hAr vs
Ar,Ar1 because in only one case were more than
two plagioclases
irradiated
together,
and in that
example (GZI-I 13. Gl6-25. G14-4) the three samples
did not lie on a straight line in a correlation
plot
when corrected for our integrated system blanks.
We believe that the different plateau ages for those
plagioclases having plateaus in Figs. 6 and 7 (exccpting sample 4) can best be interpreted
as a conscquence of slow cooling and not as a result of a
sequence of slight thermal disturbances. In this regard
we speculate
that each plutonic
plagioclase.
iike
K-feldspars, will have a different effective Ar retentivity dependent upon the effective grain sizes of the
K-bearing phases -for example. the presence of diffcrent lamellae widths in the micron range.
It appears from Figs 6 and 7 that the useful range
of laboratory temperatures
for geothermometry
with
such plagioclases is ~700-1000 C. According to our
workmg definition the 650-950 C interval for plagioclase 4 (Fig. 7) is an unacceptable
plateau. This irregularity may be due to the likely occurrence
of
nepheline inclusions (Appendix) since the coexisting
nepheline separate has a very irregular age spectrum
(Fig. 1). This supposition
of cross contamination
is
supported by evidence from Fig. 8 discussed below.
In Fig. 6 sample 113. while superficially appearing
ragged. has in fact a reasonable plateau in the interval
7X!-1050C because all dates agree within Zcr except
the small step at 0.52 on the 3Ar axis.
It is interesting to note that these plagioclases can

PLAGIOCLASE
~~

11

11

11

05

FRACTION

1,

1.0
OF 3gAr

be classified according to their .Ar,3-Ar


ratios. In
Fig. 8 the Bark Lake samples have over-lapping patterns of the same form whereas samples Gl6-25 and
GZl-I 13 have different patterns and distinctly lower
K,Ta ratios. Sample G14-4 has much higher K:C2a
ratios and a unique pattern probably because of the
presence of contaminant
nepheiine inclusions (Appendix) that were not likely removed by grinding to
-70 + 120 mesh size, since the A+-Ar
ratios for
the coexisting nepheline separate are in the range
l&40.
In summary. it seems that the plutonic plagioclases
from the dioritic Bark Lake body provide more useful
plateau segments than those from the gabbroic-anorthositic
Clamorgan
complex. It is not known if
plagioclases from more acidic plutonic rocks would
be useful for Ar isotope geothermometry.
Other examples from dioritic rocks. SAN and RS6
of ALBAREISE er ui. (1978). are from the Eastern
Pyrenes which is a region of complicated
late-stage
thermomechanical
disturbances.
Their incremental

804

GLENN

W. BERGER and DEREKYORK


,

40. PLAGIOCLASE

PLAGIOCLASE

{
i

tar
4. G/4-4
l.O?
1
./

- 800

1
FRACTION

- 1000

-800

80
0

3gArk

Fig. X. Plots of ArKi3Ar (proportional to K.Cal against


cumulative fraction of 3Ar, released for all the Haliburk~n
plagioclases. Notice the offset logarithmic scales and
appropriate horizontal arrows. The errors ( & 10) include
estimates of variation in the sensitivity of the mass spcc.
trometer. Notice that the patterns for samples 140 I ?9 and
415 essentially overlap.

IO

0.5
FRACTION

OF

OF 3gAr k

Fig. 7. Age spectra for three of the Haliburton plagloclases. Notice the offset vertical scales and appropriate
horizontal arrows.

heating schedules are not detailed enough to be informative nor can one deduce 39Ar/37Ar ratios from
their data. PlagioclaSes from hypabyssal rocks do not
appear to be useful (STUKAS. 1977). Also, the least
deformed plagioclases (samples 142. 23 and 139) from
the Corsican granodiorite (MALUSKI, 1978) do not
have plateaus in the interval w 70&1oooC as do the
Haliburton samples of this paper. This may be
because of the complicated and poorly understood
late-stage history of the Corsican area. Furthermore
Maluskis samples have higher 39Ar/37Ar ratios in
this temperature interval than at other temperatures.
unlike the Bark Lake samples. Plagioclases from Precambrian dikes have given useful chronological information (HANESand YORK, 1979) but they show the
presence of alteration products (e.g. sericite, saussurite) both optically and in their high 39Ar/37Ar ratios
over the w 700-1oooC interval.

