-
American Mineralogist, Volume 76, pages 1781-1810' 1991
The eye of the petrographer, the mind of the petrologist
Pnrpn RosrNsoNDepartment of Geology and Geography, University of
Massachusetts, Amherst, Massachusetts 01003, U.S.A-
Ansrucr
To bring meaningful numbers out of metamorphic rocks, one must
first understandwhat the numbers represent. An idealized objective
in metamorphic studies is to dem-onstrate that coexisting minerals
approached equilibrium. In a few rocks this can be donereasonably,
for example, in coarse-grained nodules from kimberlite that were
rapidlyquenched from mantle conditions. More typical, and more
interesting, are rocks showingdisequilibrium by mineral zoning and
reaction textures at various scales. These provide a
view, if correctly interpreted, into a sequence of events atd
P-T conditions. In makingquantitative estimates, a problem is to
decide correctly which minerals were in equilibriumwith which other
minerals, and what types of reactions and processes actually took
place.
These decisions depend on the integration of both petrography
and phase equilibriumtheory. Failing this, a false interpretation
of rock history can be made. Features in meta-morphic rocks in
central Massachusetts, including schists bearing strongly zoned
garnetsproduced by four different types ofreactions, and a newly
described occulTence ofwollas-tonite marble provide opportunities
to explore these ideas.
In many examples, rock history can only be charted through study
of the completegeologic setting at outcrop or larger scale.
Particularly spectacular are well-preserved py-
roxene granulites, gabbros, and diabases in the Scandinavian
Caledonides that are locallytransformed along shear zones into
eclogite and amphibolite. Lacking knowledge of the
surroundings, there would be no way to know that the original
hlgh f rocks survivedeclogite-facies overprinting with few to no
mineralogical traces. Their field setting dem-onstrates the
importance of structural environment, kinetics, and fluids in
producing dif-
ferent minerals from the same rock composition along the same
P'T path.
INrnonucrroN
In petrologic studies, as in field geology, the observa-tions
that are possible are conditioned by the experienceand training of
the observer. Petrologic research is anattempt to define the
sequence of events in the formationofthe rocks, the conditions
under which the events hap-pened, and why they happened. Petrology
can never betotally quantitative, i.e., "getting numbers out of
rocks,"but the petrologist's familiarity and deftness with
mea-surements, calculations, and the constraints imposed byphase
equilibria are essential. Also essential is observa-tional and
descriptive petrography from the atomicthrough the microscopic and
macroscopic to the petro-logic map scale, providing the framework
for interpreta-tions. Maps can only be made by integrating
knowledgegained from submicroscopic and microscopic scales usingthe
hand lens and unaided eye; thus the most truthfulpetrologic maps
are done by those most skilled in bothpetrography and mapping.
These ideas are illustrated byexamples where specialized knowledge
or backgroundopened the eyes of the petrographer, leading to
unex-pected interpretations. These ideas are explored in
threeparts, each chosen to represent a different scale
ofobser-vation and thinking but having in common the use ofphase
relations to try to understand the sequence of meta-
morphic events. The first part concerns the textural andphase
relations between garnet grains and their immedi-
ate surroundings. The second part deals with features at
the scale of a whole thin section in a wollastonite marblewhere
variations in local bulk compositions are shown to
have a dramatic effect on phase relations. The third sec-
tion covers relations between rocks at the scale of a large
outcrop or a mountainside, where an understanding of
the total setting is vital to understanding rock history.
A theme of this paper is "getting numbers out of rocks"and
concerns the need for careful and thorough petro-graphic
observations in thin sections and in the field if
one is to make meaningful maps of igneous and meta-
morphic rocks and if one is to combine theory and ex-periment to
make useful petrogenetic models' In the game
of getting numbers out of rocks, much of my attentionhas been
focused on the central Massachusetts Acadian
metamorphic high, shown on a metamorphic map of New
England in Figure I and in closer detail in Figure 2, wilh
metamorphic zones ranging from chlorite zone (C) in the
extreme northwest through Zone I (kyanite-muscovite-
staurolite) through Zone YI
(sillimanite-orthoclase-gar-net-cordierite). In this whole area I
am familiar with only
two "direct" numerical determinations. The first (Fig. 3)
is a slightly flawed number 8 found in a granulite facies
0003-o04x/9 l / l l l 2-l 78 r $02.00 1 7 8 I
-
t{ilA',5t(\ r , / t ;
i":#' / - l!Y,y,yt-'.1,r/
1782 ROBINSON: EYE OF THE PETROGRAPHER, MIND OF THE
PETROLOGIST
Sill iman ite-Orthoclase-Garnet-Cordieri le Zone
Sill imanite-Orthoclase Zone
Sill imanite-1,/uscovite Zone
Kyanite-Slaurolite Zone
Andalusite-Staurol t€ Zone
Garnel Zone
Bioti le and Chlorite Zon6s
Weakly Metamorphosed
"3 A p p r o x i m o t e S c o l e2OO Km
Fig. 1. Generalized map of Acadian and Taconian metamorphic
zones in southwestern New England. Modified from Osberg etal.
(1989) based on a compilation by Robinson (1983) and an unpublished
compilation, Metamorphic Map of the U.S.-CanadianAppalachians, by
G.W. Fisher and W.E. Trzcienski. Heavy dashed line represents the
approximate boundary where peak meta-morphism is Acadian to the
east and Taconian to the west.
pegmatite formed by spherical aggregates of graphiteplatelets
enclosed in a crystal of orthoclase. The second(Fig. a) is a much
larger number 2 discovered in a garnet-rich schist in the same area
by Elaine Padovani. Inconsidering the question of geothermometry
and geoba-rometry in such a region, it is natural to turn to the
com-positions ofgarnets in pelites as well as to considerationof
garnet stability and geothermobarometry in a newlydescribed
wollastonite-bearing marble.
INrnnpnnr.q,TroN oF GARNET coMposrrroNs rNMETAMORPHIC ROCKS
Aside from its aesthetic values, garnet has captured
theattention of quantitative metamorphic petrologists forthree
reasons: (l) It holds very high concentrations ofFeand Mn relative
to most other coexisting minerals, andhence distribution
coefficients are sensitive to tempera-ture. (2) It has high density
compared with most coexist-ing minerals, and hence reactions
involving it are likelyto be sensitive to pressure changes. (3)
Elemental difiil-sion is relatively slow in garnet compared to most
otherminerals, except at the highest temperatures, so that
dis-equilibrium features are commonly preserved, and theseprovide a
key, if properly interpreted, to the changingconditions of
metamorphism. These disequilibrium fea-tures are the subject ofthis
part ofthe paper.
The mention of metamorphic garnets in central Mas-sachusetts
opens a whole can of worms (Fig. 5). In thiscase the heads of the
wonns are the cores of the gamets,
and the tails are the rims. The roman numerals corre-spond to
the different metamorphic zones in central Mas-sachusetts, and the
compositional trajectories are differ-ent in each of the
metamorphic zones. In the subsectionsthat follow, each of the major
types of garnet zoning foundin central Massachusetts is discussed,
the importance ofcareful petrographic observations to deduce
equilibria andreaction mechanisms are emphasized, and the
particularpitfalls involved in making geothermometric
interpreta-tions are pointed out.
