Full text of DOI:10.1093/petrology/42.7.1301

JOURNAL OF PETROLOGY
VOLUME 42
NUMBER 7
PAGES 1301–1320
2001
High fO During Sillimanite Zone
2
Metamorphism of Part of the Barrovian
Type Locality, Glen Clova, Scotland
JAY J. AGUE∗, ETHAN F. BAXTER† AND JAMES O. ECKERT JR
DEPARTMENT OF GEOLOGY AND GEOPHYSICS, YALE UNIVERSITY, PO BOX 208109, NEW HAVEN, CT 06520-8109, USA
RECEIVED NOVEMBER 8, 1999; REVISED TYPESCRIPT ACCEPTED DECEMBER 5, 2000
INTRODUCTION
The redox state of sillimanite zone (650–700°C, 5–6 kbar)
metasediments of the Barrovian type area, Scotland, was investigated
using estimates of metamorphic oxygen fugacity (fO ), sulfur fugacity
The chemical and mineralogical evolution of the crust
during orogenesis is critically dependent upon the redox
state of metamorphic fluids ( James, 1955; Chinner, 1960;
Buddington & Lindsley, 1964; Miyashiro, 1964; French,
1966; Eugster & Skippen, 1967; Thompson, 1970; Harte,
1975; Ohmoto & Kerrick, 1977; Rumble, 1978; Lamb
& Valley, 1985; Frost, 1991a, 1991b; Brenan et al., 1995;
Harlov et al., 1997; Ague, 1998). The classical Barrovian
type locality in Glen Clova, Angus, Scotland, is a superb
example of redox controls on metamorphism (Chinner,
1960). Based on field relations, petrography, and the
chemical systematics of minerals and bulk rocks, Chinner
(1960) showed that large differences in bulk-rock ox-
idation ratio (OR) existed at local and regional scales
between metasediments during amphibolite facies meta-
morphism in Glen Clova [OR = molecular 2Fe2O3 ×
100/(2Fe2O3 + FeO)]. These differences were attributed
to chemical contrasts inherited, at least in part, from the
sedimentary protoliths, and were taken as evidence that
(fS ), and fluid chemistry based on new determin2ations of mineral
2
and rock compositions from 33 samples. A total of 94% of the
samples lack graphite, contain both ilmenite–hematite solid solutions
(RHOMOX) and magnetite, and had metamorphic fO about 2
2
log10 units above the quartz–fayalite–magnetite (QFM) buffer. The
regional variation in metamorphic f for these rocks was minimal,
O2
about
0·3 log10 units, reflecting either a protolith that was
homogeneous with respect to redox state, or an initially variable
protolith whose redox state was homogenized by metamorphic
fluid–rock interaction. RHOMOX inclusions in garnet por-
phyroblasts that become richer in ilmenite from the interior to the
edge of the host porphyroblast suggest that at least some syn-
metamorphic reduction of rock occurred. Significant variations in
bulk-rock oxidation ratio (OR) that are probably inherited from
sedimentary protoliths are found from one layer to the next; OR
ranges mostly between >20 and >50 [OR = molecular 2Fe2O3
× 100/(2Fe2O3 + FeO)]. These OR variations are uncorrelated
with fO and do not indicate that large, order-of-magnitude gradients
in fO a2nd redox state existed or were preserved between layers during
meta2morphism. The other 6% of the samples contain ilmenite, lack
significant differences in fO and fluid composition existed
2
between layers during metamorphism (Chinner, 1960).
Rock oxidation ratio variations in Glen Clova have
been unequivocally established, but important questions
remain regarding the metamorphic fluids. For example,
what were the quantitative variations in the fugacities of
O2 and sulfur species such as S2 in the region? Did large
magnetite, and had low fO 0–1 order of magnitude below QFM
2
in the stability field of graphite. They are characterized by com-
binations of the following: large fluid HF/H2O; metasomatic,
tourmaline-bearing veins; absence or rarity of primary organic
matter; and crosscutting late metamorphic shear zones rich in
carbonaceous material. Such observations suggest that locally low
variations in fO and fluid composition exist between
2
intercalated metasedimentary layers? Could meta-
morphic fluid–rock interaction between layers have over-
come characteristic buffer capacities inherited from
variable protoliths and act to homogenize redox states
fO conditions may have been related to the influx of reducing fluids
2
from elsewhere in the area.
KEY WORDS: metamorphism; redox; Barrovian; Scotland; oxygen fugacity of fluids over the field area? In an effort to address these
∗Corresponding author. Telephone: +1-203-432-3171. Fax: +1-203-
432-3134. E-mail: jay.ague@yale.edu
†Present address: Division of Earth and Planetary Sciences,
California Institute of Technology, MC 170-25, 1200 E. California
Blvd., Pasadena, CA 91125, USA.
 Oxford University Press 2001

JOURNAL OF PETROLOGY
VOLUME 42
NUMBER 7
JULY 2001
questions, we present a petrologic study of the Glen some pre-sillimanite zone kyanite (Chinner, 1960, 1961;
Clova rocks focused on quantification and interpretation McLellan, 1985).
of regional variations in the values of intensive variables
The oxide mineral assemblages provide a firm basis
during amphibolite facies Barrovian metamorphism. Our for subdivision of rock types. Chinner (1960) described
assemblages of ilmenite + magnetite, ilmenite + mag-
netite + hematite, and hematite + magnetite. In ad-
dition, we recognize a fourth variety containing ilmenite
alone.
results suggest that regional mapping of metamorphic
fO and fluid compositions is useful for assessing possible
2
redox state histories for the Barrovian type area.
Throughout the text, we use the term ‘redox state’ in a
general way to refer to the fO and composition of
2
metamorphic pore fluids.
PETROGRAPHIC RELATIONS
We present new compositional data for silicate and oxide
minerals in 33 rocks from the Glen Clova area. Because
Chinner (1960) provided data for garnet, muscovite,
biotite, and rhombohedral oxide for his sample 11, we
also consider this sample. Symbols are defined in Table
1, mineral assemblages are listed in Table 2, mineral
compositions are given in Tables 3–6, and analytical and
data reduction procedures are described in the Appendix.
Excellent petrographic descriptions of silicate as-
semblages have been given by many workers (e.g. Harry,
1958; Chinner, 1960, 1961; Harte & Johnson, 1969;
McLellan, 1985, 1989). Consequently, we focus on the
oxides and sulfides.
GEOLOGIC RELATIONS
The metapelitic and metapsammitic rocks of Glen Clova
form part of the Dalradian Supergroup and were de-
formed and metamorphosed during the Grampian oro-
geny (Harte et al., 1984; Fettes et al., 1986; Rogers et al.,
1989; Dempster & Bluck, 1995); peak metamorphism
occurred at >470 Ma (Oliver et al., 2000; Baxter et al.,
2000). The timing of prograde porphyroblast growth
relative to the four main deformational events (D1–D4) in
the area can be summarized as follows (McLellan, 1989):
(1) garnet: mostly syn-D2, some syn- to post-D3; (2) stau-
rolite: syn- to post-D2; (3) kyanite: pre- to syn-D3; (4)
sillimanite: mostly post-D3. Some rocks also contain stau-
rolite and kyanite that formed after sillimanite zone
metamorphism and, presumably, after D3 deformation
(see Chinner, 1961; McLellan, 1985). The inferred bed-
ding and the primary planar deformational fabric are
subhorizontal (dips are <>30°) and, thus, the rocks are
part of the ‘Flat Belt’ of the Scottish Highlands (Harte et
al., 1984).
The Glen Clova area contains much loose float; we
made every effort to sample only from fresh, in situ
outcrops. We investigated rocks in and around Chinner’s
(1960) zone of ‘Hematite-bearing gneisses and associated
semipelitic gneisses’ (Fig. 1), which our mapping indicates
is at least 200–250 m thick. Chinner (1960) reported that
at the margins of this zone the hematite-bearing rocks
are intimately interbedded at the inch scale with hematite-
free, graphite-bearing rocks, but we note that most of
the graphitic rocks described in his paper are located
2–5·5 km away from the hematite-bearing zone. We were
unable to confirm or refute the presence of such inch-
scale interbedding during the course of our field and
laboratory work.
Prograde phases
The rhombohedral oxide (RHOMOX) minerals range
primarily between ilmenite (Ilm) and hematite (Hem) in
composition. RHOMOX are found throughout the rock
matrix and as inclusions in garnet, aluminosilicate min-
eral, staurolite, and tourmaline porphyroblasts. Most
crystals are subhedral to euhedral and 10–900 ꢀm long.
Unaltered RHOMOX in magnetite-bearing rocks con-
tain exsolution lamellae; in many samples, these lamellae
underwent additional fine-scale exsolution themselves
(Fig. 2). The multiple sets of exsolution features almost
certainly developed during progressive cooling from
‘peak’ T, as expected from phase relations in the Fe2O3–
FeTiO3 system (e.g. Burton, 1991), and suggest, although
do not prove, that little chemical alteration of the grains
occurred during cooling (Harlov et al., 1997). Rocks
without magnetite contain Ilm (XIlm >0·95) that lacks
exsolution features.
Cubic solid solutions vary primarily between the mag-
netite (Mag) and ulvo¨spinel (Usp) endmembers, but are
dominated by Mag (XMag > >0·95). Mag crystals are
found in the rock matrix and, more rarely, as inclusions
within porphyroblasts. Most grains are subhedral to
euhedral and 10 and 1100 ꢀm in diameter. Rare martite
alteration of probable supergene origin is present in a
few samples.
McLellan (1985) inferred the main metamorphic re-
actions for the area. ‘Peak’ sillimanite zone conditions
were >650–700°C at >6 kbar (McLellan, 1985), were
probably accompanied by some degree of anatexis
(McLellan, 1989), and may have required heat input by
fluid advection (McLellan, 1989). No change in grade is
Rutile (Rt) is uncommon. Crystals are generally sub-
evident across the field area, and most rocks retain hedral, >100 ꢀm long, and found in rock matrices
1302