l~~fkrences,from

leaching

experiments

on pluqioclusrs

To further justify the assumption we have made


that all the reliable plateau segments observed in the
Haliburton plagioclases have chronological significance, we attempted to ensure that we were dating a
plagioclase phase and not a fine-grained K-rich alteration product such as sericite. Although there is
no optical evidence (Appendix) of significam alteration products and although (excluding G14-4) the
9Ari3Ar ratios also suggest the absence of any relntively K-rich phase contributing to the plateau seements, we decided to perform acid leaching experiments on certain plagioclases before irradiation on
the assumption that any such alteration product
would be readily removed.
We were motivated by two concerns. Firstly. we
had doubts (see Calculation of closure temperatures
above) about how accurately 39Ar/37Ar ratios reflect
K/Ca ratios for samples that often have micron-sized
lamellae of alternating K-rich and K-poor exsolution
phases. Secondly, if the ratios in Fig. 8 do faithfully
reflect K/Ca ratios we hoped to extend the temperature range of the useful plateau segment by eliminating the low- and/or high-temperature K-rich phases.

from Arj3Ar

Geothermometry

We used a 5, HF solution held at 5OC for 20 min.


following the example of EVERNDEN
and CURTIS
(1965) who studied volcanic feldspars.
We chose
plagioclase B2-149 because it has an independently
confirmed plateau segment (Fig. 2). and plagioclase
G16-25 because, in contrast. it has a much smaller
plateau segment (Fig. 7).
The results for B2-149 are presented in Fig. 9. The
most important point is that although the total age
has been lowered, the plateau age has not only been
left intact but also the plateau segment has been
extended in the age-spectrum
diagram. Despite this
the plateau occupies almost the same temperature
window (6OtSlOOO~C) as for the unetched
sample
(55@95OC) (see also Fig. IO). Evidently, the 40Ar* is
not from a fine-grained contaminant.
It also seems
that the scatter of the individual dates in the plateau
segment for B2-149 has been reduced. It is perhaps
not surprising that the etching has left the plateau
dates unchanged if the Ar* and 39ArK associated
with the plateau segments are released from contiguous micron-sized
lamellae.
because
etching
(e.g.
BERNER and HOLDEN. 1977) might be expected to
always leave some region of contiguous
lamellae
untouched

dating experiments

ETCHED

---

Gl6-25

NORMAL

ETCHED

- -- f/4/9

0408

2 /6 Ma/
_+191

/I

1
I

II

0.5

FRACTION

4-

i
i

600

NORMAL

PLAGIOCLASE

1800

O-

xo_z

OF

IO

3gArk

Fig. 10. Age spectra for duplicate


runs on Glamorgan
plagioclase
G16-25. irradiated
separately
using the same
standard.
Other comments are as in Fig. 9.

2I-

PLAGIOCLASE
IO-

82-149

NORMAL -_19488f9
.-,
I: I:

ETCHED

2 Mc?l

::
:I;,

-- -- f835.999.81

P jl:
.I

IO-

To summarize. it appears that to improve the reliability of the closure temperatures


for plutonic
plagioclases one should concentrate the heating schedule on the internal _ 60@1OooC. In addition, acid
leaching may be helpful because it can extend the
plateau segment in an age-spectrum
plot. However,
there is a concomitant reduction in the 4oAr* concentration (e.g. -5 for Gl6-25 and -50 for B2-149)
which may influence the choice of irradiation time.
Biotitr

10 -

IO-

/
0

FRACTION

0.5

IO

OF 3gArk

Fig. 9. Age spectra


for duplicate
runs on
plagioclase
B2-149. irradiated
separately
and
ent standards.
The errors in the standards
are
the integrated
ages. but not in the individual
second run was leached with a warm solution
before irradiation.

Bark Lake
using differincluded in
steps. The
of HF acid

closure temperatures

The biotites present the strangest pattern in Arrhenius diagrams of all the minerals we have analyzed. It
seems to be typical of these micas that the rate coefficients define two approximately
parallel line segments
that are offset as in Fig. 11. This feature has not been
reported before and the reason for its existence is not
known. It may be that this kink is a manifestation
of
the breakdown of biotites when heated in a vacuum.
It is possible that such a feature could reflect changes
in the interlayer spacing as the water of composition
is driven off.
Whether this speculation is accurate or not, these
biotite patterns seem to provide us with two estimates
of Ar closure temperatures
that. as for Dl-190 in

806

GLENN

&?-I49

W, BERGEH and DEREK YWK

H6lE

623 t 36

K-F.7

256 t/8

375 t 38

D/-/PO B/O

__._

{ ~~~i~~

___1
*a

---I
.9

1.0

1000/T,

I.1

-__._._._.
i.2

_.A
ii

PK)

Fig. 11. An Arrhenius


diagram
for a hornblende
and K-feldspar
from the Bark Lake drorlte anu .i
biotite from the Dudmon diorite. Plotted are calculated diffusion frequency factors (D ,<?I for AI-* and
corresponding
laboratory
temperatures
(TL). The slope of the best-tit line yields an experimental
act1
vation energy E and the Y-intercept gives a pre-exponential
frequency factor DO/~. Errors in data po~nr\
and closure temperatures
(T,) are f IO. The closure temperatures
are calculated only from the solid data
points. For this biotite, the preferred closure temperature
is the weighted mean of the IWO possibic
estimates.