In the discussion that follows, the terms "prograde"and
"retrograde" appear frequently because they are partof the everyday
language of metamorphic petrologists, yetthey are fraught with
ambiguity in the context of complexP-7 paths. A prograde path is
commonly thought of asone involving a temperature increase, a
dehydration, orboth. Yet prograde metamorphism to a high-pressure
as-semblage might involve mainly a pressure increase withlittle or
no increase of temperature. To consider progradeas the direction in
which minerals grow would be entirelyerroneous because it is quite
possible for some mineralsto grow with decreasing temperature. The
ambiguity maybe epitomized by considering a hypothetical
equilibriumA : B with a positive P-Z slope. P-T trajectoies,
alsowith positive slopes, may be drawn to cross this reactionin
different up-temperature directions so that A goes toB or
alternatively so that B goes to A. Both senses of thereaction could
independently be described as prograde.Thus, the best that one can
say is that prograde meta-morphism involves a movement more or less
directly
-
Fig. 2. Generalized map of Acadian and younger metamor-phic
zones in west-central Massachusetts and adjacent states.Modified
from Schumacher et al. (1989). Zones I to VI afterTracy et al.
(1976). Diamonds indicate locations of sillimanitepseudomorphs
after andalusite; closed circles indicate orthopy-roxene in
metamorphosed mafic and intermediate igneous rocks;open circles
indicate locations of relict granulite-facies assem-blages in
Pelham gneiss dome; diagonal shading indicates zonesof retrograde
metamorphism at New Salem and Quabbin Hill.The area of zone I
between the retrograde zones and the Meso-zoic Connecticut Valley
fault appears to be an area of intenselate Paleozoic metamorphic
and tectonic overprinting.
away from lhe P-T origin in a way that may or may notinvolve
significantly increasing pressure.
Zoning produced during prograde garnet growth
In the lowest grade zones in central New England, thegarnets are
commonly well formed, are nearly euhedral,and show a characteristic
zoning pattern with Mn-richcores zoning to less Mn-rich rims with
first an increasingMg/Fe ratio and then a decreasing ratio (Figs. 6
and 7).With some reservations, these may possibly be taken
asexamples of garnet gpowth under the simplest conditions,where
diffusion in garnet is minimal to nonexistent. Onlythe surface of
the garnet is in equilibrium with the matrix.
I 783
Fig. 3. The number 8 formed by spherically arranged graph-ite
platelets enclosed in orthoclase of a pegmatite in Zone VI,central
Massachusetts. For location see Robinson et al. (1986),stop 7.
A sequence of garnet compositions can be studied thatrecords
garnet growth. Unless inclusions are preserved,only garnet rim and
matrix compositions can be used forgeothermometry. Inclusions may
reequilibrate withingarnet, but this requires diffusion. Note that
garnet growthneed not be prograde but may be retrograde,
depending
Fig. 4. The number 2 formed by scattered garnet in
silli-manite-orthoclase-garnet-cordierite-biotite schist from Zone
Yl,central Massachusetts. For location see Robinson et al.
(1986),stop I 1.
ROBINSON: EYE OF THE PETROGRAPHER. MIND OF THE PETROLOGIST
I \
At t(({ \\ \ l ' \ \rrt( t-ra-
\
z
ozoE.6
-
/
1784
Fe 'lo zo 30 Mg -
oo
Fig. 5. Summary plot of compositional trajectories of
zonedgarnets from central Massachusetts in terms of Fe, Mn, and
Mg.Numerals I-VI refer to metamorphic zones in central
Massa-chusetts, and P refers to partially resorbed granulite-facies
garnetfrom the Pelham dome. From Robinson et al. (1989).
critically on the P-7path and the reaction isopleths.
Thisconcept of growth, under either a progade net-transferreaction
or a retrograde net-transfer reaction, is illustrat-ed in Figure 8.
In aluminous rocks, Mn is most stronglypartitioned into garnet as
compared with any other phase.For this reason, it is the element
that appears most abun-dantly in early formed garnet and is also
the last to bedepleted from garnet in any garnet resorption
reaction in
ROBINSON: EYE OF THE PETROGRAPHER, MIND OF THE PETROLOGIST
Mg+
Fig. 7. Compositional trajectories oftwo typical zoned gar-nets
from zone I, central Massachusetts in terms of Fe, Mn, andMg. Core
compositions are at most Mn-rich points next to spec-imen number.
Specimen 908 is staurolite-garnet-biotite-mus-covite schist;
specimen 4F5 is similar but also with kyanite.From Tracy et al.
(1976).
which the garnet reequilibrates. (It is interesting that insome
skarns containing mostly andradite garnet, how-ever, Mn is not
partitioned into garnet but into pyrox-ene.)
What are the means by which examples of progradegrowth zoning in
garnet are destroyed or altered to ob-scure their story? First of
all, prograde or retrograde dif-fusional ion exchange between
garnet and its matrix (Fig.9), here taken to be biotite, could
destroy the near-rimrecord. A second possibility is a retrograde
(or possiblyprograde) diffusional net-transfer reaction (Fig. l0)
bywhich the outside of the garnet is partially resorbed whilethose
components retained by garnet actively diffuse to-
Fig. 6. Compositional maps of a zoned garnet in sample 908from
staurolite-garnet-biotite-muscovite schist of zone I,
centralMassachusetts. Contours based on Fel(Fe + Mg + Mn +
Ca),etc., are Fe 79-880/0, Mg 7-llo/o, Ca 2-5o/o, and Mn
l-90l0.Adapted from Tracy et al. (1976).
PROGRADE. NET TRANSFER,GARNET GBOWTH REACTION
Fe
BETROGRADE, NET TRANSFER.GABNET GROWTH REACTION
Fe
Fig. 8. AFM muscovite projections and Fe-Mn-Mg plotsshowing the
nature of prograde and retrograde garnet-growthreactions. In each
example, reaction proceeds from the solid tothe dashed tie lines,
and in each example there is a net increasein the modal amount of
garnet. The prograde reaction involvesdehydration but may not
involve significant temperature in-crease, as shown by the nearly
parallel tie lines in the AFM view.The retrograde reaction does
involve a significant temperaturedrop, as shown by the crossing tie
lines in the AFM view. Mod-ified from Roll (1987).
Mg
-
ROBINSON: EYE OF THE PETROGRAPHER. MIND OF THE PETROLOGIST 1
785
t
t3t2t 1t6
Gamet
Gamett *- t ;
t 1
ts
DistanceFig. 10. Hypothetical composition profiles ofgarnet and
bi-
otite for a retrogtade net-transfer garnet resorption reaction
fortime lo through r-. Diffirsion through garnet is a strong
functionof temperature, whereas biotite is assumed to remain
homoge-neous throughout the period ofgarnet equilibration. Because
ofresorption, the garnet interface retreats in favor ofbiotite.
Mod-ified from Spear (1991).
10'C/]VIa
%w725 (initial)
At6
! r
Distance --+- Biotite
- 1 2
+ L ?