AGUE et al.
HIGH fO DURING SILLIMANITE ZONE METAMORPHISM
2
Fig. 1. Location map. Unpatterned area, amphibolite facies metasedimentary rocks; cross-hatch pattern, post-peak metamorphic intrusions
(‘Newer’ igneous rocks); stipple, para-amphibolite (‘Green Beds’). Samples used for northwestern, central, and southeastern profiles enclosed by
dashed lines. Post-metamorphic, shallow-level, porphyritic dikes and sills omitted for clarity.
and as inclusions within porphyroblasts. Rt crystals are
typically rimmed or partially replaced by RHOMOX. It
is unclear if these rim or replacement textures developed
during prograde metamorphism or if they are retrograde.
Additional Rt parageneses that are almost certainly retro-
grade are described below.
Table 1: List of symbols
Symbol Definition
Symbol
Definition
Some samples contain small amounts (<>1 vol. %)
of sulfides. Pyrite (Py), pyrrhotite (Po), and chalcopyrite
(Ccp) crystals are found mainly in the rock matrix and
much less commonly as inclusions within porphyroblasts.
Crystals range from anhedral to euhedral and most are
between 5 and 1000 ꢀm in diameter. Ccp, Po, and Ni-
sulfide are also found as tiny rounded ‘droplets’ within
some Py. Evidence for some low-temperature supergene
alteration of sulfides to Hem and limonite is present in
most samples. Because supergene reactions may have
modified the prograde sulfide assemblages (e.g. alteration
of Po to Py), interpretation of metamorphic sulfide as-
semblages is somewhat hazardous.
Alm
Adr
Ann
Bt
Almandine
Andradite
Annite
Pyr
Pyrophanite
Quartz
Qtz
RHOMOX
Rt
Rhombohedral oxide
Rutile
Biotite
Ccp
Gk
Chalcopyrite
Geikielite
Grossular
Garnet
Sil
Sillimanite
Spessartine
Staurolite
Sps
St
Grs
Grt
Hem
Ilm
Ky
Usp
Zrn
Ulvospinel
¨
Hematite
Ilmenite
Kyanite
Zircon
a
f
Activity
Mag
Ms
Phl
Po
Magnetite
Muscovite
Phlogopite
Pyrrhotite
Pyrope
Fugacity
P
T
X
Pressure
Temperature
Mole fraction
Retrogression of Fe–Ti oxide phases
Small, Ilm-rich (XIlm > >0·85) RHOMOX
Rt grains
Prp
Py
are commonly found (1) along cleavages and margins of
biotite and, more rarely, muscovite crystals, and (2) in
contact with and as partial rims around Mag grains. This
Pyrite
OR
Rock oxidation ratio
All mineral symbols except Gk, Pyr, and RHOMOX from Kretz
(1983).
Ilm
Rt probably formed from retrograde Ti release
1303

JOURNAL OF PETROLOGY
VOLUME 42
NUMBER 7
JULY 2001
of Rt, Ti-rich Ilm, Ti-poor Hem, and, rarely, titanite.
These pseudomorphs are inferred to have originally been
prograde crystals because the pseudomorphic re-
placement has generally preserved relict exsolution tex-
tures characteristic of cooling from high T. Fe2O3–FeTiO3
phase relations predict that at T below >500°C, Ti-
poor Hem and Ti-rich Ilm should coexist (e.g. Burton,
1991). Thus, the pseudomorphed RHOMOX probably
comprise retrograde minerals that may have equilibrated
Table 2: Prograde mineral assemblages
Sample Grt
Ky
Sil
RHOMOX Mag Py
Po
Ccp
X
36A
37C-2
38A
39B
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
40A
41C
to some degree with nearby retrograde Ilm
Rt as-
X
X
X
X
X
sociated with micas and Mag. RHOMOX grains that
are far from micas and Mag or that are preserved as
inclusions within garnet, aluminosilicate, or tourmaline
porphyroblasts are typically not pseudomorphed.
42A
45A
46B
X
X
X
X
X
?
X
47A-1
48E
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
49A
50B
X
X
X
PRESSURE AND TEMPERATURE
P–T estimates for selected samples were computed with
the TWEEQU program of Berman (1991) using the
program’s default thermodynamic data and activity mod-
els. Mole fractions were computed using version 0·95 of
the CMP program available at the TWEEQU Web site
(www.gis.nrcan.gc.ca/twq.html). The mole fraction of
hydroxyl in micas was adjusted to include F and Cl
substitution for OH (e.g. Ague & Brimhall, 1988; McMul-
lin et al., 1991). We calculated simultaneous solutions of:
(I) the garnet–biotite Fe–Mg exchange (GARB; Alm +
Phl = Py + Ann) and the SGAM reactions of McMullin
et al. (1991; Alm + Ms = Ann + 2 Sil + Qtz, Prp +
Ms = Phl + 2 Sil + Qtz); or (II) GARB and GRAIL
(Alm + 3 Rt = Sil + 2 Qtz + 3 Ilm). Method (II) was
applied to sample 62A, which contains rutile that appears
not to be retrogressed (see above). Reactions involving
grossular in garnet were not used because grossular
contents are generally small and subject to significant
uncertainties stemming from estimated andradite con-
tents.
Results cluster in the sillimanite field (Fig. 3), averaging
657°C and 5·7 kbar; we used 660°C and 5·7 kbar to
calculate fluid composition herein. T was also estimated
using GARB at 5·7 kbar for five samples that lack al-
uminosilicate minerals (36A, 45A, 49A, 50B, 52C). The
average T of 668°C agrees with the simultaneous P–T
estimates of Fig. 3. Because larger garnet grains preserve
bell-shaped Mn growth zoning profiles, the temperature–
time evolution of these rocks was insufficient to cause
complete diffusional equilibration of cations in garnet.
Our P–T results are fully consistent with those of McLel-
lan (1985).
51B
X
X
X
X
52C
54A
55A
56A
57B
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
58A
59A
60A
61B
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
62A
63A-2
64B-2
65A
66A
67A
68A
69B
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
?
X
X
X
X
70A
71A
Chin11∗
X
X
X
X
n.r.
n.r.
n.r.
All samples contain quartz, plagioclase, biotite, and mus-
covite. Sulfide assemblages not listed if strong limonite al-
teration present. n.r., not reported; ?, possible trace.
∗Sample 11 of Chinner (1960).
from micas and Mag (e.g. Ague & Brimhall, 1988). The
small Usp contents of the Mag grains probably reflect,
at least in part, such retrograde Ti loss. It is unclear if
retrograde Ilm nucleated and grew during cooling, or if
pre-existing prograde metamorphic grains were com-
positionally modified during cooling.
REACTIONS
Nearby matrix RHOMOX are largely pseudomorphed A fundamental conclusion of Chinner (1960) was that,
by fine-grained assemblages containing variable amounts for any given sample, silicates and oxides reacted together
1304

AGUE et al.
HIGH fO DURING SILLIMANITE ZONE METAMORPHISM
2
Table 3: Rhombohedral oxide compositions
Sample Type∗ Ti
Al
Cr
Fe3+
V
Fe2+
Mn
Mg
Zn
Total
XIlm
XHem
XGk
XPyr
36A
37C-2
38A
39B
40A
40A
41C
42A
45A
46B
47A-1
47A-1
48E
M
0·735
1·730
1·292
1·791
1·298
1·753
1·016
0·596
0·883
1·940
0·915
1·941
0·643
0·800
0·628
1·015
1·592
0·802
0·803
0·866
0·835
1·378
0·970
1·734
0·869
0·738
0·822
1·682
1·782
1·957
1·007
0·911
0·932
1·395
1·711
1·701
1·738
1·661
0·761
1·097
1·325
1·829
0·622
0·922
1·726
1·820
0·516
0·005
0·001
0·002
0·002
0·006
0·002
0·002
0·010
0·010
0·007
0·006
0·002
0·009
0·005
0·003
0·002
0·001
0·005
0·007
0·009
0·004
0·001
0·004
tr.
0·007
0·004
0·006
0·005
n.d.
2·505
0·522
1·395
0·405
1·397
0·493
1·958
2·786
2·211
0·112
2·164
0·115
2·686
2·385
2·734
1·956
0·805
2·377
2·375
2·247
2·312
1·235
2·034
0·521
2·252
2·507
2·321
0·633
0·431
0·083
1·977
2·167
2·115
1·201
0·572
0·589
0·518
0·669
2·465
1·787
1·334
0·331
2·740
2·136
0·544
0·358
2·968
0·012
0·009
0·014
0·005
n.d.
0·715
1·692
1·278
1·737
1·257
1·736
0·996
0·573
0·864
1·855
0·878
1·867
0·625
0·686
0·585
0·997
1·555
0·791
0·794
0·852
0·826
1·359
0·960
1·706
0·846
0·720
0·795
1·629
1·767
1·904
0·956
0·886
0·922
1·354
1·644
1·670
1·703
1·614
0·716
1·089
1·295
1·796
0·602
0·906
1·702
1·787
0·516
0·018
0·030
0·007
0·043
0·020
0·010
0·017
0·019
0·012
0·080
0·032
0·073
0·009
0·112
0·041
0·017
0·035
0·009
0·008
0·013
0·008
0·016
0·008
0·023
0·019
0·017
0·026
0·018
0·008
0·050
0·048
0·023
0·009
0·039
0·033
0·025
0·017
0·016
0·042
0·006
0·026
0·025
0·012
0·015
0·011
0·018
n.d.
0·001
0·009
0·005
0·010
0·022
0·006
0·001
0·002
0·004
0·004
0·005
0·002
0·007
0·002
tr.
0·002
0·001
0·001
0·002
n.d.
100·00 0·360
99·26 0·849
99·97 0·643
99·11 0·872
99·30 0·630
98·99 0·868
99·43 0·500
101·48 0·289
98·84 0·436
99·53 0·930
99·03 0·439
97·38 0·934
99·30 0·316
100·58 0·345
100·33 0·294
99·99 0·501
100·00 0·780
100·97 0·398
101·18 0·399
100·84 0·428
101·12 0·415
100·81 0·682
100·00 0·484
99·92 0·857
100·28 0·425
99·73 0·362
100·29 0·401
100·68 0·817
100·10 0·885
100·35 0·953
100·05 0·479
99·75 0·445
99·85 0·464
100·40 0·679
99·87 0·824
100·12 0·837
99·69 0·854
98·80 0·809
100·65 0·360
100·12 0·548
99·73 0·651
100·07 0·902
101·09 0·306
100·12 0·456
100·24 0·852
100·34 0·894
98·59 0·258
0·630
0·131
0·351
0·102
0·350
0·123
0·491
0·700
0·556
0·028
0·542
0·029
0·676
0·598
0·685
0·491
0·202
0·597
0·597
0·565
0·581
0·309
0·512
0·131
0·564
0·629
0·585
0·158
0·108
0·021
0·495
0·543
0·531
0·301
0·143
0·148
0·130
0·168
0·618
0·449
0·335
0·083
0·688
0·537
0·136
0·090
0·742
tr.
0·009
0·015
0·004
0·022
0·010
0·005
0·009
0·010
0·006
0·040
0·016
0·036
0·004
0·056
0·020
0·008
0·018
0·005
0·004
0·006
0·004
0·008
0·004
0·011
0·009
0·008
0·013
0·009
0·004
0·025
0·024
0·011
0·005
0·019
0·017
0·013
0·009
0·008
0·021
0·003
0·013
0·013
0·006
0·008
0·006
0·009
n.d.
GS
M
0·005
0·003
0·005
0·011
0·003
tr.
M
IG
RM
IG
n.d.
n.d.
n.d.
0·002
0·004
0·003
0·001
n.d.
0·006
0·007
0·010
tr.
0·002
0·002
0·003
0·001
n.d.
IG
0·001
0·002
0·002
0·003
0·001
0·004
0·001
tr.
M
IG
IG
n.d.
RM
IG
n.d.
n.d.
n.d.
0·008
tr.
0·012
0·008
0·005
0·009
0·008
0·010
0·009
0·008
0·012
0·007
0·012
0·006
0·005
0·010
0·013
0·001
0·004
0·002
0·004
0·005
0·008
0·006
0·002
0·005
0·003
0·004
0·006
0·010
0·010
0·006
0·012
0·010
0·002
0·001
n.d.
0·002
0·001
0·002
0·002
0·001
0·002
0·001
0·001
0·001
0·001
0·001
0·003
0·002
tr.
49A
50B
51B
52C
54A
55A
56A
57B
57B
58A
58A
59A
60A
61B
61B
61B
62A
63A-2
64B-2
65A
66A
66A
66A
66A
66A
67A
68A
68A
68A
69B
70A
71A
71A
Chin11†
M
IG
0·002
0·002
0·004
0·004
0·003
0·004
0·002
0·002
0·010
0·006
0·002
0·005
0·019
0·002
0·002
0·001
0·001
0·001
0·006
0·002
0·002
0·004
0·002
0·002
0·002
0·004
0·006
0·005
0·004
0·004
0·002
0·002
n.d.
IG
tr.
tr.
M
0·001
0·001
tr.
tr.
IGT
IGK
IK
tr.
tr.
0·001
0·001
0·002
0·001
0·002
0·003
0·001
tr.
tr.
M
tr.
M
0·001
tr.
M
RM
IG
0·001
0·002
tr.
0·002
0·002
0·002
0·001
0·001
tr.
IG
IGi
IG
0·001
0·003
0·001
tr.
tr.
0·032
0·006
0·002
0·003
0·002
0·001
0·002
0·032
0·005
0·016
0·030
0·002
tr.
0·016
0·003
0·001
0·002
0·001
tr.
ISM
IG
IG
0·000
0·004
0·006
0·001
0·001
0·001
0·002
0·003
0·004
0·005
0·001
tr.
tr.
IG
tr.
IG
tr.
IGi
IG
0·001
0·002
0·001
0·002
0·001
0·001
0·002
0·002
0·003
0·008
0·001
0·001
tr.
0·001
0·016
0·002
0·008
0·015
0·001
tr.
ISM
IK
IK
M
M
IG
0·002
0·005
tr.
0·001
0·003
tr.
PKSt
IK
tr.
IK
0·006
0·001
tr.
tr.
tr.
IG
0·012
0·014
n.d.
0·006
0·007
n.d.
ISM
—
n.d.
n.d.
Six oxygen formula basis. n.d., not determined; tr., trace (<0·001). Fe2+, Fe3+ calculated based on stoichiometry. Total is total
weight percent incorporating calculated FeO and Fe2O3.
∗GS, at garnet–sillimanite contact; IG, inclusions in garnet rim; IGi; inclusions in garnet interior; IGK, inclusions in garnet
and kyanite; IGT, inclusions in garnet and tourmaline; IK, inclusions in kyanite; ISM, inclusions in sillimanite–muscovite
aggregates; M, matrix grain; PKSt; at contact of plagioclase and post-sillimanite kyanite and staurolite; RM, retrograde grain
in matrix.
†From Chinner (1960).
1305