Fig. 11, often agree within ILT. For the biotites, we


have had to relax somewhat our criteria for reliability
because often the pre-selected heating schedule caused
each of the two Ar pulses to be released in fewer than
five steps. Hence we accepted any well-defined correlation of four or more points as reliable for biotites.
Because we do not know yet whether the first or
second pulse should be preferred for routine closure
temperature calculations, whenever our reliability criteria have been met WC have chosen the mean of the
two temperatures
(weighted by inverse variances) as
the best estimate.
We performed
duplicate incremental
heating experiments on two biotites using unirradiated samples.
We chose G4-22 because its closure temperature
seemed to diverge the most from the general trend of
the biotites, and G16-25 because its two estimates of
closure temperature seemed to disagree unacceptably
(Table I). The heating schedule was concentrated
on
the second Ar pulse simply because this represented
the bulk of the Ar.
For both samples there was little change in the
relative release of Ar as a function of temperature for
the first few steps even though we began the dupli-

cates IO&2ooC higher. From G&7.! we cxrra~~ed ten


fractions corresponding
to the second Ar p&c. .A,
shown in Table 1 the scatter in the Arrhenius plot war
so large that a least-squares
fit wazl not ever;
attempted. We therefore rejected the 420< &surc
temperature as unreliable. For G 16-25 the tnnr: point>
representing the second pulse do define a straight linl:
so that the 393~C closure temperature
IS considered
reliable.
There is independent support Ihr our biotltc closure
temperatures. In particular. we calculate that diffusion
parameters measured by No~woou ( lY741 on hydra.thermally heated micas yield Ar closure temperature5
in the range 29&380X? (for ? = 5 C,Ma). not tom
different from those considered
reliable in Iablc 1
(viz. 350-4OOC). From metamorphic
facies arguments
in a study on Alpine biotites, WAGNER rt tri. I lY77)
estimated an Ar closure temperature
of 300 + JO <.
By rather general arguments. DALLMEYEK (lY78r preferred temperatures
of 3m34.5 C. Others (see DALRYMPLE and LANPHERE, 1969) would prefer it much
lower closure temperature.
The geologically derived
estimate of 225C by TURNER and FORBES (1976)
depends on a very large extrapolation
and on .I

Geothermometry

from Ar?Ar

number of assumptions
that have not been aired in
the literature. Until these differences are reconciled,
we consider the reliable Ar closure temperature
for
biotites in Table 1 to be geologically reasonable. We
emphasize that comparisons
of independent methods
must use the same samples, if possible, because of the
probability that differences in chemical composition
and natural grain size affect mica Ar retentivity.

For the hornblende in Fig. Il. the first five points


represent i 3,, of the Ar* and also provide no significant chronological
information, as is typical of Ar
released below -900 C for most hornblendes
(e.g.
BERGER. 1Y75). The next seven points represent
1900,,
of the Ar and define an excellent straight line, and
correspond to the first seven steps in Fig. 2 (a reliable
plateau). The last few percent of the Ar released from
the Haliburton
hornblendes
often have very large
errors in Arrhenius plots and do not lie on a straight
line nor on a plateau in an age-spectrum plot. Changing the cooling rate by a factor of ten from the
assumed 5CiMa changes the closure temperature
by
less than 30 C.
Because of the ambiguity of the age-spectrum
of
hornblende
B5-139 in Fig. 4, we performed a duplicate run with a slightly different heating schedule
(Fig. 5). The corresponding
rate parameters are plotted in Fig. 12. The data from run 1 show a SUMS
value near the Y5,, confdence limit. For run 2 the six
fractions
in the approximate
range 103~1090 C

dating experiments

(steps between 0.35 and 0.92 in Fig. 5) detine a


straight line with a corresponding
closure temperature indistinguishable
from that of run I. For these
reasons we tentatively consider this closure temperature to be reliable.
Another difficulty one may encounter in attempting
to calculate reliable closure temperatures
is illustrated
by hornblende G16-25. In spite of an excellent plateau
(Fig. 3). there is no
over 90,, of the age-spectrum
well-defined straight line for the corresponding
data
in an Arrhenius plot. Therefore we consider the corrcsponding closure temperature to be unreliable.
Although
our reliable closure temperatures
for
hornblendes
are 1~200
C higher than those estimated by other arguments (see DALR~WI.E~ and LA\PHERE. 1969; DALLME~ER. 197X). we bclickc rhq art
geologically reasonable
because they xe consistent
with the estimated maximum temperature
of metamorphism for this area of r600-700 C from metamorphic grid arguments (CH~SUUK~H. Ic)71). Furthermore. the peak of regional mrtamorphlsm
III the
Granville Province is thought not to have heen contemporaneous
throughout
the prokincc. and may
have occurred near 1050 Ma in the Haliburton Hiphlands (MACINTYRE tt trl.. lY67). consistent Mith our
preferred geological cooling curve (bclo~). Finally. In
view of the probabledependence
of Ar retentivity m
amphiboles on chemical composition
(ONIOM
(I t/i..
1969). any attempts
at independent
confirmation
of
our hornblende
temperatures
should uht identical
samples.