Fig. 9. Hypothetical composition profiles ofgarnet and bio-tite
for retrograde ion exchange for time lo through l-. Diffusionin
garnet is a strong function of temperature, whereas biotite
isassumed to homogenize throughout the period ofequilibrationwith
garnet. From Spear (1991).
ward the interior. A third possibility, recently exploredby
Spear and Florence (1991), is internal diffusive mod-ification of
the primary zoning profile, which would pro-duce a new and
obviously false record oforiginal garnetgrowth compositions. In a
thermal model (Fig. I l) ofdiffusional ion exchange between garnet
and matrix bio-tite done by Spear and Florence (1991), using
diffusiondata from Cygan and Lasaga (1985) and assuming a
qua-si-infinite reservoir of biotite host, we can see the limitsof
diffusional reequilibration at a cooling rate of l0 "Cper m.y.
beginning al.725 t and tailing down to a limitclose to 500 'C.
Karabinos (1983) has described garnet-zone pelites in
1.00
.96
-
l 786
Fig. 12. Textural map of garnet from eastern Vermont,showing two
cycles ofprograde growth. From Karabinos (1983).Line C-R is the
compositional profile shown in Figure 13, andTU is the point
oftextural unconformity supposed by Karabinosto be the outer limit
of the original g:rnet at the end of the stageofretrogression. I
argue that resorption proceeded slightly deeperinto the
crystal.
Hollocher (1981, 1987) has describedjust such an ex-ample of
surface dissolution of garnet to produce chlorite(Figs. l5 and 16)
in the New Salem retrograde zone, cen-tral Massachusetts (see Fig.
2). In this case there are de-tailed analyses of the full spectrum
of relict primary zon-
ROBINSON: EYE OF THE PETROGRAPHER, MIND OF THE PETROLOGIST
RzR3
l m m
R1
end of lstp.rogrodestoge
end ofretrogression
T
MnF"
c
[.,.-...T] oo,n.t
MnFe
o
E.ooE,
IR
. lo
, .o2o . l 2
i .o,M g
F e
2mm ITU
Dis tonceFig. 13. Composition profile C-R in the garnet shown in
Fig-
ure 12, modified from Karabinos (1983). Karabinos suggests
thatthe zone with high Mn/Fe ratio was produced during
retrogradegarnet resorption; I believe it was produced early in the
secondstage ofgarnet growth.
-G-_.t
\!- -a r
-a -
-A.AA'41- -o--o--q--o--a--a- ' -
t t l
I
cl - l orortr , minor mico
ffi cntorire, minor muscovife
lr.-Tf, stourotite E ur..ouite, minor biofireFig. 15. Variably
resorbed garnets from the New Salem ret-
rogradezone studied by Hollocher (1987). Numbers indicate
thevariable spessartine content obviously left over from the
periodof initial prograde garnet growth.
c
end of 2ndp.rogrodesrqge
Fig. 14. History ofgarnet growth, resorption, and growth forthe
sample illustrated in Figures 12 and 13, as portrayed byKarabinos
(1983).
( R 2 ) U O | A(R3) 64-438
-
l l T t t t l +Somole NS 7 |Ret rogroded Gornet
Fig. 16. Drawing of partially resorbed garnet from New Sa-lem
retrograde zone by Hollocher (1981). Much of garnet is re-placed by
chlorite, but where it remains it retains the spessartinecontents
from the much earlier stage ofprograde garnet $owth.
ing in garnets that are partially dissolved in a
regionalhydration reaction. This demonstrates that the volumeof
garnet that was in equilibrium with the matrix duringthe retrograde
reaction was zero, but the surface of thegarnet was being actively
corroded to produce new matrixminerals.
In a more complex situation, garnet is growing, butdiffusion is
causing reequilibration within garnet and pro-gressively destroying
the earlier tape recorder message.
1787
Fig. 18. Compositional trajectories oftwo typical zoned gar-nets
from zones IV and V, central Massachusetts, in terms ofFe, Mn, and
Mg. Core compositions are at most Mg-rich pointsnext to specimen
number. Both specimens are sillimanite-ortho-clase-garnet-biotite
schists in which garnet has been resorbed bythe reaction garnet +
orthoclase + HrO : sillimanite + biotite.From Tracy et al.
(1976).
Spear (1991) has recently explored this situation in detailand
shown that the last grown part ofthe prograde zonedgarnet, which is
likely to record the highest temperature,is also the first to be
destroyed by retrograde ion exchangeor resorption reactions.
Zoning produced by retrograde garnet resorption
In regions of higher grade metamorphism, the recordof earlier
history, if there was one, was destroyed by dif-fusion, and the
garnets are largely unzoned. What hap-pens beyond the peak of
metamorphism depends criti-cally on environmental conditions and
whether garnetresorption reactions take place. The garnet
illustrated inFigure 17 is an irregularly shaped one from the
silliman-ite-orthoclase-muscovite zone in central Massachusetts.It
appears to have a continuous exterior rim of Mn andFe enrichment
and Mg depletion wherever the garnet istouching the matrix and
regardless of what phase is di-rectly adjacent in the matrix. This
is interpreted to meanthat the exterior part of the garnet was
equilibrated witha pervasive phase, probably an aqueous fluid, that
was in
ROBINSON: EYE OF THE PETROGRAPHER, MIND OF THE PETROLOGIST
Mg+
g l m m
Fig. 17. Composition maps of a garnet in sample 9338 from the
sillimanite-orthoclase-muscovite zone (Zone IV), central
Massachusetts. Contours based on Fe/(Fe + Mg + Mn + Ca), etc.,
areFeT3-77Vo; Mg l2-180lo; Mn 5-80/0. Continuous rims of
higher Fe, lower Mg, and higher Mn around this irregular garnet
are characteristic ofthis zone and indicate a continuous
retrogfade,
net-transfer, hydration resorption reaction through the medium
ofan interstitial fluid. Modified from Tracy et al. (1976)'
-
l 788 ROBINSON: EYE OF THE PETROGRAPHER, MIND OF THE
PETROLOGIST
Fig. 19. A series of model sections at constant T, P,
andhumidity showing equilibrium compositions of coexisting
garnetand biotite in an Fe-Mn-Mg ternary. From Robilson et al.
(1982).
easy communication with other matrix ferromagnesianminerals. The
fact that the Mn-content increases outward(Fig. l8) suggests the
action ofa net-transfer, fractional,garnet-resorption, retrograde
hydration reaction, as illus-trated in Figure 10, such as GAR + KSP
+ H,O : SILL+ BIO. Commonly the amount of fluid available for sucha
reaction would be limited and, because of limited dif-fusion, would
involve only a limited volume of the gar-net, while involving
virtually all of the matrix. The re-action is fractional because
the garnet interior isprogressively removed from participation in
the retro-grade reaction, so that a decreasing volume ofthe
exterioris subjected to more and more extreme reactions.