JOURNAL OF PETROLOGY
VOLUME 42
NUMBER 7
JULY 2001
Table 4: Magnetite (Mag) and rutile (Rt) compositions
Sample
Phase
Type∗
Ti
Al
Cr
Fe3+
V
Fe2+
—
Mn
Mg
Zn
Total
XMag
62A
37C-2
39B
40A
40A
41C
52C
54A
56A
61B
66A
71A
71A
Rt
IG
M
M
IG
M
M
M
M
M
M
M
IG
M
1·954
0·030
0·003
0·045
0·041
0·004
0·005
0·002
0·002
tr.
tr.
0·002
0·002
0·004
n.d.
0·053
1·919
1·963
1·898
1·914
1·974
1·971
1·983
1·981
1·987
1·989
1·970
1·974
0·002
0·003
0·005
n.d.
0·002
tr.
tr.
99·69
99·71
—
Mag
Mag
Mag
Mag
Mag
Mag
Mag
Mag
Mag
Mag
Mag
Mag
0·017
0·023
0·012
0·005
0·008
0·011
0·008
0·009
0·006
0·005
0·018
0·017
1·026
1·002
1·042
1·040
0·994
1·004
1·002
1·002
1·000
0·999
1·001
1·003
tr.
0·003
0·001
0·001
0·001
tr.
0·001
tr.
0·959
0·981
0·949
0·957
0·987
0·986
0·992
0·991
0·994
0·994
0·985
0·987
0·001
0·002
tr.
100·47
100·29
99·66
n.d.
n.d.
0·001
0·001
tr.
n.d.
n.d.
0·002
0·001
0·001
0·001
0·001
0·001
0·003
0·001
0·007
0·006
0·004
0·004
0·005
0·004
0·006
0·003
0·010
tr.
99·60
tr.
101·20
100·39
100·09
100·36
100·30
100·19
100·91
tr.
tr.
tr.
tr.
tr.
tr.
tr.
0·001
0·001
tr.
0·001
0·001
0·003
tr.
tr.
tr.
tr.
tr.
tr.
tr.
Four oxygen formula basis. n.d., not determined; tr., trace (<0·001). Magnetite Fe2+, Fe3+ calculated based on stoichiometry.
All Fe as Fe3+ for rutile. Total is total weight percent. Total incorporates calculated FeO and Fe2O3 in magnetite.
∗IG, inclusions in garnet; M, matrix grains.
as part of a thermodynamic system. The primary reaction
2O2 + 3Py = Mag + 3S2
for Py-bearing assemblages, and
2O2 + 3Po = Mag + 1·5S2
(7)
(8)
that we used to estimate fO involves both silicates and
2
magnetite
4Qtz + 2Mag + 2Sil = O2 + 2Alm.
(1)
for Po-bearing ones. For samples that lack Mag and Py,
we used the reaction 3O2 + 4Po = 2Hem + 2S2.
The main advantages of this reaction are that (1) nonideal
activity corrections for Alm and Mag are well constrained
for the compositional ranges of interest, and (2) Qtz and
aluminosilicates are essentially pure phases. For rocks
lacking aluminosilicate minerals or garnet we used
The fHF /fH O estimates used measured biotite com-
2
positions (Table 5) following Zhu & Sverjensky (1992).
2+
2+
2+
2+
The log10(aMg /aFe ) and log10(aMn /aFe ) for the fluid
were estimated with
Fe2+ + 1/3Prp = Mg2+ + 1/3Alm
(9)
3Qtz + 2Mag + Ms = O2 + Ann + Alm (2)
2Qtz + 2Mag + 2Ms = O2 + 2Ann + 2Sil. (3)
and
Fe2+ + 1/3Sps = Mn2+ + 1/3Alm.
(10)
Calculation of log10K for reaction (9) involves a slight
extrapolation of fluid species thermodynamic data beyond
the recommended maximum P of 5 kbar (Shock et al.,
(4) 1997), but we emphasize that we are concerned with
regional gradients in fluid composition that are in-
dependent of log K, not absolute values. Because thermo-
2+
We also considered the hematite analogs of reactions
(1)–(3):
8Qtz + 6Hem + 4Sil = 3O2 + 4Alm
6Qtz + 6Hem + 2Ms = 3O2 + 2Ann + 2Alm (5)
4Qtz + 6Hem + 4Ms = 3O2 + 4Ann + 4Sil. (6)
dynamic data for Sps are lacking, we plot log10(aMn
/
2+
aFe ) −log10K for reaction (10).
However, these reactions are complicated by uncertain
activity–composition relations for RHOMOX (see
below). This uncertainty is particularly problematic for
the Mag–Hem reaction (6Hem = 4Mag + O2) because
ACTIVITY–COMPOSITION
RELATIONS AND THERMODYNAMIC
DATA
fO depends on the sixth power of the Hem activity. Thus,
2
we do not consider this reaction further.
We used the activity model of Ghiorso (1990) for RHO-
MOX. However, Evans & Scaillet (1997) showed that
Given an estimate of the fO , fS was estimated using
2
2
1306