TL(C)
II00

I I

1000

507

800

900

HORNBLENDE

-5

E35- 139
RUNf/o)

RUN 2 f.I

N5

3.5

664 2 44

705?85

cc0
2
-o -6
-

1000/T,

(OK)

Fig. 12. An Arrhenius diagram for duplicate runs on the hornblende


in Fig. 5. SLIMS ib ;I stattatical
parameter
(text) and n is the number
of data points fitted to a straight
lme to yield the clo\urr
temperatures.
Other comments are as in Fig. 11.

808

GLENN W. BERGER and DEREK YORK

Feldspar closrrre temprrutnres

The Arrhenius diagram in Fig. I I suggests that lattice disruption begins in the K-feldspar near 950 C irt
LVXUO.
Hence the <9OOY data points are probably
the most suitable for the calculation of closure temperatures. We calculate a temperature for the Bark
Lake K-feldspar of 256 i 18C (? = 5C/Ma) or
230 + 18C (? = OSC,Ma) based on an excellent
linear correlation of five data points (Table I ).
As far as the plagioclases in Figs 6 and 7 are concerned, we show in Table 1 the results of two
approaches to calculating closure temperatures. The
first of these, the one-component approach, involves
treating the data as for the K-feldspar, biotites and
hornblendes and searching for any linear segments in
the Arrhenius diagrams that would correspond to plateaus in the age-spectrum plots. The second method,
the multi-component approach, is to sum the 4nAr*
volumes of only those fractions corresponding, or
nearly so. to the age-spectrum plateau segments. The
diffusion
coefficients
in this multi-component
approach are then calculated and plotted in the usual
way. Although we think this method gives the most
reliable results for reasons outlined here, we mention
the results of the one-component method below and
the indirect support for them.
We prefer the multi-component
approach for
reasons of internal consistency. Only this method
gave a closure temperature for plagioclase B2-149
(225 k 8C, O.SC/Ma), that agrees with the temperature for the coexisting K-feldspar (230 rtr 18C.
O.SC/Ma), as it must because they have concordant
plateaus. With the one-component method. we derive
a temperature of 151 f 10C (OS-C/Ma) for this
plagioclase. In the multi-component
approach, it
makes no difference statistically for this plagioclase
whether we use the 700-950C steps or the 60%950C
steps, even though we can argue that an acceptable
plateau begins with the 7ooC step.
Summary,

Our two ad hoc criteria for a reliable closure temperature are the presence of an acceptable age plateau
over at least five heating steps and a straight line in
an Arrhenius diagram for the same five data. Four
samples exhibited clearly defined age plateaus (biotites B12-415 and G4-22. hornblendes G2 1. 1I3 and
G 16-25) yet had poor linear correlations in Arrhenius
plots. Only one sample (plagioclase Gl4-4) having an
unacceptable plateau showed a good linear correlation in an Arrhenius plot. The remaining unreliable
closure temperatures represent samples that failed to
meet both criteria.
It is possible that the above biotite 812-415 would
have met both criteria had we changed the heating
schedule. since a detailed heating schedule did increase the number of correlating points for biotite
Gl6-25. On the other hand, such a heating routine
increased the scatter in an Arrhenius diagram for bio-

tite G4-22. Hornblendes G2l-I I3 and Glb-25 also


gave poor results in spite of relatively detailed heating
schedules.
From these comments and Table 1 it is clear that
our best results have come from the dioritic bodies.
Although we cannot explain why these bodies pro.
duced a proportionately greater number of reliable
closure temperatures than the gabbroic complex, such
a distinction may serve as a guide to the selection of
samples in future studies of this type

The results from Table 1 are plotted in Fig. 13. This


is the first geological cooling curve to be calculated
from self-contained isotopic data, and it seems to be
geologically reasonable in that the maximum temperatures are near the -600 7OOC expected for the
regional metamorphism in this area from metamorphic grid arguments (CHESWORTH. 1971). The conc;Ivity of this curve suggests that the uplift rate in the
Haliburton Highlands was rapid at first and then
slowed considerably. If our preferred curve (solid) is
valid, then the geological constraint that much of this
area was at the surface by _ 500 Ma (there are Ordovician sediments on the basement rocks a few km
south of the Haliburton intrusives- Ontario Depart.
ment of Mines. Map 1957b) requires a late-stage surge
in the uplift rate (see below).
Although we are tempted to make a detalfod Interpretation of the local uplift history from this cooling
curve, we are limited by our ignorance of the temporal dependence of the geothermal gradient. Nevertheless we can deduce average uplift rates. fhc maxi..
mum lithostatic pressure of _ 7 kbar (7 - ill5 kP:K)
(CHESWORTH,