Theeffective volume of reequilibration of garnet continues tofall,
until it approaches zero and the garnet ceases to be
Fig. 21. Fractional garnet resorption illustrated schematical-ly
as a series of steps. At stage I bulk composition I consists
ofgarnet Gl and biotite Bl. The inner part of garnet Gl ceases
toequilibrate with matrix biotite and the effective bulk
composi-tion moves to 2, where it now consists of garnet G2 and
biotite82. Still more of the garnet ceases to equilibrate with
matrix,moving the bulk composition to 3, where it consists of
garnetG3 and biotite B3. The process continues until the efective
bulkcomposition contains no garnet, i.e., that difrrsion in garnet
isso slow that the mineral is incapable of reacting with
biotite.From Robinson et al. (1982).
in equilibrium with the matrix minerals, although, aspointed out
above, its surfaces may still be subject tosurface corrosion that
could contribute to the bulk chem-istry of matrix assemblages.
In 1982 we modelled this process in a very simple wayin an
Mn-Fe-Mg ternary (Robinson et al., 1982) that isnow being
duplicated more elegantly in computer pro-grams by Frank Spear and
his students. The first step wasto select a series ofequilibrium
sections (Fig. 19) cone-sponding to a series of changing
metamorphic conditionsof temperature, pressure, and humidity. The
figure showsputative garnet-biotite tie lines at a series of
temperaturesand a series of progressively more humid conditions
withfalling temperature. A similar series of templates could
600"c6250C
650"C
675"C700"c
Fig. 20. Model equilibrium garnet composition Fe-Mn-Mg
trajectories for two different initial garnet-biotite ratios
duringprogress ofthe hydration resorption reaction garnet +
orthoclase + H2O : sillimanite + biotite. In the upper diagram the
hydrationtakes place with falling temperature from 700 to 600 "C;
in the lower diagram at constant temperature of 625 'C. In each
case theequilibrium garnet composition trajectory is longer and
more curved for the more biotite-rich bulk composirion. Frorn
Robinsonet al. (1982).
-
ROBINSON: EYE OF THE PETROGRAPHER, MIND OF THE PETROLOGIST 1
789
Fig. 22. Comparisons of model equilibrium composition
trajectories and fractional crystallization trajectories during
retrogadegarnet resorption for two paths of cooling and hydration
starting with the same initial bulk composition. Note that the
fractionalpaths are more strongly curved than the equilibrium ones.
The actual zoning path for the garnet in 9338 (Fig. 18) is shown
for
comparison. From Robinson et al. (1982).
be constructed for retrograde hydration at constant
tem-perature. Using such templates, a series of
equilibriumgarnet-biotite tie lines (Fig. 20) has been constructed
fora series of changing conditions for two different
bulkcompositions, both originally on the same gamet-biotitetie line
at the peak of metamorphism. One can see thatthe equilibrium paths
of garnet for the two bulk compo-sitions are quite different and
that garnet compositionchanges more for bulk compositions that are
poorer ingarnet. This simple geometrical concept was well
under-stood by Ferry and Spear (1978) when they designed
theirgeothermometric experiments. In order to avoid disequi-librium
problems with zoned garnet, they designed theirexperiments with
large ratios of garnet to biotite so thatthere would be little
tendency for composition change ingarnet and they could study
composition changes in bi-otite, which reached equilibrium easily.
It should be em-phasized that the equilibrium paths illustrated in
this fig-ure are only indirectly related to the zoning in
naturalresorbed garnet, which is produced by a fractional
dis-equilibrium process.
The fractional aspect is illustrated in Figure 2l in aseries
ofsteps involving alternate equilibration and frac-tionation, in
which progressively less and less garnet par-ticipates, thus
decreasing the garnet to biotite ratio. Asthe fraction
ofparticipating garnet decreases, the changein composition of the
remaining portion becomes moreand more extreme, while changes in
biotite compositiondecrease dramatically. For this reason the
fractional crys-tallization paths (Fig. 22) arc more strongly
curved thanthe equilibrium ones, but the exact details depend
criti-cally on the extent of fractionation. Obviously, this
type
ofreaction ceases when the diffttsion is so slow that
garnetcannot participate at all.
A serious consequence ofthis type ofreaction, in caseswhere
garnet and biotite start out in roughly equal pro-portions, is that
none of the original biotite composition
I
RETROGRADECONTINUOUS
Fig. 23. Consequences for geothermometry of a
diffirsionalretrograde continuous net-transfer garnet-resorption
hydrationreaction, here wiewed in an orthoclase AFM projection. As
theresorption reaction proceeds, equilibrium garnet and
biotitecompositions both move toward Fe enrichment. The
originalgarnet composition Gl is protected from reaction in the
interior,whereas all or nearly all of the original biotite Bl in
the matrixis destroyed, thus removing the possibility of directly
determin-ing the original peak metamorphic temperature. Garnet G2
andmatrix biotite 82 may be used to estimate the temoerature
underwhich the retrograde reaction ceased. When garnet Gl is
pairedwith biotite 82, with which it was never in equilibrium, a
com-pletely false temperature estimate is made, which can
normallybe much higher, even hundreds ofdegrees, than the original
peak
temperature. From Robinson et al. (1982).
i l
Fe
-
1790
1.00
.96
ROBINSON: EYE OF THE PETROGRAPHER. MIND OF THE PETROLOGIST
bo
T
( r .
v .88()
.84
^ 800o
E 7oo
F ooo
.80200 400 600
Radius (pm)
900
\.
;;;i;;;^;/
500 L0 2N 400 600 800 1000
Radius (pm)Fig.24. Profiles of garnet composition and calculated
garnet-
biotite temperature at temperatures from 725 to 500 qC. In
theparticular model produced, the bulk garnet/biotite ratio
begannear l: l, and a diflirsional retrograde net-transfer
garnet-resorp-tion reaction with a major biotite reservoir was
assumed to takeplace between 725 and 675'C. This was followed by
pure ionexchange from 675 and 500 qC. Note that although the
interiorofthe garnet is hardly changed in composition,
Z(calculated) forthe core rises dramatically from 725 to 900'C as a
result of amajor increase in Fe/(Fe + Mg) of matrix biotite as the
garnetrim is resorbed. This is followed by a decline from 900 to
875"C as the biotite decreases in Fe/(Fe + Mg) during ion
exchange.From Spear and Florence (1991).
remains that was originally in equilibrium with the garnetcore,
unless a $ain remained in some unusual protectedlocation. Thus,
when an unaltered garnet core composi-tion, such as Gl in Figure
23, is paired to make a tem-perature estimate with a highly altered
matrix composi-tion such as B2, the result is always higher than
the originalpeak temperature. In some cases the result may be
hun-dreds of degrees higher, as pointed out by Robinson etal.
(1975) and illustrated in a recent calculated profile
Gie.2q by Spear and Florence (1991).The style of reaction just
discussed is illustrated in a
spectacular way by zoned garnets studied in detail in thePelham
gneiss dome, Massachusetts (Fig. 25), by Roll(1987). These rocks
occur in the Mount Mineral For-mation within the area of Zone I
(kvanite-muscovite-
Fig. 25. Map showing the Pelham dome relative to the re-gional
geology and metamorphic zones of west-central Massa-chusetts.