AGUE et al.
HIGH fO DURING SILLIMANITE ZONE METAMORPHISM
2
Table 5: Biotite (Bt) and muscovite (Ms) compositions
Sample Phase
Si
Ti
Aliv
Alvi
FeT
Mn
Mg
Ba
Na
K
F
Cl
Total
36A
45A
46B
49A
50B
52C
54A
55A
56A
57B
61B
62A
66A
70A
71A
Bt
Bt
Bt
Bt
Bt
Bt
Bt
Bt
Bt
Bt
Bt
Bt
Bt
Bt
Bt
2·768
2·767
2·751
2·771
2·759
2·756
2·733
2·741
2·735
2·736
2·742
2·756
2·732
2·737
2·730
2·696
3·100
3·108
3·102
3·155
3·122
3·127
3·107
3·097
3·093
3·098
3·109
3·096
3·095
3·108
3·104
3·077
0·121
0·133
0·118
0·128
0·135
0·131
0·107
0·121
0·120
0·151
0·097
0·098
0·135
0·126
0·118
0·060
0·055
0·048
0·060
0·042
0·039
0·040
0·041
0·043
0·045
0·052
0·039
0·055
0·043
0·045
0·051
0·028
1·232
1·233
1·249
1·230
1·242
1·244
1·267
1·259
1·265
1·264
1·258
1·244
1·268
1·263
1·270
1·304
0·900
0·892
0·898
0·845
0·878
0·873
0·893
0·903
0·907
0·902
0·891
0·904
0·905
0·892
0·896
0·923
0·357
0·321
0·381
0·340
0·330
0·337
0·370
0·378
0·381
0·339
0·388
0·392
0·367
0·369
0·416
0·558
1·727
1·709
1·769
1·666
1·730
1·729
1·758
1·743
1·730
1·724
1·745
1·790
1·747
1·719
1·750
1·738
1·218
1·288
1·313
1·289
1·228
1·392
1·247
1·182
1·212
1·292
1·265
1·232
1·246
1·185
1·337
0·879
0·192
0·216
0·120
0·234
0·188
0·192
0·163
0·183
0·199
0·190
0·189
0·101
0·183
0·208
0·167
0·205
0·010
0·018
0·008
0·035
0·008
0·009
0·004
0·004
0·006
0·004
0·002
0·003
0·003
0·004
0·001
0·015
0·001
0·001
tr.
1·163
1·105
1·049
1·064
1·155
0·993
1·166
1·185
1·150
1·062
1·133
1·148
1·100
1·186
1·008
1·378
0·074
0·085
0·082
0·120
0·090
0·085
0·076
0·081
0·085
0·082
0·080
0·076
0·079
0·088
0·077
0·117
0·008
0·007
0·009
0·008
0·008
0·006
0·004
0·006
0·006
0·006
0·006
0·005
0·009
0·006
0·005
n.d.
0·028
0·027
0·049
0·016
0·031
0·027
0·054
0·040
0·038
0·031
0·042
0·031
0·047
0·043
0·046
0·076
0·125
0·094
0·110
0·067
0·116
0·124
0·183
0·146
0·143
0·127
0·155
0·136
0·133
0·139
0·129
0·160
0·850
0·878
0·829
0·892
0·883
0·883
0·832
0·850
0·855
0·883
0·853
0·868
0·864
0·848
0·804
0·769
0·824
0·864
0·828
0·876
0·839
0·830
0·776
0·809
0·804
0·833
0·795
0·808
0·817
0·808
0·810
0·765
0·060
0·082
0·108
0·068
0·092
0·068
0·064
0·081
0·052
0·096
0·075
0·173
0·094
0·074
0·077
tr.
0·003
0·001
0·003
0·001
tr.
96·12
95·76
95·94
95·87
95·48
95·61
96·11
95·68
96·11
95·96
95·48
95·88
95·80
95·27
95·98
95·13
94·65
94·19
94·35
94·89
94·52
94·00
95·23
95·10
94·58
94·84
94·73
94·85
94·30
94·95
94·39
94·34
0·001
tr.
tr.
tr.
tr.
0·005
0·003
tr.
0·001
0·001
n.d.
tr.
Chin11∗ Bt
Ms
36A
45A
46B
49A
50B
52C
54A
55A
56A
57B
61B
62A
66A
70A
71A
0·009
0·006
0·005
0·014
0·008
0·009
0·009
0·009
0·010
0·010
0·006
0·008
0·010
0·007
0·008
n.d.
0·022
0·011
0·024
0·019
0·024
0·017
0·011
0·026
0·011
0·011
0·009
0·009
0·013
0·021
0·011
n.d.
Ms
Ms
Ms
Ms
Ms
Ms
Ms
Ms
Ms
Ms
Ms
Ms
Ms
Ms
tr.
tr.
0·001
0·001
tr.
0·001
tr.
tr.
0·001
0·001
0·001
0·001
tr.
tr.
tr.
tr.
tr.
tr.
0·001
tr.
tr.
tr.
0·001
tr.
tr.
0·001
n.d.
Chin11∗ Ms
0·014
Eleven oxygen formula basis. n.d., not determined; tr., trace (<0·001). All Fe as Fe2+. Total is total weight percent. Oxygen
equivalents of the fluorine and chlorine atoms have not been subtracted from the weight percent totals.
∗Sample 11 of Chinner (1960).
current RHOMOX models overestimate fO for highly Fe2TiO4 system, the appropriate Mag activities for our
2
oxidized Mt Pinatubo dacites (see below).
samples are between about 0·8 and 1·0 (Andersen &
Retrograde Ti loss commonly alters the bulk com- Lindsley, 1988; Ghiorso & Sack, 1991); we used a very
position of Mag. However, because ‘peak’ metamorphic conservative range between 0·75 and 1·0 herein. Other
T was < >700°C, the expected XMag for spinel coexisting components such as MgAl2O4 could also have affected
with the measured RHOMOX compositions would have Mag activity, but they are present in only trace amounts
been large and between about 0·80 and 1·0 (Andersen (Table 4); their concentrations are generally significant
& Lindsley, 1988; Ghiorso & Sack, 1991). The Ti-poor only at much higher T (Ghiorso & Sack, 1991).
Mag inclusions in garnet (Table 4) support this idea
Po that coexists with Py in the Fe–S system deviates
because they should have been shielded from retrograde from ideal FeS by as much as Fe0·875S at high T. However,
equilibration with matrix phases and thus preserve pro- natural high-T Po compositions commonly equilibrate
grade compositions. For T of 650–700°C in the Fe3O4– down to very low T, at which many complex phase
1307

JOURNAL OF PETROLOGY
VOLUME 42
NUMBER 7
JULY 2001
1308

AGUE et al.
HIGH fO DURING SILLIMANITE ZONE METAMORPHISM
2
Fig. 3. Pressure–temperature estimates. Ε, method (I); Χ, method
(II) (see text). Samples 46B, 54A, 55A, 57B, 61B, 62A, 66A, 70A, and
71A. Al2SiO5 phase relations from Berman (1988).
pure Al2SiO5 (XAl SiO >0·99) and thus can be modeled
5
using unit activity2. Qtz was assumed to be pure SiO2.
Thermodynamic data for Mag and most silicate com-
ponents were taken from Berman (1988). Data for Hem,
Ann, Alm, and F-bearing micas were taken from Ghiorso
(1990), McMullin et al. (1991), Berman (1990), and Zhu
& Sverjensky (1992), respectively, and are internally
consistent with Berman (1988). Results computed with
these standard state properties are nearly identical to
those computed using recently revised properties for Alm
and Mag (Berman & Aranovich, 1996). For example,
the two datasets yield fO values for reaction (1), the main
2
reaction we used to estimate fO , that differ by only 0·04
log units. We used data from 2Johnson et al. (1992), as
updated by Shock et al. (1997) and Sverjensky et al. (1997),
for sulfide minerals, graphite, and most fluid species.
Data for the gases CO, HF, and COS are not included
in the Johnson et al. (1992) database and were thus taken
from Robie et al. (1979), Stull & Prophet (1971), and
Wagman et al. (1982), respectively.
Speciation calculations were carried out for C–O–H–S
fluids composed of H2O, H2, CO2, CO, CH4, COS, S2,
SO2, and H2S. Fugacity and activity coefficients for H2O
and CO2 were from Kerrick & Jacobs (1981), whereas
the other fugacities were computed using the equations
of Shi & Saxena (1992) assuming ideal mixing of non-
ideal gas species. Solution of the non-linear system of
speciation equations was carried out using standard pro-
cedures (e.g. Ferry & Baumgartner, 1987; Holloway,
1987).
Fig. 2. Backscattered electron images of rhombohedral oxides (RHO-
MOX) included within interior (a) and rim (b) of garnet in sample
66A. RHOMOX inclusions are more ilmenite rich towards rim of
garnet. Ilmenite-rich lamellae, dark gray; hematite-rich lamellae, light
gray. Multiple sets of exsolution lamellae in (a) should be noted.
Composition of hematite-rich domains in (a) (including fine ilmenite-
rich lamellae): XIlm = 0·32; XHem = 0·67; XGk = 3·9 × 10−4; XPyr
=
7·8 × 10−3. Ilmenite-rich domains in (a) (including fine hematite-rich
lamellae): XIlm = 0·84; XHem = 0·13; XGk = 1·0 × 10−3; XPyr = 2·3
× 10−2. White, equant grain at left in (b) is zircon (Zrn).
transitions occur. In view of these retrograde com-
plications, we make the simplification of setting the
activity of Po to 1·0. This simplification introduces some
error, but the activity corrections for even highly non-
stoichiometric Po have minor impact on the resolution
of the large variations in fugacities of interest here. Py is
relatively pure and its activity was set to 1·0.
Activity coefficients for garnet components, Ann, and
Ms were computed following Berman (1990), McMullin
et al. (1991) and Chatterjee & Froese (1975), respectively.
Because the rocks are in the sillimanite zone and
nearly all samples contain Sil, thermodynamic data for
sillimanite were used in the calculations; use of Ky has
FUGACITIES AND MINERAL
CHEMISTRY
Oxygen fugacity
little impact on the results. Reconnaissance analyses Results are given in terms of ꢂQFM defined here as the
indicate that the aluminosilicate minerals are relatively difference, at 660°C and 5·7 kbar, between the log10 fO
2
1309

JOURNAL OF PETROLOGY
VOLUME 42
NUMBER 7
JULY 2001
ꢂQFM Mag estimates. The six samples that do cluster
near the 1:1 correlation line (37C-2, 39B, 52C, 61B,
66A, 71A) all have RHOMOX compositions that lie on
the Ilm-rich side of the Hem–Ilm solvus. Agreement
between ꢂQFM Hem and ꢂQFM Mag for these samples
may be due to the fact that most experiments in the
Fe2O3–FeTiO3 system have used Ilm-rich compositions
(see Ghiorso, 1990). Consequently, the activity–
composition relations for Ilm-rich RHOMOX are prob-
ably reasonably well understood. We note that ꢂQFM
Hem for sample 39B, which plots somewhat farther from
the 1:1 line than the other five samples, is based on the
composition of matrix RHOMOX. The RHOMOX in
this sample may have undergone some retrograde Ti
gain by exchange with matrix phases during cooling
(see above), and its ꢂQFM Hem may be somewhat
underestimated.
The samples that plot well above the 1:1 line have
compositions that are either on the solvus or to the Hem-
rich side (Fig. 4). Problems with current RHOMOX
activity models near the solvus have been documented
(Evans & Scaillet, 1997), and the Hem-rich side of
the solvus remains to be fully explored experimentally.
Consequently, the ꢂQFM Hem values for these samples
are inferred to be less reliable than those for Ilm-rich
compositions. Furthermore, the RHOMOX in samples
that plot well above the 1:1 line contain complex ex-
solution textures. It is possible that these grains originally
Fig. 4. ꢂQFM computed using hematite-bearing reactions (ꢂQFM
Hem) vs ꢂQFM computed using magnetite-bearing ones (ꢂQFM Mag).
Results obtained using reactions (1) and (4) for aluminosilicate +
garnet-bearing assemblages, reactions (2) and (5) for aluminosilicate-
free assemblages, and reactions (3) and (6) for garnet-free assemblages.
When possible, ꢂQFM Hem was estimated using compositions of
RHOMOX grains included in garnet rims or in aluminosilicate min-
erals. For samples lacking RHOMOX inclusions in porphyroblasts,
compositions of RHOMOX in the matrix as far removed as possible
from micas were used (Table 3). Length of black bars corresponds to
ꢂQFM Mag calculated for the range in magnetite activities from 0·75
to 1·0 (see text). In this and other plots, analytical variability in Grt straddled the solvus and were thus composed of two
almandine content yields uncertainties in ꢂQFM Mag of > 0·03
log10 units (not shown). ꢂQFM Hem and ꢂQFM Mag agree reasonably
case, our ‘integrated’ compositions, which are averaged
well for samples with RHOMOX compositions on the ilmenite-rich
coexisting RHOMOX compositions at peak T. In this
over all exsolution domains in RHOMOX grains, would
be in error and could not be used in thermodynamic
calculations. Resolution of the above issues requires a
better understanding of phase relations among rhom-
bohedral oxides.
side of the ilmenite–hematite solvus (37C-2, 39B, 52C, 61B, 66A, 71A).
High fO sample 11 of Chinner (1960) denoted by open rectangle.
Samples2 46B and 62A lack magnetite, but are plotted on the 1:1
correlation line for comparison purposes.
To avoid problems with the thermodynamic treatment
of RHOMOX, we focus on ꢂQFM values computed
using Mag-bearing reactions (Table 7). The two ex-
ceptions are samples 46B and 62A, which lack Mag
and require ꢂQFM estimates based on Hem-bearing
reactions. Because these two samples contain Ilm-rich
RHOMOX that lacks exsolution features, they should
be treated reasonably well with the activity model and
standard state thermodynamic data used herein.
The agreement between ꢂQFM estimates made using
reactions (1), (2), and (3) for Mag-bearing samples, or
reactions (4), (5), and (6) for Mag-free samples (Table 7),
indicates that the thermodynamic data and activity mod-
els employed are internally consistent.
for a sample and that for the quartz–fayalite–magnetite
(QFM) buffer (QFM log10 fO = −17·46). Because T–
2
fO trajectories for rocks are generally subparallel to that
fo2r QFM, use of ꢂQFM notation largely removes un-
certainties caused by errors in T estimates and, therefore,
highlights real fO variations between samples. Fur-
2
thermore, the reactions used to estimate fO are fairly
2
insensitive to P variations, thus minimizing errors caused
by errors in the P estimate.
ꢂQFM values computed using Hem-bearing reactions
(ꢂQFM Hem) are plotted against the corresponding
Mag-bearing reactions (ꢂQFM Mag) in Fig. 4. Most
rocks have ꢂQFM Hem between +1·75 and +3·0 and
ꢂQFM Mag of +1·8 to +2·3. However, ꢂQFM values
for the Mag-free samples 46B and 62A are 2–3 orders
of magnitude lower than those for the other rocks (ꢂQFM
−0·25 to −0·75; Fig. 4).
Mineral chemistry
Closer inspection reveals that the ꢂQFM Hem es- Prograde RHOMOX vary from XHem >0·02 to >0·75
timates are as much as >10 times greater than the (Table 3). XHem values for inferred retrograde RHOMOX
1310