1971)

implies

;I depth

of hurlal

of

-21 km. With a maximum temperature of 7(H)I at


1000 Ma ago (Fig. 13) this gives an average gcothermal gradient of 34 C km. greater than the
I5--2OCikm estimated for present shield arcax. If the
rocks cooled to 14OC by 5SOMa (preierKcd solid
curve). then they would have reached ti depth crf
3.Xkm, representing an average uplift r:itc oi
0,04mm,yr between 1000 Ma and 5%) %1?1:1This
would have increased to 0.08 mm:yr between 550 Ma
and 500Ma. Using the alternative dashed curve we
deduce an average uplift rate of 0.05 mmyr for the
interval 1000 550 Ma. These approximate rdles 21rc
similar to those deduced by DAI.I.MEY~ ef d., ( 1975)
for the Reading Prong, although they used different
estimates of Ar closure temprraturcs.
Some idea of the variation in uplift rate imphed t-q
the different cooling rates in the Fig. I3 caption can
be obtained if we assume ;I constant geothermal
gradient of 34C!km. In this case the uplift rate vitrics
from 0.6 mm/yr (20C,Ma) through 0.06 mmyr to
0.009 mm/yr (0.3C/Ma). In reality the geothermal
gradient is likely to have been high initially and therrafter to have decreased in some unknown fashion. For
comparison, denudation rates of 0.4-1 .Omm: yr have
been estimated for the Alps (CLARK and J&X H, 1969).

Geothermometry

8OC

809

from 4Ar/39Ar dating experiments

HBDE
BARK LAKE
GLAMORGAN

810

PLAG

DUDfUON

KFS

60C

i?
2

4oc

2oc

1000

900

800

K-Ar

700

600

500

AGE (Ma)

Fig. 13. A derived cooling curve for the Haliburton


mafic intrusions
and, by implication,
for the local
Haliburton
Highlands.
Sample numbers are shown, The error bars are f la and indicate only those
samples considered to have reliable Ar closure temperatures
as defined in the text. The solid curve is
preferred to the alternative
late-stage dashed segment (see text). All points to the right of sample 25
inclusive have been calculated
with an assumed cooling rate of O.SC/Ma while the others have used
5 C,/Ma. The errors in the apparent ages include all known sources except errors in decay and isotopic
constants. The cooling rate, represented by the tangent to our preferred curve, varies from > 20C/Ma at
975 Ma through
-2C/Ma
at 925 Ma to 20.3C/Ma
at 700 Ma. This cooling curve differs in minor
ways from an earlier version (BERGERet al.. 1979) largely because of new analyses of our OBED biotite
reference mineral. These had the effect of reducing some dates by about 2,,
In passing we mention that an earlier version of
this cooling curve (BERGER et al., 1979) was used to
deduce minimum average horizontal
crustal movements of c 5 cm/yr for the interval 98&820 Ma.

CONCLUSIONS
We have described a method of geothermometry
that yields both time and temperature from 40Ar/39Ar
dating experiments.
This method determines
K-Ar
apparent ages from 40Ar/39Ar age spectra and Ar isotopic closure temperatures
from rate parameters that
are derived routinely in the same experiments
and
which are combined with Dodsons closure temperature theory.
We have applied this technique to different minerals from three basic intrusions within the high-grade
metamorphic
terrane near Haliburton,
Ontario, and
derived a cooling curve. The results are geologically
reasonable,
supporting
our assumptions
and procedures;
however. independent
support from hydrothermal experiments
and fission-track
data is desirable.
Our ad hoc isotopic criteria for the reliability of
calculated Ar closure temperatures
have been tested

by duplicate experiments, For a hornblende and two


biotites we have varied the laboratory heating schedule and observed certain benefits. For two plagioclases, we used acid leaching before irradiation
to
demonstrate
the reproducibility
and to test the
chronological
significance of plateau segments within
plagioclase age spectra. The presence of plateau segments within the interval -60&1OOOC
for these
plagioclases appears to provide a new and sensitive
K-Ar clock, one that can extend the range of application of the 40Ar/39Ar stepwise heating method.
We have also observed that the rate parameters for
biotites define an offset pattern of two straight line
segments in an Arrhenius diagram. One segment corresponds to Ar released below + 750C and the other,
above -900C. We are thus provided with two estimates of closure temperature, and as a first approximation, we have taken the weighted mean as the best
value for this temperature.
However, it is notable in
Table 1 that the closure temperatures
calculated for
the low-temperature
pulse of Ar are less scattered
than those for the high-temperature
pulse. It may be
that one should prefer closure temperatures
derived
from the low-temperature
pulse for such biotites. Certainly one can rationalize such a preference at this

u:o

GLENN W. BERCX:.Kand DEREK YORK

stage in view of the probability that the biotite crystal


has suffered less change ir7 LUIIO by 750C than bq
IOGQC.