Narrow belt shown in black in the southern part ofthedome is the
Mount Mineral Formation containing relict granu-lite-facies
assemblages. Adapted from Roll (1987).
staurolite), and they retain relics of an older granulite-facies
metamorphism shown by relict high orthoclase withsillimanite
inclusions, garnet cores with pyrope contentup to 350/0, coarse
sillimanite, and relict associations of
Fig. 26. Schist sample Q-P25c from the lower part of theMount
Mineral Formation, east limb of Pelham dome, centralMassachusetts.
Contains a sheared relic ofhigh orthoclase about2.5 cm long with
tails of secondary plagioclase and muscoviteset in a matrix of
schist rich in muscovite, biotite, kyanite, andgarnet.
800 1000
IN H I
I
o
I P E L H A MD O M E
-
Fig. 27. Photomicrograph of schisr sample W67B from theupper
part of the Mount Mineral Formation, east limb of Pel-ham dome. To
left is the margin of a high orthoclase megacrystenclosing quartz,
plagioclase, and sillimanite (stout needles farleft). The
immediately surrounding matrix (right) contains mus-covite,
biotite, and coarse kyanite (large grain close to orthoclaseright
center).
garnet with coarse rutile. These occur in a variety of rockswith
a mylonitic character related to late shearing in thedome, and they
commonly occur side by side with rocksthat have been so completely
reconstituted that they showfew or no traces of the older
metamorphism and onlyfeatures of late mineral growth.
The sample in Figure 26 contains a sheared relic ofhigh
orthoclase with tails of secondary plagioclase andmuscovite set in
a matrix of schist rich in muscovite,biotite, kyanite, and garnet.
Figure 27 shows the marginof a high orthoclase megacryst enclosing
sillimanite. Thematrix in the immediate area contains muscovite.
biotite.and coarse kyanite. In Figure 28, large garnet
megacrystssurrounded by orthoclase and sillimanite are enclosedby
thick rims of biotite produced by retrograde hydrationreactions.
One of the samples studied in detail was a my-
Fig. 28. Schist sample l60C originally collected by L.D.
Ash-walin 1972 from the middle of the Mount Mineral Formation,east
limb of the Pelham dome. Garnet megacrysts up to 1.5 cmlong,
surrounded by orthoclase and sillimanite, are envelopedby thick
rims ofbiotite produced by retrograde hydration reac-tions.
t79r
@Fig. 29. Garnet composition maps from sample l60X (see
previous caption) with analytical points indicated by dots.
Topmap shows contours of pyrope content ranging from 34olo downto
l2o/o at 2olo intervals. Bottorn map shows contours of spessar-tine
content of 2o/o and 30/0. Vertical ruling indicates areas ofbrown
biotite; dashed horizontal ruling indicates areas of palegreen
biotite. Adapted from Roll (1987).
lonitic garnet gneiss, sample 160X, from which the
garnetcomposition map in Figure 29 and the composition tra-jectory
diagram in Figure 30 were produced. These gar-nets show extreme
zoning in pyrope content and Fe/Mgratio and a slight marginal
increase in Mn. The pyropecomposition map shows that the large
garnet was beingresorbed along a series ofcracks that look like
fiords onsome island off the Norwegian Coast. The relict cores
insample l60X contain 350/o pyrope and l.l0lo spessartine.From the
core to the rim, the pyrope content decreasescontinuously to l0-l
lo/o and spessartine increasessmoothly to a maximum of about 3ol0.
A second garnetfrom a similar rock (Fig. 3l) is a more regular
egg-shapedporphyroclast showing an irregular pyrope-rich
interiorand a thin pyrope-poor, spessartine-enriched rim and ismore
amenable to diffirsional modelling.
Roll modeled the time required to produce such re-sorption rims
at 580 lC, 5 kbar using three different setsof difusion coemcients
(Fig. 32). Under such conditionsit would take 40 m.y. for
diffirsional penetration of thegarnet to a depth of 0.05 cm, rather
longer than appearedto be available during Acadian metamorphism.
Based onthese results, she suggested that the retrograde
resorptionzoning might have been more easily produced in two
ep-isodes, one during cooling at the end ofan early, probably
ROBINSON: EYE OF THE PETROGRAPHER, MIND OF THE PETROLOGIST
-
t792
Fe 10 20 30 Mg_
40
Fig. 30. Compositional trajectories ofzoned garnets from
theMount Mineral Formation. Pelham dome. central Massachu-setts, in
terms of Fe, Mn, and Mg. C indicates core, R indicatesrim.
Composition maps for these garnets are given in Figure 29(160X),
Figure 31 (W67B), and Figure 33 (X43A). All comefrom
porphyroclastic schists with other relics of original
neargranulite-facies conditions, in which garnet has been
resorbedby the reaction garnet + orthoclase + HrO : alurninum
silicate+ biotite. From Roll (1987).
Proterozoic, high-grade metamorphism and a secondduring Acadian
retrograde hydration. These views re-quire reinterpretation based
on new detailed isotopicstudies by R. D. Tucker (Tucker et al.,
1988; Tucker andRobinson, l99l; Robinson and Tucker, l99l). Zircon
andmonzvite from a sillimanite-orthoclase pegmatite withinthe
schist in question yielded nearly concordant U-Pbages of 357 and
367 Ma, respectively, and monazite fromthe schist gave 367 Ma,
precluding an early metamor-
ROBINSON: EYE OF THE PETROGRAPHER, MIND OF THE PETROLOGIST
a O8Oo
I
l i o7sE
580oC, 5 kbar
for 40 Ma
o o o o 02 0,o4 005 0 0a o 10
Distonce (cm)Fig.32. Model penetration distances produced in
hornoge-
neous garnet 2 mm in diameter of 350/o pyrope composition
byretrograde reequilibration at 580 'C, 5 kbar for 40 m.y.
Gradi-ents are based on three sets ofdiffusion coefrcients: F:
Freer(1979);C: Cygan and Lasaga (1985); E : Elphick et al.
(1985).Under such conditions it would take 40 m.y. for diffusional
pen-etration of the garnet to a depth of 0.05 cm. The model does
nottake into account the movement ofthe garnet interface producedby
resorption. From Roll (1987).
phism much before 360 Ma. The quartzite at the top ofthe Mount
Mineral Formation contains virtually concor-dant single detrital
zircons of a range of ages includinggrains with ages of 458, 440,
and 439 Ma, showing thatthe quartzite itselfcould be no older than
Early Silurian.The same quartzite contains the extremely
retrograde-zoned, Mn-rich garnet described below and has
yielded
F
cE
Fig. 31. Composition maps of egg-shaped porphyroclastic garnet
from sample W67B of the Mount Mineral Formation (see Fig.27).
Contours based on Me/(Fe + Mg + Mn + Ca), etc., are Py 16-220/0 at
20lo intervals, Sp 4-690 at l0lo intervals,andGr 4-5o/oat 1olo
intervals. Maps show rimward decrease in pyrope and increase in
spessartine consistent with resorption ofgarnet by reactionwith
matrix through the medium of interstitial fluid. Adapted from Roll
(1987).