AGUE et al.
HIGH fO DURING SILLIMANITE ZONE METAMORPHISM
2
Table 7: Fugacity and rock compositions
Sample
ꢂQFM(1)
ꢂQFM(2)
2·05–2·30
ꢂQFM(3)
FeO
Fe2O3
OR
ꢂHF/H2O
0·100
Elevation (m)
36A
37C-2
38A
39B
325
440
505
405
365
445
460
263
315
370
450
390
310
340
430
480
510
385
420
460
490
490
525
450
410
350
370
400
450
485
495
445
385
1·86–2·11
1·80–2·05
1·85–2·10
1·86–2·11
1·91–2·16
1·96–2·21
7·2
3·00
31·6
40A
41C
42A
45A
46B
1·93–2·18
−0·34†
0·266
0·417
−0·36∗
−0·33‡
47A1
48E
1·92–2·17
1·84–2·09
7·3
3·49
35·4
49A
50B
1·96–2·21
1·94–2·19
0·194
0·294
51B
1·78–2·03
52C
1·70–1·95
1·87–2·12
1·97–2·22
0·227
0·136
0·230
0·047
0·355
54A
55A
56A
57B
1·89–2·14
2·01–2·26
1·84–2·09
1·93–2·18
1·82–2·07
1·67–1·92
4·6
8·0
2·84
4·41
43·5
39·7
1·87–2·12
1·91–2·16
2·02–2·27
1·87–2·12
1·69–1·94
−0·75∗
1·77–2·02
58A
59A
60A
61B
6·6
10·2
8·0
3·77
2·76
1·81
40·9
21·7
18·5
1·73–1·98
−0·69†
1·77–2·02
−0·63‡
0·216
0·595
62A
63A-2
64B-2
65A
66A
67A
68A
69B
1·91–2·16
1·80–2·05
1·85–2·10
1·94–2·19
2·17–2·42
1·81–2·06
1·87–2·12
1·92–2·17
1·70–1·95
2·51–2·76
5·6
7·6
4·28
2·85
51·1
28·9
1·87–2·12
1·79–2·04
0·326
70A
71A
Chin11§
1·91–2·16
1·68–1·93
2·56–2·81
1·90–2·15
1·66–1·91
2·60–2·85
5·3
5·2
3·7
3·52
2·23
46·0
32·4
74·8
0·190
0·279
12·01
ꢂQFM (1), ꢂQFM (2), ꢂQFM (3) computed using reactions (1), (2), and (3), respectively, except where noted otherwise. Range
in values corresponds to the magnetite activity range 0·75–1·0. FeO and Fe2O3 are bulk-rock weight percents. OR, rock
oxidation ratio. ꢂHF/H2O is the difference in log10( fHF /fH O) between the sample and log10( fHF /fH O) for a standard biotite having
2
XMg = 0·5 and log10(XF/XOH) = −1·5 at 660°C and 5·7 k2bar.
∗Computed using reaction (4).
†Computed using reaction (5).
‡Computed using reaction (6).
§Sample 11 of Chinner (1960).
intergrown with micas are smaller than >0·15 and are cases, retrograde St or Ky commonly contain significant
significantly smaller than the XHem of crystals included in quantities of RHOMOX. These yield XHem values that
nearby porphyroblasts (Table 3).
are equal to or smaller than those for inclusions in
‘Felted mats’ of Sil intergrown with micas and, in some porphyroblasts (Table 3). It is unclear if the smaller
1311

JOURNAL OF PETROLOGY
VOLUME 42
NUMBER 7
JULY 2001
Fig. 6. Spessartine/almandine mole fraction ratio (XSps/XAlm) vs
ꢂQFM. High fO sample 11 of Chinner (1960) denoted by open
2
rectangle.
necessarily with fO . The XSps/XAlm ratio is small for
2
ꢂQFM <0, whereas a wide range of XSps/XAlm values
seem compatible with ꢂQFM of >+2 (Fig. 6). Sample
11 of Chinner (1960) is somewhat anomalous relative
to our samples and has the largest XSps/XAlm and fO
2
(Fig. 6). We note in this regard the possibility that
the bulk garnet composition given by Chinner (1960)
may be more Mn rich than the peak-T composition
at garnet rims.
T
The XMg/XFe ratio for biotite changes little over the
Fig. 5. Chemical traverses from core to rim across typical garnet
in sample 66A. ꢁVI and ꢁVIII denote cation sums for six-fold and
eight-fold coordinated sites, respectively.
ꢂQFM range between −1·0 and +2·3 (Fig. 7; FeT
is all Fe as FeO). However, XMg/XFe increases at the
T
largest ꢂQFM, consistent with the expectation that
higher fO should favor more Mg-rich compositions
(e.g. Wones & Eugster, 1965). The XMg/XFe of muscovite
2
values resulted from input of reducing fluid, retrogression,
or both. Retrograde effects are difficult to rule out because
the RHOMOX are closely intergrown with micas.
Samples 57B and 68A appear to have two coexisting
prograde RHOMOX compositions that occur as dis-
crete grains. Using the solvus diagram of Ghiorso
(1990), compositions in 68A suggest a T near 680°C,
in good agreement with the T estimates based on
thermobarometry, whereas compositions in 57B are
inconsistent with equilibration at a single T. The
significance of these results is unclear given continuing
uncertainties regarding T–activity–composition relations
for RHOMOX.
T
(Ms) decreases between ꢂQFM −1·0 and +2·3, and
then increases somewhat for higher ꢂQFM (Fig. 7).
Fe3+ is the dominant form of Fe in Glen Clova Ms
(Chinner, 1960). Under reducing conditions, the total
Fe content of Ms is relatively small because the amount
of Fe3+ available is limited relative to more oxidized
T
conditions. Consequently, XMg/XFe ratios for Ms in
the low-fO rocks are large. The slight increase in XMg/
2
T
XFe above ꢂQFM +2 reflects the stability of magnesian
compositions at higher fO .
2
Garnet Fe3+ is generally largest in cores and smallest
in rims (Fig. 5). Fe3+ and Ti4+ decrease with Ca (Fig.
5), such that the ratio of grossular to andradite
component remains fairly constant. Garnet Fe3+ con-
tents are small and no correlation between ꢂQFM
and garnet Fe3+ is evident, consistent with Chinner
(1960). The Fe3+ systematics are apparently sensitive
Changes in RHOMOX inclusion
compositions from core-to-rim in garnet
Some samples were examined in detail to determine
if RHOMOX inclusion compositions changed from
core to rim in garnets. This effort was hampered by
the relatively small size and paucity of inclusions in
mostly to crystal chemical factors and the availability many of the garnet crystals. None the less, RHOMOX
of Ca, rather than fO .
included in the interiors of garnet porphyroblasts in
Garnet Mn content2is in general positively correlated samples 61B and 66A were measured to have sig-
with rock Fe3+/Fe2+ (Chinner, 1960), although not nificantly smaller XIlm than RHOMOX inclusions near
1312

AGUE et al.
HIGH fO DURING SILLIMANITE ZONE METAMORPHISM
2
Fig. 8. Rhombohedral oxide (RHOMOX) ilmenite mole fractions
for samples 61B and 66A. Χ, RHOMOX inclusions in garnet; Α,
inclusions in kyanite rims; Φ, inclusions in sillimanite + muscovite
aggregates. Generalized changes in RHOMOX compositions during
garnet growth denoted by dashed arrows.
orders of magnitude less than the other rocks (Fig. 9;
see below).
Fig. 7. Mg/FeT (all Fe as Fe2+) mole fraction ratio vs ꢂQFM for
biotite (Bt; upper panel) and muscovite (Ms; lower panel). High
REGIONAL VARIATIONS
Regional variations in fO
2
fO sample 11 of Chinner (1960) denoted by open rectangles.
2
Regional fO profiles were constructed for the northwest,
2
central, and southeast parts of the field area (Fig. 1).
Because the fabric of the rocks is nearly horizontal,
sample data were projected onto vertical sections such
that the profiles are essentially perpendicular to the fabric.
A small number of samples (46–49) are located between
the northwest and central profile areas and are thus not
shown in Figs 10 and 11.
ꢂQFM for the northwestern and southeastern profiles
shows no large variations as a function of elevation (Fig.
10a and c). The main feature of the central profile is the
strong localized minimum in ꢂQFM associated with
sample 62A (Fig. 10b).
garnet rims, or those included within Sil + Ms
aggregates and Ky rims (Figs 2 and 8; Table 3). One
interpretation of this shift towards larger XIlm is that
reducing fluids, either locally or externally derived,
caused reduction of the mineral assemblage during
garnet growth. Unfortunately, uncertainties in activity–
composition relations for RHOMOX preclude full
quantification of possible fO changes recorded by the
2
increases in XIlm, although calculations using the Ghiorso
(1990) model suggest that drops in fO as large as an
order of magnitude occurred during2 garnet growth.
Because RHOMOX compositions between the small
XIlm values in Grt interiors and the large XIlm values
in rims were not observed, any such reduction was
probably rapid compared with the rate of Grt growth.
Regional variations in activities of Mg, Fe,
and Mn species in solution
2+
2+
Surprisingly, the calculated log10(aMg /aFe ) of the fluid
varies little and shows no significant correlations with
elevation or ꢂQFM (Fig. 11). Nearly all values are within
10% of the mean. Other related ratios, such as log10
Relationships between fO and fS
2
2
Most sulfide-bearing samples record fS at or slightly
below the Py–Po buffer (Fig. 9). The2 average fS of
+
+
(aMgCl /aFeCl ), would show exactly the same patterns of
2
Py-bearing rocks is slightly higher than that for Po-
regional variation [of course, the absolute values would
bearing ones, consistent with thermodynamic pre- be different from log10(aMg2+/aFe2+)]. These results strongly
dictions. The two samples with the smallest ꢂQFM suggest that no large gradients in fluid Mg/Fe existed
(46B, 62A) are the only ones that could have been in between layers. ꢂQFM for all samples must have been
equilibrium with graphite. These record fS values 2–3 low enough so that Fe2+ species were dominant, otherwise
2
1313