Finally. we note that some samples and rock types


appear to be more amenable to this method of geothermometry than others. Specifically. we found more
difficulty with minerals from the gabbroic-anothositic
rocks than with those from the dioritic rocks.

DALLMEY~R R. D.. SIITT~R J. F. and BAKI.K


IJ
f { il.?>l
Incremental Ar/3L,Ar ages of Ixot~te anti h~rri~hlcndi:
from the north-eastern Reading Prong: Thclr hc,~rtng CILI
late Proterozoic thermal and ICC~~~~IL
III<IO~~.i: -,! SI~C

:lrn. Bull. 86. 1435 1443.


DAL.KYMPLI G. B. and LANPIU KI \I

:Ir(/on Dtrlit~cq.I.reeman.
DAI.YKMPI.I G B. and LANPHEHI
age spectra of some undisturhsd
c,hin~.Cconwc~hint. .4<,1rr38, 7 I 7

I lOf+i~ IcI,;

.~,~,~I:

&I. -4. ilVY41


,\I- h!
tcl-1cytri,tl \.IIII~~~c% t ttll 2.

33

./lC!i,70HI&n?lUlt.s
We arc indebted to Professor D J
DVNIX)IJ for providing the samples: to PIx:K KC:YHIIIA for
his meticulous care in mineral
separations.
potassium
analyses and aid with Ar citractions:
to LARKI. SC.HI ITS
for thin-section descriptions of most of the samples. and to
Dr D. c. GREI:N for this thin-section
descriptions
for
sample G 13-4 and his useful discussions. We are grateful to
Professor R. E. J~R~IS for making available the gamma
spectrometry facilities. Support for this research was provided b) a Negotiated Development Grant and a Research
Operating Grant from the National Research Council 01
Canada. Revision of this manuscript has benefited from
suggestinns made by two anonymou\ revicuerc and by Dr
G T~IRUI K

REFERENCES
AI.IIARLIIX. F.. FERAIII G.. KANEOKA I. and ALLIXXI. J. c.
(197X1 Ar-Ar
dating: the importance of K-feldspars
on multimineral data of polyorogenic areas. .I. &o/. 86.
>XI 59x.
BEKC;EK G. W. (1975) Ar .Ar step heating of thermall)
overprinted biotite. hornblende and potassium feldspar
from Eldora. Colorado.
Etrrrll Plnr~c/. St,i. Zxtt. 26.
3x7 30x.
BI:KC;F.KCr. W. and YOKK D. (1981) Ar. Ar dating of thu
Thanet gabbro. Ontario: Looking through the Grenvillian metamorphic veil and implications for paleomagnetism. Ctl~. J. Gcrfll SC,;. 18, 266 773
BERCXK G. W. and YOKh D. (1979) OAr,Ar
dating of
multicomponent
magnetizations
in the Archean ShelIe!
Lake granite. northwc:stern Ontarm
CLIU. J. Eurrh SCI.
16, 1933-1941.
BERWR G. W., YORI< D. and Duhi.or
I>. J. (1979) C:alibration of Grenvillian
paleopole!, by Ar,Ar
dating.
Nuturt~277. 4& 4%
B~RNEK R. A. and HOLI)I.N G. R. JK. (lY77) Mcchamsm of
feldspar weathering: some observational
evidence. GPOlogy 5, 369%372.
BUCHA& K. L. and Dr NLW D. J. llY76) Paleomagnetism ol
the Haliburton
intrusions:
superimposed
magnetizations. metamorphism and tectonics in the late Precambrian. J. Geophys. Rcs. 81. 2951 2967.
BUCIZAX K. L., BERC;EKG. W.. MCWILLIAMS M. 0.. YORK
D. and DUNLOP D. J. (iY77) Thermal
overprinting
01
natural remanent magnetization and K/Ar ages in metamorphic rocks. J. Geonqj~. Gro&cfr.. 29, 401 410.
CHESWCIRTH W. (1971) Metamorphic
conditions in a part
of the Haliburton
Highlands
of Ontario.
L.ithos 4.
219-22).
CLARK S. P. JK. and Jh;c;t~ E. (1969) Denudation
rate in
the Alps from geochronologic and heat flow data. .4n1. .I
Sci. 267. I I43 I 160.
DALLMFIIR
R. D. (1975) Incremental
A@Ar
ages ol
biotitc and hornblende
from retrograded
basement
gneisses of the Southern Blue Ridge: their bearing on the
.age of Paleozoic
metamorphism.
:lrn. .I. St,i. 275.
444-460.
DALLMCYEW R. D. (1978) 4Ar!3Ar
incremental-release
ages of hornblende and biotite across the Georgia Inner
Piedmont: their bearing on Late Paleozoic-Early
Mesozoic tectono-thermal
histnrl. .41t1 .I. %I. 278, 124 l4Y.