-
Fig. 33. Composition maps of small fragmented garnet fromsample
X43A from the Mount Mineral Formation on the south-west limb of the
Pelham dome. Contours based on Mg/(Fe +Mg + Mn + Ca), etc., arePy
l4-34o/o at 20lo intervals, Sp l-40loat l0lo intervals,Gr l-4o/o at
lYo intervals. Although only I x 3mm in size, it contains almost
the same range of pyrope com-positions as the much larger gamet in
Figure 29. Probably thegarnet is a sheared fragment from the
interior of a once largergarnet that was only tectonically exposed
to the matnx over arelatively short period near the end of the late
reequilibration.Adapted from Roll (1987).
a recrystallization age of metamorphic monazite of 297Ma,
consistent with metamorphic sphene ages (Tuckerand Robinson, 1990),
and Rb-Sr and Nd-Sm mineral iso-chrons (Gromet and Robinson, 1990)
nearby in the dome.Thus, it appears that the relict
granulite-facies metamor-phism was Acadian and the
kyanite-muscovite overprintwas late Paleozoic, not Acadian. The
boundary betweenthis late overprint and true Acadian isograds
farther eastis being actively investigated.
In sample X43 (Fig. 33) almost the same range of py-rope
compositions was obtained (Fig. 30) in a grain I x3 mm in size. In
this case it seems hardlv likely that such
Fe lO 20 30 40
M q +
Fig.34. Summary diagram of the zoning trajectories for gar-nets
in seven specimens of the Mount Mineral Formation stud-ied by Roll
(1987). X43A, 160X, and W67B are from the schistsdescribed above.
M22A is from a sillimanite-orthoclase peg-matite. The Fe-Mn-rich
garnet is from a nearby garnet quartziteshown with contours for
spessartine and grossular in Figure 35.There is little to no zoning
in pyrope content, and the trajectoryis consistent with a
resorption reaction involving formation ofFe-rich biotite.
Specimens l60M and, especially, Y33 belong torocks that appear to
have been totally reconstituted, with garnetnewly grown during the
late hydrous metamorphic overprint, asdiscussed in the text.
t793
Fig. 35. Composition maps of egg-shaped porphyroclasticgarnet
from quartzite sample M2l near the upper contact of theMount
Mineral Formation on the east limb of the Pelham dome.Contours at
intervals of l9o based on Mn/(Fe + Mg + Mn +Ca), etc., Sp 3-140/0,
Gr 6-80/0. Maps show rimward increase inspessartine consistent with
resorption ofgarnet by reaction withthe matrix biotite through the
medium of interstitial fluid. De-pression hachures are omitted from
closely spaced contours nearedge ofgarnet. Adapted from Roll
(1987).
a steep gradient could have been preserved through evena single
retrograde event at 580 'C, and more probablythe garnet is a
sheared fragment from the interior of aonce larger garnet that was
only tectonically exposed tothe matrix over a relatively short
period near the end ofthe late reequilibration.
Figure 34 summarizes the zoning trajectories fot gar-nets in
seven diferent specimens in the Mount MineralFormation studied by
Roll. Three are from the schistsalready described, and a fourth is
similar in form but isfrom a sillimanite-orthoclase pegmatite. The
Fe-Mn richgarnet, from the quartzite at the top of the formation,
isshown in maps in Figure 35, with contours for grossularand
spessartine. In this specimen there is little to no zon-ing in
pyrope content and the trajectory is consistent witha resorption
reaction involving formation of Fe-rich bi-otite.
In Figure 36 biotite compositions from all seven rocksare
plotted in terms of Ti per I I O atoms and the ratioMg/(Mg+Fe). Ti
content for a given Mg ratio has beenshown to be a function of
metamorphic grade in rockssaturated with ilmenite or rutile as well
as aluminum sil-icate (Robinson et al., 1982; Schumacher et al.,
1990). Inthese rocks variable Ti content seems to be related in
part
to extent of participation in retrograde resorption reac-tions
and proximity to grains of Ti oxide minerals. Theextreme example is
biotite from deep within the "fiords"of the garnet l60X in Figure
29, which are colorless topale green in thin section and contain
less than 0.025 Tiper I I O atoms. Because these fiord biotite
grains wereclearly produced by the reaction of a K-bearing
matrixsolution upon the enclosing walls of garnet, it appears
ROBINSON: EYE OF THE PETROGRAPHER, MIND OF THE PETROLOGIST
-
@@ wet
wa
% to a o
s v ' o . ovrr,V_l 16ox
16oM y- v a
A
Ol O,2 O.3 04 05 S" OZ OB 09
25
20
15
10
o5
BIOTITE
aM 2 1
a
aoO a
$oa
o
o
t794
Mg/(Mg+Fe)Fig. 36. Biotite compositions from all seven rocks
plotted in
terms of Ti per I I O atoms and the ratio Mg/(Mg + Fe).
Samplesshown are W67 (squares), X43 (circles), l60X (rriangles),
y33and 160M (inverted triangles), M22 (small diamonds), and
M2l(large diamonds). Inclusion biotite (closed symbols) is
moremagnesian than corresponding rim (open symbols) or matrix(ruled
symbols) biotite. Dotted triangles indicate biotite in l60Xthat
fill cracks in garnet. From Roll (1987).
+ M
ROBINSON: EYE OF THE PETROGRAPHER, MIND OF THE PETROLOGIST
Ti
1 lOx'
\ \ \ ,continuous
+ hydrationion"id"ng"
Fig. 37. Schematic AFM projection, showing the
contrastingeffects on biotite compositions ofcontinuous retrograde
hydra-tion on matrix biotite and of retrograde ion exchange on
inclu-sion biotite. As hydration proceeds, equilibrium tie lines
be-tween garnet rims and matrix biotite move to the left,
whereasinclusion biotite moves to the right. The original peak
biotitecomposition is no longer preserved but presumably had a
com-position intermediate between matrix biotite and biotite
includ-ed in garnet cores. From a diagram prepared by M.A. Roll
(per-sonal communication, I 986).
Fig. 38. Composition maps of euhedral garnet in sample Y33from
schist near the upper contact of the Mount Mineral For-mation on
the southwest flank of the Pelham dome. This schistis believed to
have been completely reconstituted during the latemetamorphic
reequilibration. Maps show rimward decrease rnpyrope, spessartine,
and grossular content consistenl with ret-rograde garnet growth.
Adapted from Roll (1987).
that under the ambient retrograde conditions estimatedat about
550'C, the matrix solution had not dissolved ordid not contain
enough Ti to produce a biotite with anormal Ti content typically
found in matrix locations inclose physical contact with Ti-oxide
grains.
In three of the samples, M2l, W67, and X43, com-positions were
obtained both on matrix biotite and onbiotite included within
garnets. Inclusion biotites (Fig.36) are consistently more
magnesian than matrix biotites,in agreement with the two different
processes by which
Fe 10 zo Mo*
ao
Fig. 39. Comparisons of composition trajectories of garnetin
W67B and in Y33. In each example, garnet core has the mostMg-rich
composition. W67B is consistent with a retrograde gar-net
resorption reaction, whereas Y33 is consistent with retro-grade
growth. Adapted from Roll (1987).