JOURNAL OF PETROLOGY
VOLUME 42
NUMBER 7
JULY 2001
Fig. 9. ꢂPyPo vs ꢂQFM. ꢂPyPo is the difference between log10
fS for the sample and log10 fS for the pyrite–pyrrhotite buffer at
2
660°C and 5·7 kbar. For rocks2 that contain magnetite (all those at
ꢂQFM above graphite stability), gray and black bars denote pyrite-
bearing and pyrrhotite-bearing samples, respectively. Length of bars
corresponds to ꢂQFM calculated for the range in magnetite activities
from 0·75 to 1·0 (see text). It should be noted that only the two
magnetite-free, pyrrhotite-bearing samples (46B, 62A) fall within the
stability field of graphite.
2+
2+
the log10(aMg /aFe ) would vary more strongly as a
function of fluid Fe3+/Fe2+
In contrast, log10(aMn /aFe ) varies significantly and
.
Fig. 10. Regional profiles of ꢂQFM across layering for the north-
western, central, and southeastern parts of the field area (Fig. 1). Z is
the elevation of the sample in meters above sea level. Length of black
bars corresponds to ꢂQFM calculated for the range in magnetite
activities from 0·75 to 1·0 (see text, Table 7). Sample 62A plots off-
scale at ꢂQFM −0·75. Samples 61B and 66A discussed in text.
2+
2+
shows a weak positive correlation with fO (Fig. 11). In
2
2+
2+
most cases the log10(aMn /aFe ) variations are not abrupt
and restricted to individual beds, but rather are grad-
ational at the 10–100 m scale. Examples include the
2+
2+
gradual decrease in log10(aMn /aFe ) from 460 to 360 m
in the northwest profile (Fig. 11a), and the ‘C’-shaped
variation between 300 and 425 m in the southeast profile
(Fig. 11c). The sharper discontinuity in the central profile
corresponds to the extremely low fO sample 62A, and
may be of metasomatic origin (see be2low). Metamorphic
fluid–rock interaction was insufficient to homogenize
fluid Mn2+/Fe2+ in the sequence, possibly owing to low
Mn concentrations in fluids (i.e. large solid–fluid partition
coefficient for Mn).
would be at the same fO as fluid in a rock containing
99% Mag and 1% Hem, 2even though the OR values for
the two rocks differ considerably.
In fact, a wide range of OR values, between about 20
and 50, are compatible with essentially the same ꢂQFM
of about +2 (Fig. 12). Consequently, the observation of
intimately interbedded rocks of differing OR (Chinner,
1960) does not necessarily indicate that the rocks had
very different fO and fluid compositions. Extremes of the
OR range are,2 however, associated with extremes of
fO . Sample 62A, equilibrated at ꢂQFM <0, has OR
<20, whereas sample 11 of Chinner (1960) has an OR
2
DISCUSSION OF fO SYSTEMATICS
2
of >75 and ꢂQFM > +2·5. These extremes of the
Chinner (1960) used the rock oxidation ratio (OR) as an
index of rock oxidation state, where OR = 2Fe2O3 ×
100/(2Fe2O3 + FeO) (molecular basis). Although clearly
fO range are discussed in a later section.
2
The homogeneity of the bulk of the metamorphic fO
2
powerful, OR is not a direct indicator of fO because values can be interpreted in two ways. One possibility is
rocks of differing OR can equilibrate at the 2same fO . that the lack of large variations reflects a protolith that
For example, for equilibrium at a given P and T, por2e was fairly homogeneous with respect to mineral and fluid
fluid in a rock that contains 99% Hem and 1% Mag compositions to start with (although OR values must
1314

AGUE et al.
HIGH fO DURING SILLIMANITE ZONE METAMORPHISM
2
The second possibility is that the sequence was ori-
ginally more heterogeneous, such that sharp, layer-to-
layer variations in redox state controlled by protolith
composition existed. Metamorphic fluid–rock interaction
between layers must then have acted to smooth these
short wavelength variations so that the entire sequence
attained a relatively uniform redox state. The observed
increases in the XIlm and decreases in the XHem of RHO-
MOX inclusions from core to rim in garnet (Figs 2 and
8) suggest that some syn-metamorphic changes in redox
state did occur.
If redox states were homogenized during meta-
morphism, the reaction histories would have been crit-
ically dependent on fluid compositions and the amount
of fluid transport. The highly oxidized, ꢂQFM +2 fluids
of Mag-bearing rocks would have been fairly inefficient
oxidizers or reducers because such fluids consist mostly
of H2O and CO2, and are poor in many species needed
for redox reactions including H2, CH4, and CO (Wood
& Walther, 1986; Frost, 1991a; Ague, 1998). For example,
the reactions
3H2 + 3Hem + Ms + 3Qtz = Ann +
Alm + 3H2O
(11)
0·75CH4 + 3Hem + Ms + 3Qtz = Ann + Alm +
0·75CO2 + 1·5H2O (12)
require either 1 mole of H2 or 0·25 moles of CH4 to
reduce 1 mole of Hem and produce garnet and biotite.
Typical ꢂQFM +2 Glen Clova fluids, however, probably
2+
Fig. 11. Elevation (Z) in meters above sea level vs fluid log10(aMn
/
2+
2+
2+
aFe ) − log10 K [reaction (10)] and log10(aMg /aFe ) [reaction (9)] for
northwestern, central, and southeastern profiles.
had XH of only >10−4 and vanishingly small XCH
<
2
4
>10−7 (Table 8). Consequently, large fluid–rock ratios
would have been needed to homogenize the rock package.
Quantification of these ratios is difficult owing to the
uncertain activity–composition relations for RHOMOX,
but calculations using simple model reactions such as (11)
and (12), and representative fluid compositions (Table 8)
and modes (Chinner, 1960) imply that if homogenization
occurred, minimum fluid–rock ratios were >102 (Wood
& Walther, 1986; Ague, 1998). Fluid exchange between
beds would probably have been dominated by diffusion
and mechanical dispersion across layers (e.g. Ague &
Rye, 1999; Baxter & DePaolo, 2000), although advection
may also have played a role. Modeling results suggest that
disruption of protolith bed-scale fugacity heterogeneities
may occur in as little as 102−103 years (Ague, 1998).
Fig. 12. Rock oxidation ratio (OR) vs ꢂQFM. High fO sample 11 of
2
Chinner (1960) denoted by open rectangle.
PETROGENESIS OF SAMPLES AT
THE EXTREMES OF THE fO RANGE
2
have differed from bed to bed). Metamorphism of such
a sequence would not be expected to produce large
Rocks equilibrated at extremely low fO
2
gradients in fO and fluid composition between layers, Two samples (46B, 62A), making up >6% of the sample
unless ‘exotic’,2 nonequilibrium fluids from outside the set, are from layers that equilibrated at fO 2–3 orders
sequence were introduced.
of magnitude lower than the rest of the rock2s. The layers
1315

JOURNAL OF PETROLOGY
VOLUME 42
NUMBER 7
JULY 2001
Table 8: Fluid mole fractions
Species
Representative
Representative
Representative
Representative
Sample 62A
ꢂQFM +1·8
ꢂQFM +1·8
ꢂQFM +2·4
ꢂQFM +2·4
ꢂQFM −0·75
XH = 0·50∗
2O
XH = 0·95∗
2O
XH = 0·50∗
2O
XH = 0·95∗
2O
H2
1·18×10−4
5·00×10−1
4·96×10−1
2·13×10−4
1·28×10−7
3·39×10−3
6·00×10−5
3·02×10−4
1·56×10−6
1·78×10−4
9·50×10−1
4·44×10−2
4·35×10−5
5·90×10−8
5·10×10−3
6·00×10−5
6·17×10−5
1·56×10−6
5·93×10−5
5·00×10−1
4·98×10−1
1·07×10−4
8·09×10−9
1·70×10−3
2·39×10−4
1·52×10−4
1·56×10−6
8·91×10−5
9·50×10−1
4·71×10−2
2·29×10−5
3·91×10−9
2·55×10−3
2·39×10−4
3·25×10−5
1·56×10−6
3·01×10−2
8·04×10−1
1·71×10−1
2·10×10−3
1·53×10−2
2·68×10−3
7·55×10−9
1·33×10−4
3·11×10−9
H2O
CO2
CO
CH4
H2S
SO2
COS
S2
aGraphite
5·39×10−3
−0·25
1·10×10−3
−0·25
1·36×10−3
−0·25
2·90×10−4
−0·25
1·00×100
−2·95
ꢂPyPo†
∗Calculated for representative range in XH O based on McLellan (1985).
2
†ꢂPyPo is the difference between log10fS for the fluid and log10fS for the pyrite–pyrrhotite buffer at 660°C and 5·7 kbar.
2
2
are at least meter scale in thickness, but whether contacts into the low fO rocks, and draw the connection that these
2
with surrounding rocks are sharp, gradational, tectonic, fluids also drove syn-metamorphic reduction. The HF/
or otherwise could not be determined in the field. Either H2O relations might also have been inherited from sedi-
the sedimentary or diagenetic precursors to these two mentary protoliths, but it is difficult to envision what
samples were highly reduced, or reduction resulted from sedimentary process could produce a negative correlation
metamorphic fluid–rock interaction. Some evidence from between HF/H2O and ꢂQFM (Fig. 14).
graphite and fluorine systematics suggests the latter pos-
The above evidence suggests that infiltration of reduced
sibility.
metamorphic fluids drove reaction and lowered fO in
2
Graphite, inferred to have been derived from sedi- samples 46B and 62A. If these fluids were at or near
mentary organic matter, is common in low-OR rocks far graphite saturation, they would have contained abundant
removed (kilometer scale) from the study area (Chinner, H2 and CH4 (sample 62A, Table 8). Because these species
1960). In contrast, graphite is absent from the lowest are powerful reducing agents, alteration would have
fO sample (62A). This sample probably originally lacked occurred even if fluid–rock ratios were rather small—of
2
or contained only traces of organic matter and, thus, its the order of one or less (Wood & Walther, 1986; Ague,
protolith may not have been highly ‘reduced’. Con- 1998). The fluid could have been derived from graphitic
sequently, the low fO could have been produced during rocks, such as those that surround the sequence studied
2
here, or from reduced magmas that intruded during peak
metamorphism.
metamorphism. If the low fO was produced during meta-
morphism, the localized reduction suggests channelized
fluid infiltration.
Traces of graphite may be included within a few mica
grains in 46B, but the identification is uncertain. On the
other hand, the sample is cut by millimeter- to centimeter-
wide, post-peak metamorphic shear zones containing
abundant carbonaceous material, plagioclase, muscovite,
and chlorite, and minor calcite (Fig. 13). Reducing fluids
were therefore able to infiltrate the rock, albeit not
necessarily on the prograde path.
2
High fO sample
2
Sample 11 of Chinner (1960) has both large OR and
ꢂQFM relative to our samples (Fig. 12). These char-
acteristics were probably inherited from the protolith
The low fO samples record larger HF/H2O than the
others (Fig. 124). In addition, sample 46B is cut by veins
that contain abundant F-rich tourmaline crystals as much because (1) the large Mn and Fe contents of the rock
as 1 cm long of probable metasomatic origin (see Dutrow suggest a protolith with high bulk-rock Fe2O3/FeO [see
et al., 1999). We suggest that the HF/H2O systematics discussion by Chinner (1960)], and (2) no local or regional
reflect the local infiltration of chemically active fluids sources for highly oxidized metamorphic fluids needed
1316