Plurirt. Sci. Lrtt. 25, Xtc ~212.


MACINTYRE

R. M., YOK~CD. and Mo~n~rorx.


IV. VI (19071
Potassium argon age determinatl~~n:, III the Matl~~~-Ban
croft area in the Grenvillc
lrl~~lncc of rtic c;I~;I~I;II:
Shield. Curl. J. Earth Sci. 4, X 15 SZV
MALUSKI H. ( 1978) Behaviour
of lxotltca. Cnnl)tubl~l~,..
plagioclases and K-feldspar!.
m re\ponrc
tin /cctotmevents, with 4Ar Ar radiometric
method. c\;rmnlc Ijt

NOKWCK~ c. B. (lY74) Radiogcnlc argon diffllbl<Bn 111the


biotitc micas. linpublished
M.Sc ihtxi\
Dent 411(;c~l
Sci.. Brown Clniv.
ONIONS R. K., SMIT~I D. G W.. BA~ZI)S(~~XUU:Ii ;uI~:
MORTC)N R. D. (1969) Influsncc 111chemical c~mlx~>ltlolt
on argon retentivity in metamorphic
calc~c. ampbib&\
from south Norway. Eurth P/~o~i S?i. I.t~i 5. .;70 X5
OZIMA M. and TAUGAMI Y. (1980) Activation cncrgy lot
thermal release of Ar from some DSDP
$uhmarine
rocks. Grochirt~. Cvs~~~o~hin~..4~rrr 44. 1-lI 141.
PALMER H. C.. HAYATSC. A.. A.. WAHOSA C. C. and PI I ~1
S. (1979) A paleomagnetic and K-i\r study of the t.:mfruville gabbro. Ontario. CUU. J. k<r~lr ,S,i. 16. 4% -47 1,
PEPIN R. O., RI~~NOLIX J. H. and Tr KhtR G ( 10041
Sh~wh
emplaced Argon in a stony metcllrltc:
1 k,lmp.trl\o!,

Geothermometry

from Ar,jgAr

with natural argon in its dlfFusion. J. Gropll.~s. Res. 69,


I~O&l41 I.
STI IC;LK R. H. and J~~GEK E. (1977) Subcommission of
gcochronologq:
conlention
on the use of decay cons,ants in @eo and cosmochemistry.
Eurrll Plmc~.
Sci.
Lvr
36, 359 362.
STI 6.4s V. (1977) Plugioclase release patterns: a high resolution Ar Ar stud!. Unpublished
Ph.D. thesis. Dept.
of Cicol.. Dalhousle Univ.
Tt~tf.1
N. and
Mc DOLIGALL
I. (197X) Anomalous
Ar Ar release spectra for blotltes from the Berridale
Batholith.
Nem South Wales. Australia.
Short Pap. 4th
Int. Conf. Geochron. Cosmochron. Isotope Geol. L.S.
r/ctj/. Slrrr. O~U-Fiic Rep. 7X 701. pp. 427 430.
Tt KUI K D. L. and FORI~S R. B. (1976) K-Ar studies in two
deep basement
drill holeT: a new geologic estimate of
argon blocking temperature
for biotite. Eos. 57, 353.
T~,KYI K C;. (1969) Thermal histories of meteorites
by the
ZAr-JAr method.
In !Il<~rwriric~
Rtwtrrch
led. P. M.
Millman). pp. 407 417. Reidel.
TORNI K G. ( 1979) A Monte Carlo fragmentation
model for
the production
of meteorites:
Implications
for gas retention ages. P~~~c~. lctrh Lufw
Pltrrrrr.
SC;. CCU$.
pp.
19l? iY3l.
TI.KUI R G.. E\RIC;HT M. C. and CAI~OGA~. P. H. (1978) The
earl! histor! of chondritc
parent bodies inferred from
Ar-Ar
ages. Pvw
Yrh Lmu
P/met.
SC;. CotIf:.
pp.
9s9 105.
WAV.I.K G. .4.. RLIZICRG. M. and JAGER E. (1977) Cooling
age> derived bk apatite fission-track,
mica Rb-Sr and
K-Ar dating: the uplift and cooling history of the (entral Alps. .$Ic,~orit, 11; Pud,,~[r XXX.
YORK D. (1969) Least-squares
fitting of a straight line with
correlated
crrorb. huh
PIIIWI.
SC,;. Lcrr. 5, 32@324.
YOKK I).. ~AI\I:SI. k. and Bt.KctK G. W. (1970) Determlnation of geological
time ulth a nuclear reactor and a
mab\ hpcctrometrr.
In P~IIc,. N.4 TO .Adwr~crd
Study Imr.
OH ~lcriwrion
I~J. KJeller.

Stclnnesl.