-
Fig. 40. P-T diagram with garnet and biotite
compositionisopleths for the reaction garnet + muscovite : biotite
+ alu-minum silicate. Note that isopleths change orientation
whencrossing aluminum silicate field boundaries and when
crossingthe muscovite-out reaction in lower right. Quadrants of
garnetproduction (P) and garnet consumption (C) are indicated for
the
t795
kyanite and sillimanite zones. Heavy lines with arrows
representthree retrograde P-T trajectories illustrated in
quasi-binary sec-tions in Figure 41: one parallel to garnet
isopleths (Fig. alA),one parallel to biotite isopleths (Fig. 1O,
and one intermediate(Fig. 41B). Adapted from Spear and Selverstone
(1983).
ROBINSON: EYE OF THE PETROGRAPHER, MIND OF THE PETROLOGIST
GAR+MUS=BlO+Als i l
ltI
a'
/ - - - z -
=ffi7='j=- ---
1 - - t - - t - '
r, oc Blo + Alsil = GAR + KSP + H2o
XFE GAR XpE BIO
their compositions were achieved. The matrix biotiteformed by
continuous retrograde net-transfer garnet re-sorption in which
biotite grew in amount and becameprogressively more Fe enriched.
The inclusion biotite didnot change in amount but became Mg
enriched as a resultoflocal ion exchange with the enclosing garnet,
probablyproducing a small Mg-impoverished compositional wellin the
garnet, such as that described by Spear et al. (l 990).
It is apparent that if Fe-enriched matrix biotite ismatched with
the relict core garnet composition withwhich it was never in
equilibrium, an artificially high-temperature estimate will be
obtained (see also Fig.2q.In the case of sample l60X such a
procedure gives anestimated temperature of 1000 oC. By contrast,
matchingof inclusion biotites with immediately surrounding gar-nets
yields estimates of temperature of final ion exchange
equilibration of about 450 qC. A major problem is tolocate and
analyze, or to estimate, the composition ofbiotite that was
originally present in the rock at the peakof granulite-facies
conditions. Spear et al. (1990) man-aged to locate biotite
inclusions within sillimanite thathad apparently experienced little
of the retrograde re-equilibration described here. In the Pelham
dome rocks,the best estimate of original biotite composition was
madeby choosing one intermediate between matrix and inclu-sion
extremes (Fig. 37) and also one that would yield anestimated
temperature and pressure consistent with thepresence of sillimanite
in the rock. On this basis, a biotitecomposition chosen for sample
l60X with Mg/(Mg+Fe): 0.66 yields conditions of equilibration with
sillimanite,garnet, and high orthoclase of 700 "C and 6.8 kbar
(Roll,I 987) .
-
1796
Zoning produced during retrograde garnet growth
The two garnet grains with short zoning trajectories inFigure 34
are from schists of the Mount Mineral For-mation that appear to
have been completely reconstitutedduring the second more hydrous
metamorphism. Thesegarnets are typically euhedral (Fie. 38) with
Mn- and Ca-enriched cores and might be supposed to have formed
byprograde Efowth zoning. However, unlike such garnets,they do not
show a typical increase in the Mg/(Mg + Fe)but rather a decrease,
resembling in this way zoning pro-files produced by retrograde
garnet-resorption reactions.Chemical trajectories of the two types
of garnet are com-pared in Figure 39. The garnet in sample W67B
showsthe typical outward increase in both Ca and Mn associ-ated
with diffusive rim resorption of the Mg componentinto other
ferromagnesian minerals. The garnet in sampleY33, on the other
hand, shows rimward decrease of Caand Mn that can only be
associated with garnet growthbecause there is no other mineral in
the rock that couldbe proportionally absorbing more Mn than the
garnet.
Peter J. Thompson (1985) has described interestinggarnets from
the Mount Monadnock area, New Hamp-shire, in which there is a
similar strong rimward decreasein Mg/(Mg * Fe) accompanied by a
similar decrease inMn but with a striking rimward increase in Ca.
In thisexample, it appears that the pyrope content ofthe garnetwas
undergoing progressive resorption with respect to thecordierite in
the rock, but at the same time the grossularcontent ofthe garnet
was undergoing progressive growthwith respect to the anorthite
component of plagioclasethat slightly overbalanced the resorption.
In this example,as in all of the others discussed here, the Mn
profile isthe most reliable predictor of growth vs. resorption
be-cause Mn is more strongly fractionated into aluminousgarnet than
any other phase.
ROBINSON: EYE OF THE PETROGRAPHER. MIND OF THE PETROLOGIST
Fig. 41. Binary representations in terms of ?"-P and XF" ofthe
phase relations along the three retrograde trajectories
illustratedin Figure 40. For details see text. Modified and
expanded from Roll (1987).
r .o o.8 0.6 0.4 0.8 0.6 0.4X F e X r e
o.8 0.6 0.4Xre
It is a common habit of metamorphic petrologists tothink of
garnet growth in terms of a prograde reactionand resorption as a
retrograde (i.e. decreasing tempera-ture) phenomenon. It is further
habitual to visualize suchreactions in terms of binary Z-X loops.
Study of com-position isopleths for a reaction on a P-T diagram
illus-trates the true complexity even in an oversimplified
qua-si-binary situation. For this purpose, Ihe Xr. isopleths
forgarnet and biotite derived by Spear and Selverstone (1983)for
the reaction garnet + muscovite : aluminum silicate+ biotite serve
as an admirable example, as shown inFigure 40. In the habitual
view, one may cross the iso-pleths by increasing pressure and
increasing temperatureslightly such that garnet is produced,
biotite is consumed,and progressively lower X." isopleths for both
mineralsare crossed. Similarly, one may cross isopleths by
de-creasing pressure and temperature such that garnet is re-sorbed,
biotite grows, and progressively higher Xr" iso-pleths for both
minerals are crossed. These two directionstoward garnet production
or consumption (P and C inFig. a0) He in only two of the possible
four quadrantsbetween the isopleths.
Relations within the remaining two quadrants are morecomplex. In
these quadrants the directions ofgarnet andbiotite composition
change are opposite, and for eachP-I trajectory, garnet production
or consumption willdepend critically on the garnet-biotite ratio in
the rock.The relations in these quadrants are illustrated by
PT-X-.sections from Figure 40 taken to follow three specific
tra-jectories. Figure 4lA shows a trajectory constrained alongthe
Garnet Xo. : 0.85 isopleth for the kyanite zone. ThisPZ-X section
is oriented with increasing T at the top,which is also the
direction of decreasing P. Within thekyanite zone, as specified,
garnet composition is constant,but along the same trajectory at
lower temperature in the
S i l l I r \ B i o
-
Fig.42. Pavement outcrop of typical migmatitic schist of
thesillimanite-orthoclase-garnet-cordierite zone, central
Massachu-setts. Large garnets and leucosomes are believed to be the
prod-uct offluid-absent partial melting. See Robinson et al.
(1986),stop 4.
garnet-chlorite zone and also at higher temperature in
thesillimanite zone, garnet is more Fe-rich. Taking a typicalbulk
composition along the dashed line and following itin Figure 4lA in
the kyanite zone only, it can be seenthat garnet is consumed with
increasing temperature (i.e.,the bulk composition on the
garnet-biotite tie line liesprogressively closer to the biotite
end) and is producedas temperature decreases.