AGUE et al.
HIGH fO DURING SILLIMANITE ZONE METAMORPHISM
2
Fig. 13. Photomicrograph of post-peak metamorphic shear zone in sample 46B. Deformed muscovite grains (Ms) lie within chlorite-rich matrix.
Note abundant dust-like carbonaceous material. Field of view 1·35 mm wide; plane-polarized light.
(ꢂQFM >+2) at sillimanite zone conditions, although
rare low fO rocks (ꢂQFM <0) were also present. The
2
high fO of ꢂQFM >+2 is comparable with the
2
highest known from arc settings (e.g. dacite from Mt
Pinatubo; Evans & Scaillet, 1997). With the exception
of the rare low fO rocks, variations in metamorphic
2
fO across the study area were generally small, about
20·3 log10 units. This minimal variability may reflect
either (1) a protolith that was initially homogeneous with
respect to redox state or (2) a protolith characterized by
significant differences in redox state from bed to bed
that was homogenized by metamorphic fluid–rock
interaction. Rhombohedral oxide (RHOMOX) in-
clusions in garnet grains that become richer in ilmenite
Fig. 14. ꢂHF/H2O vs ꢂQFM. ꢂHF/H2O is the difference in
and poorer in hematite towards garnet rims suggest
log10( fHF /fH O) between the sample and log10( fHF /fH O) for a standard
2
2
that some syn-metamorphic reduction of rock occurred.
In detail, the ꢂQFM >+2 rocks comprise intimately
intercalated, graphite-free metasedimentary layers with
whole-rock oxidation ratios (OR) that vary mostly
between >20 and >50 from one layer to the next
[OR = molecular 2Fe2O3 × 100/(2Fe2O3 + FeO)].
These OR variations are probably inherited from
biotite having XMg = 0·5 and log10(XF/XOH)= −1·5 at 660°C and
5·7 kbar. High fO sample 11 of Chinner (1960) denoted by open
2
rectangle. F in this sample was below detection (Chinner, 1960). Its F
content was set to the smallest value measured from our sample suite
and thus probably represents a maximum.
to oxidize the rock are evident. One possibility is that
sample 11 is an unusual relic that escaped to some degree
redox equilibration with the rest of the sequence, perhaps
because of extremely low porosity and/or permeability.
Another possibility is that the bulk garnet composition
of Chinner (1960) used in our calculations has lower
Fe/Mn and almandine activity than the peak-T rim
sedimentary protoliths, are uncorrelated with fO , and
2
do not indicate that large, order-of-magnitude gradients
in fO and redox state existed between these layers
durin2g metamorphism. As pointed out by Chinner
(1960), metamorphic fluid–rock interaction did not
homogenize fluid compositions at scales of several
composition. If so, then the fO estimate made using the
2
kilometers, because the high fO , ꢂQFM >+2 rocks
2
bulk garnet composition would be somewhat too high.
are in regional contact with Fe2O3-poor, graphite-
bearing lithologies. The rare low fO rocks (ꢂQFM <0)
in our field area that occur among2 the dominant high
CONCLUSION
fO (ꢂQFM >+2) rocks may have been reduced by
2
Estimated fO for the bulk of the study area was about regionally migrating fluids derived from these graphite-
2
+2 log10 units above the quartz–fayalite–magnetite buffer bearing surroundings.
1317

JOURNAL OF PETROLOGY
VOLUME 42
NUMBER 7
JULY 2001
Chinner, G. A. (1960). Pelitic gneisses with varying ferrous/ferric ratios
from Glen Clova, Angus, Scotland. Journal of Petrology 1, 178–217.
Chinner, G. A. (1961). The origin of sillimanite in Glen Clova, Angus.
Journal of Petrology 2, 312–323.
Deer, W. A., Howie, R. A. & Zussman, J. (1992). An Introduction to the
Rock-Forming Minerals. Harlow: Longman.
Dempster, T. J. & Bluck, B. J. (1995). Regional metamorphism in
transform zones during supercontinent breakup: Late Proterozoic
events of the Scottish Highlands. Geology 23, 991–994.
Dutrow, B. L., Foster, C. T., Jr & Henry, D. J. (1999). Tourmaline-
rich pseudomorphs in sillimanite zone metapelites: demarcation of
an infiltration front. American Mineralogist 84, 794–805.
ACKNOWLEDGEMENTS
We thank D. M. Rye, E. L. McLellan, R. L. Masters,
and C. J. Carson for thoughtful discussions, M. S. Ghiorso
for sharing both software and insightful advice for cal-
culating activity–composition relations among Fe–Ti ox-
ides, B. W. Evans, B. R. Frost, and an anonymous referee
for critical reviews, and S. S. Sorensen for editorial
handling. Financial support from the National Science
Foundation, Division of Earth Sciences, grant EAR-
9405889, is gratefully acknowledged.
Eugster, H. P. & Skippen, G. B. (1967). Igneous and metamorphic
reactions involving gas equilibria. In: Ableson, P. H. (ed.) Researches
in Geochemistry II. New York: John Wiley, pp. 492–520.
Evans, B. W. & Scaillet, B. (1997). The redox state of Pinatubo dacite
and the ilmenite hematite solvus. American Mineralogist 82, 625–629.
Ague, J. J. (1998). Simple models of coupled fluid infiltration and redox Ferry, J. M. & Baumgartner, L. P. (1987). Thermodynamic models of
REFERENCES
reactions in the crust. Contributions to Mineralogy and Petrology 132,
180–197.
Ague, J. J. & Brimhall, G. H. (1988). Magmatic arc asymmetry
and distribution of anomalous plutonic belts in the batholiths of
molecular fluids at the elevated pressures and temperatures of crustal
metamorphism. In: Carmichael, I. S. E. & Eugster, H. P. (eds)
Thermodynamic Modeling of Geological Materials: Minerals, Fluids, and Melts.
Mineralogical Society of America, Reviews in Mineralogy 17, 323–365.
California: effects of assimilation, cratonal thickness, and depth of Fettes, D. J., Graham, C. M., Harte, B. & Plant, J. A. (1986). Lineaments
crystallization. Geological Society of America Bulletin 100, 912–927.
and basement domains: an alternative view of Dalradian evolution.
Ague, J. J. & Rye, D. M. (1999). Simple models of CO2 release from
Journal of the Geological Society, London 143, 453–464.
metacarbonates with implications for interpretation of directions French, B. M. (1966). Some geological implications of equilibrium
and magnitudes of fluid flow in the deep crust. Journal of Petrology
40, 1443–1462.
between graphite and a C–O–H gas phase at high temperatures
and pressures. Reviews of Geophysics 4, 223–253.
Andersen, D. J. & Lindsley, D. H. (1988). Internally consistent solution
models for the Fe–Mg–Mn–Ti oxides: Fe–Ti oxides. American Min-
eralogist 73, 714–726.
Frost, B. R. (1991a). Introduction to oxygen fugacity and its petrologic
importance. In: Lindsley, D. H. (ed.) Oxide Minerals. Mineralogical
Society of America, Reviews in Mineralogy 25, 1–9.
Baxter, E. F. & DePaolo, D. J. (2000). Field measurement of slow Frost, B. R. (1991b). Stability of oxide minerals in metamorphic rocks.
metamorphic reaction rates at temperatures of 500° to 600°C. Science
288, 1411–1414.
In: Lindsley, D. H. (ed.) Oxide Minerals. Mineralogical Society of America,
Reviews in Mineralogy 25, 469–487.
Baxter, E. F., Ague, J. J. & DePaolo, D. J. (2000). Tectonometamorphic
history of the Barrovian type-locality constrained by precise Sm/
Nd garnet ages from Glen Clova, Scotland. Geological Society of America,
Abstracts with Programs, 32, A-153.
Berman, R. G. (1988). Internally-consistent thermodynamic data for
minerals in the system Na2O–K2O–CaO–MgO–FeO–Fe2O3–
Al2O3–SiO2–TiO2–H2O–CO2. Journal of Petrology 29, 445–522.
Berman, R. G. (1990). Mixing properties of Ca–Mg–Fe–Mn garnets.
American Mineralogist 75, 328–344.
Berman, R. G. (1991). Thermobarometry using multi-equilibrium
calculations: a new technique, with petrological applications. Canadian
Mineralogist 29, 833–855.
Berman, R. G. & Aranovich, L. Ya. (1996). Optimized standard state
and solution properties of minerals I. Model calibration for olivine,
orthopyroxene, cordierite, garnet, and ilmenite in the system FeO–
MgO–CaO–Al2O3–TiO2–SiO2. Contributions to Mineralogy and Petrology
126, 1–24.
Brenan, J. M., Shaw, H. F., Ryerson, F. J. & Phinney, D. L. (1995).
Mineral–aqueous fluid partitioning of trace elements at 900°C and
2·0 GPa: constraints on the trace element chemistry of mantle and
deep crustal fluids. Geochimica et Cosmochimica Acta 59, 3331–3350.
Buddington, A. F. & Lindsley, D. H. (1964). Iron–titanium oxide
minerals and synthetic equivalents. Journal of Petrology 5, 310–357.
Burton, B. P. (1991). The interplay of chemical and magnetic ordering.
In: Lindsley, D. H. (ed.) Oxide Minerals. Mineralogical Society of America,
Reviews in Mineralogy 25, 303–321.
Ghiorso, M. S. (1990). Thermodynamic properties of hematite–
ilmenite–geikielite solid solutions. Contributions to Mineralogy and Petro-
logy 104, 645–667.
Ghiorso, M. S. & Sack, R. O. (1991). Fe–Ti oxide geothermometry:
thermodynamic formulation and the estimation of intensive variables
in silicic magmas. Contributions to Mineralogy and Petrology 108, 485–510.
Harlov, D. E., Newton, R. C., Hansen, E. C. & Janardhan, A. S.
(1997). Oxide and sulfide minerals in highly oxidized, Rb-depleted,
Archaean granulites of the Shevaroy Hills massif, South India:
oxidation states and the role of metamorphic fluids. Journal of
Metamorphic Geology 15, 701–717.
Harry, W. T. (1958). A re-examination of Barrow’s older granites in
Glen Clova, Angus. Transactions of the Royal Society of Edinburgh 63,
393–413.
Harte, B. (1975). Displacement of the kyanite isograd by local variations
in oxygen fugacity within pelitic schists in Glen Lethnot, Angus,
Scotland. Transactions of the American Geophysical Union 56, 466.
Harte, B. & Johnson, M. R. W. (1969). Metamorphic history of
Dalradian rocks in Glens Clova, Esk, and Lethnot, Angus, Scotland.
Scottish Journal of Geology 5, 56–80.
Harte, B., Booth, J. E., Dempster, T. J., Fettes, D. J., Mendum, J. R.
& Watts, D. (1984). Aspects of the post-depositional evolution of
Dalradian and Highland Boundary rocks in the southern Highlands
of Scotland. Transactions of the Royal Society of Edinburgh, Earth Sciences
75, 151–163.
Holloway, J. R. (1987). Igneous fluids. In: Carmichael, I. S. E. &
Eugster, H. P. (eds) Thermodynamic Modeling of Geological Materials:
Minerals, Fluids, and Melts. Mineralogical Society of America, Reviews in
Mineralogy 17, 211–233.
Chatterjee, N. D. & Froese, E. (1975). A thermodynamic study of the
pseudobinary join muscovite–paragonite in the system KAlSi3O8–
NaAlSi3O8–Al2O3–SiO2–H2O. American Mineralogist 60, 985–993.
1318