~~I/J\,.x

itI C;cw/wrvi,~fr~~

Noraa!
pp. II9 422.

leds

A.

D.

md

Co.wocher~ri.s-

Brunfeld

and

E.

APPENDIX
Thiwvtaiorl

c0))lf)le~lr.s

In the following, the B. D or G in the sample number


indicates respectiveI> the Bark Lake. Dudmon or Glamorgan intrusions. while the number preceding the dash gives
the sltc number of Buchan and Dunlop (1976) and the
number after is their sample number.
B2-IJY:

h~~~~~tlrot~~o~~~l~i~ qrrrtruhr.

rLww

-0.2 x 0.3
to
from
Biotitc- grain
sire
ranges
.kO.X )/ 3 mm: brown: altered partly to chlorite through
xreen hiotite: some magnetite.
hematite and quart7 inclus~ons: some evidence-of grain fracturing.
Hornblende
maximum <i/-c 1 I x 0.5 mm; a hit of green
hornblende
exists: borne alteration
to chlorite plus magnetlte: magnetite inclusions.
largest grain 2 3.5 x 5 mm, others
K-feldspar
_orthoclase:
1 I X to 2 3 this ~17~:some small degree of sericitation
011 some graln\:
the odd grain has a bit of quartz inelusion.
maximum
size 23.5 x 5
Plag~~>clase calcic andesine;
mm: ;I ;maII hlr of ycricitc production:
no inclusions.
undeformed

dating

812-4

XII

experiments

15: hypidiomorphic

qw~u/ur

rtta~t~c~

Biotite--up
to 3 x 4mm in size: mostl!
brown, some
altered to green. a little of this further altered to chlorite
with magnetite and perhaps sphene. Inclusion5 of quartz.
magnetite, calcite. plagioclase or reolites in \omr grain\:
grains are well formed. a small hit of bcndmg
\\lth
straining ohserbed in 2V.
up to I r I mm m SILC: all green. lncluslonb
Hornblende
of quartz and magnetite:
no alteration
be!und grecnlng;
undeformed.
Plagioclase-andesine:
up to X x 3 mm in \Ire: II,, 111.
cluslons: undeformed:
some cericitc production
ES-/3Y.-

/1~,pitlr~,,tlorphi~

qrurfrtlur

rt\rw

up to ~2 x 2 mm: hrov.n IO green v,ith CLIHornblende


dence of alteration
to chlorite plu\ magnetite
in a fee\l
grams: some lncluslons
of quart7. \phenc.
maenctitc.
some zeolite: undeformed.
Plagioclase-andesine:
up to 12 )I Imm:
rckltlvcl!
free
of inclusions: undefnrmed;
a littlc scrlciti7ation.
Dl-IYiJ

Poikilltic
texture: hornblende
biotite is light brown. unaltered,
well twinned.

is green to blue green:


undeformed:
plagioclasc 14

Gl4-4
A nepheline-hornblende
plagioclasc
gneiss with subidioblastic amphibole
prisms in a granoblastic
matrix.
Amphibole-hastingsitic;
average length 17 mm: osscntially no inclusions. contains no relic proxcnc: comprlsc\
1.35 0 of the rock.
slight aIteratIon to
Nepheline --> 35,, of rock; ubiquitous
prey green gieseckite; inclusions of plagloclasr.
apatite.
amphibole
and rare microline are a common fcaturc.
Plagioclase= 20,, of the rock: well twinned: occurs a:,
fresh subidioblastic
or rounded interstitial
grains up to
1 mm in diameter;
calcic oligiocla,e:
a fe\\ patcheh of
nepheline are the only inclusions.

GI6--75: /r!pidionlorphic,

yrur7ulur

rt~ww

Biotite cup to 2 x I mm: predomlnantlb


brown Lvith some
alteration to green and a small bit altered a5 far as chlorlte: some inclusions
of magne,ite.
reolite and pcrhapb
quartz; some slight bendmg in a few crystals.
Hornblende
-up to - I x I mm; occurs as uralite Ialteration of clinopyroxene)
at houndartes
and along cleavage planes of primary grains: all green: inclusions
01
magnetite and perhaps quart7 and rcolitc: undeformed.
Plagioclase-about
Ab,,
in composltion:
maulmum
z 3.5 x 2 mm: relatively free of inclusion\:
undeformed:
a little sericitization.
_
G4-3-1: /~~pidiotwrphic

ftwirt

Biotite
up to I x 0.3 mm; mostly green. some brown.
magnetite inclusions: a little chlorite: undclhrmcd.
G-71-1 13; pa,lidrortlorphic,

qrm~~lur

rp\rm

Hornblende
-up to 2 x 0.75 mm: all green: \;omc magnetite inclusions; no alteration;
undeformed
Plagioclasesodic labradorite:
up to 3 r 2 mm: some
quartz and muscocite
inclusions.
u Ilttle \erlcitiration:
undeformed.