AGUE et al.
HIGH fO DURING SILLIMANITE ZONE METAMORPHISM
2
James, H. L. (1955). Zones of regional metamorphism in the Pre-
Cambrian of northern Michigan. Geological Society of America Bulletin
66, 1455–1488.
(105 Pascals) pressure and at higher temperatures. US Geological Survey
Bulletin 1452, 456 pp.
Rogers, G., Dempster, T. J., Bluck, B. J. & Tanner, P. W. G. (1989).
A high-precision U–Pb age for the Ben Vuirich granite: implications
for the evolution of the Scottish Dalradian Supergroup. Journal of
the Geological Society, London 146, 789–798.
Rumble, D., III (1978). Mineralogy, petrology, and oxygen isotopic
geochemistry of the Clough Formation, Black Mountain, western
New Hampshire, U.S.A. Journal of Petrology 19, 317–340.
Shi, P. & Saxena, S. K. (1992). Thermodynamic modeling of the
C–H–O–S fluid system. American Mineralogist 77, 1038–1049.
Shock, E. L., Sassani, D. C., Willis, M. & Sverjensky, D. A. (1997).
Inorganic species in geologic fluids: correlations among standard
molal thermodynamic properties of aqueous ions and hydroxide
complexes. Geochimica et Cosmochimica Acta 61, 907–950.
Johnson, J. W., Oelkers, E. H. & Helgeson, H. C. (1992). SUPCRT92:
a software package for calculating the standard molal thermodynamic
properties of minerals, gases, aqueous species, and reactions from 1
to 5000 bars and 0° to 1000°C. Computers and Geosciences 18, 899–947.
Kerrick, D. M. & Jacobs, G. K. (1981). A modified Redlich–Kwong
equation for H2O, CO2, and H2O–CO2 mixtures at elevated tem-
peratures and pressures. American Journal of Science 281, 735–767.
Kretz, R. (1983). Symbols for rock-forming minerals. American Min-
eralogist 68, 277–279.
Lamb, W. M. & Valley, J. W. (1985). C–O–H fluid calculations and
granulite genesis. In: Tobi, A. & Touret, J. (eds) The Deep Proterozoic
Crust in the North American Provinces. Dordrecht: D. Reidel, pp. 119–131.
McLellan, E. L. (1985). Metamorphic reactions in the kyanite and Stull, D. R. & Prophet, H. (1971). JANAF thermochemical tables.
sillimanite zones of the Barrovian type area. Journal of Petrology 26,
789–818.
Standard Reference Data Series, US National Bureau of Standards 37, 1141
pp.
McLellan, E. L. (1989). Sequential formation of subsolidus and anatectic
migmatites in response to thermal evolution, eastern Scotland. Journal
of Geology 97, 165–182.
McMullin, D. W. A., Berman, R. G. & Greenwood, H. J. (1991).
Calibration of the SGAM thermobarometer for pelitic rocks using
data from phase-equilibrium experiments and natural assemblages.
Canadian Mineralogist 29, 889–908.
Miyashiro, A. (1964). Oxidation and reduction in the Earth’s crust
with special reference to the role of graphite. Geochimica et Cosmochimica
Acta 28, 717–729.
Ohmoto, H. & Kerrick, D. M. (1977). Devolatilization equilibria in
graphitic systems. American Journal of Science 277, 1013–1044.
Oliver, G. J. H., Chen, F., Buchwaldt, R. & Hegner, E. (2000). Fast
tectonometamorphism and exhumation in the type area of the
Barrovian and Buchan zones. Geology 28, 459–462.
Sverjensky, D. A., Shock, E. L. & Helgeson, H. C. (1997). Prediction
of the thermodynamic properties of aqueous metal complexes to
1000°C and 5 kb. Geochimica et Cosmochimica Acta 61, 1359–1412.
Thompson, J. B., Jr (1970). Geochemical reaction and open systems.
Geochimica et Cosmochimica Acta 34, 529–551.
Wagman, D. D., Evans, W. H., Parker, V. B., Schumm, R. H., Harlow,
I., Bailey, S. M., Churney, K. L. & Nuttall, R. L. (1982). The NBS
tables of chemical thermodynamic properties. Journal of Physical and
Chemical Reference Data, Supplement Number 2 11, 392 pp.
Wones, D. R. & Eugster, H. P. (1965). Stability of biotite: experiment,
theory, and application. American Mineralogist 50, 1228–1272.
Wood, B. J. & Walther, J. V. (1986). Fluid flow during metamorphism
and its implications for fluid–rock ratios. In: Walther, J. V. & Wood,
B. J. (eds) Fluid–Rock Interactions during Metamorphism. New York:
Springer, pp. 89–107.
Robie, R. A., Hemingway, B. S. & Fisher, J. R. (1979). Thermodynamic
properties of minerals and related substances at 298·15 K and 1 bar
Zhu, C. & Sverjensky, D. A. (1992). F–Cl–OH partitioning between
biotite and apatite. Geochimica et Cosmochimica Acta 56, 3435–3467.
1319

JOURNAL OF PETROLOGY
VOLUME 42
NUMBER 7
JULY 2001
was required for several fine-grained samples. Data were
typically collected along linear traverses perpendicular to
the long axes of the exsolution lamellae. At least five spot
analyses on 1–2 grains per thin section were carried out
for unexsolved RHOMOX (10 ꢀm beam), Rt (5 ꢀm
beam), Mag (focused beam), garnet (focused beam), and
micas (15 or 20 ꢀm beam). Linear traverses across garnet
involved as many as 50 determinations. Unless noted
otherwise, reported garnet analyses are for inferred ‘peak’
compositions having the highest Mg and lowest Mn
contents (usually several tens of ꢀm from the rim).
Structural formulae for RHOMOX and magnetite
were computed assuming 2 moles of cations per three
oxygens for RHOMOX, and 3 moles of cations per four
oxygens for magnetite. These procedures were judged
successful because total oxide sums incorporating es-
timated FeO and Fe2O3 are always near 100 wt %
(Tables 3 and 4). FeO and Fe2O3 in garnet can be
estimated in several ways; we assumed 2 moles of oc-
tahedrally coordinated cations per 12 oxygens. This
procedure gives the best results when tested against the
wet chemical determinations by Deer et al. (1992), and
gives total cation site occupancies very close to the
ideal number of three for the four-fold and eight-fold
coordinated sites (Table 6).
APPENDIX: ANALYTICAL AND DATA
REDUCTION PROCEDURES
Mineral composition determinations and electron micro-
scopy used the JEOL JXA-8600 electron microprobe at
Yale University. Quantitative analyses employed wave-
length-dispersive spectrometers, 20 nA Faraday cup cur-
rent, 15 kV accelerating voltage, natural and synthetic
standards, off-peak and fluorescence-corrected, non-lin-
ear mean atomic number background corrections, and
ꢃ(ꢄ z) matrix corrections. Electron microscope image
processing and qualitative energy-dispersive chemical
analyses were carried out with systems mounted on the
JXA-8600; the early stages of our study used a Kevex
Delta system and the later stages used an EDAX Phoenix
system.
All Fe was assigned to FeO for the matrix corrections.
This procedure can lead to inaccurate results for Fe3+
-
rich phases if the ‘excess’ oxygen associated with Fe2O3
is not accounted for. Consequently, oxygen was in-
corporated by difference into the matrix correction it-
erations for RHOMOX and Mag such that the sum of
all oxides and the ‘excess’ oxygen was 100 wt %. Similarly,
H2O was specified by difference in the matrix corrections
for micas such that the sum of the oxides and the H2O
was 100 wt %.
Bulk-rock X-ray fluorescence analyses were performed
by X-ray Assay Laboratories, Don Mills, Ontario,
Canada. FeO was determined using standard wet chem-
Bulk compositions for the exsolved RHOMOX were
estimated by making as many as 40 spot determinations
on several grains per thin section. A 5 or 10 ꢀm diameter
beam was used in nearly all cases, but a focused beam ical titrations.
1320