Figure 4lC shows a trajectory constrained along thebiotite
isopleth Xr.: 0.45 for the kyanite zone. This sec-tion is also
oriented with increasing T at the top, but inthis case, this is
also the direction of slightly increasingP. Taking the same typical
bulk composition along the
Feo MsoFig. 43. AFM projection showing mineral compositions
in
typical assemblages of pelitic schist of Zone VI, central
Massa-chusetts. From Robinson et al. (1982).
t797
O.5mm
Ftg. 44. Composition maps of garnet in sample FW407
ofsillimanite-orthoclase-garnet-cordierite-biotite schist from
ZoneVI, central Massachusetts. Garnet shows significant Fe
enrich-ment and Mg depletion only where in direct contact with
adja-cent biotite, permitting retrograde ion exchange. From Tracy
etal. (1976). See Robinson et al. (1986), stop 6.
dashed line and foltowing it in Figure 4lC,it can be seenthat
garnet is produced in an up-temperature direction(i.e., the bulk
composition lies progressively closer to thegarnet ends of the tie
lines) and is consumed in a down-temperature direction.
Figure 4lB is a PT-X*. section for a trajectory betweenthe two
bounding examples in 4lA and 4lC. On thistrajectory, with
increasing temperature, garnet Xo. de-creases while biotite Xo"
increases, consistent with thedecreasing distribution coefficient
between these twophases. Note that for Fe-poor compositions (X),
garnet isconsumed with increasing temperature even though thegarnet
itself decreases in X.. until the bulk compositionconsists only of
biotite. Similarly for Fe-rich composi-tions (Y), garnet is
produced with increasing temperature.With falling temperature along
the same trajectory, gar-net is produced in Fe-poor compositions
(X) and is con-sumed in Fe-rich compositions (I').
The garnet-biotite two-phase region in a profile such asthat in
Figure 4lB is divided into regions of production
and consumption by a vertical dash-dot line, which ispositioned
differently for each trajectory. To the left ofthis boundary, as in
composition Y, garnet is producedwith increasing temperature and
consumed with decreas-ing temperature. To the right of this
boundary, as in com-position X, garnet is consumed with increasing
temper-ature and produced with decreasing temperature. By
ROBINSON: EYE OF THE PETROGRAPHER. MIND OF THE PETROLOGIST
B i o :
-
r 798
Fig. 45. Compositional trajectories of zoned garnets fromZone
Yl, central Massachusetts, in terms of Fe, Mn, and Mg.Typical
compositions are the most Mg rich, and zoning to moreFe-rich
compositions occurs only where garnet is in direct con-tact with
biotite or cordierite. From Tracv et al. (1976).
comparing Figures 4l B and 4lC, it will be seen that themore
nearly parallel the P-T trajectory is to the biotiteisopleths, the
larger the region for retrograde gamet con-sumption and the smaller
the region for retrograde pro-duction, until, in the extreme
example parallel to the bi-otite isopleth, there is no region for
retrograde production.
In the trajectory of Figure 4lB, the potential for ret-rograde
garnet production is proportional to the distancefrom the
production-consumption boundary. This is il-lustrated in two
calculated examples showing the changefrom 600 to 550 'C, in which
Xo. garnet changes from0.832 to 0.845. For a bulk composition Xo" :
0.593, themole percent garnet will increase from 30 to 33,
whereasfor a bulk composition XF, : 0.507, the mole percentgarnet
will increase from 5 to 10.5. In the case ofgarnetgrowth by
fractional crystallization, the interior of thegarnet will be
effectively locked away from the exteriorbulk assemblage so that
the effective bulk compositionwill become progressively more
biotite rich. This will drivethe effective bulk composition toward
lower Xo., furtherinto the region of retrograde production of
garnet withdecreasing temperature. If, in the above examples of
cool-ing from 600 to 550 oC, there is fractional
crystallizationduring retrograde $owth such that half of the garnet
istaken out of reaction at 575 "C (see dotted lines in Fig.4lB),
then the effective bulk composition will be even
Fig. 46. AFM orthoclase projection of garnet and
biotitecompositions, illustrating the relatively simple
interpretation re-quired in cases of retrograde ion exchange. From
Robinson etal. (1982).
ROBINSON: EYE OF THE PETROGRAPHER, MIND OF THE PETROLOGIST
Ir/g
Al
MgFe
Fig. 47 . Photograph of even-grained white quartzose marblewith
brightly reflecting elongate grains of white wollastonite.Specimen
was collected from a gas pipeline trench on the eastside of
Rattlesnake Mountain by Berry ( I 989), from the heart ofzone VI,
central Massachusetts.
more biotite rich and still more capable of growing gar-net: in
the two cases, from 30 to 33.7o/o or from 5 toll.7o/0.
In Figure 4lB, it will be seen that the best opportunityfor
retrograde growth zoning with a substantial garnetcomposition
change will be in a trajectory at a slight angleto biotite
isopleths where a small biotite-rich composi-tion region will be
subject to garnet growth. Fractionalcrystallization will continue
to maintain the effective bulkcomposition in this biotite-rich
region where garnet growthcan continue. Such a situation of
garnet-growth zoningwith falling temperature appears to be the
origin of thezoned garnets in completely reconstituted schists such
assample Y33 (Fig. 39).
Zoning produced during retrograde ion exchange
The pelites of the sillimanite-orthoclase-garnet-cordi-erite
zone (Zone VI) in central Massachusetts are typifiedby leucosomes
of quartz, feldspar, and cordierite com-monly containing garnet up
to 2 to 3 cm in diameter (Fig.42). Such textural features have been
described as theproduct of fluid-absent melting (Tracy and
Robinson,1983). Typical assemblages and mineral compositions
areshown in an orthoclase projection in Figure 43. Garnetis
typically about 5-100/o more magnesian than garnet inZones IV and
V, indicating considerable progress ofthestrong dehydration
reaction biotite + sillimanite : or-thoclase * garnet + HrO. Very
locally there is evidenceof late retrograde hydration reactions
between garnet orcordierite and orthoclase to produce sillimanite +
biotite,but this is not typical. The common four-phase assem-blage
in projection may indicate that HrO was behavingas an inert
component in the absence ofan aqueous fluid.Typically both larger
and smaller garnets are homoge-neous from core to rim except at
points where the garnetsare in direct contact with another
ferromagnesian min-eral, either cordierite or biotite (Fig. aq.
This style ofzoning was first described in this region by Hess
(1969,
.-------.s--4.4o7o-.?4.st.FWl54FWt22
-
pS76oroesoefilcARNEr- f_l wotlAsToNm- WV)PtAGtoctlsE+'Jy' GMINS
Lj'^J BAARNG L-I BEARING I2-4 CNJCfiE
Eu REGoN REGIoN REGoN
Fig. 48. Thin-section sketch of wollastonite marble,
showingsubdivision of the rock into regions. wollastonite-bearing
regionshave a veinlike character, separated by areas with
quartz-calcite.Plagioclase-calcite regions occur predominantly away
from di-opside, whereas close to diopside, plagioclase and calcite
areseparated by garnet. From Berry (1989).
l97l). In these locations, the Fe-enriched garnet rim
isaccompanied by zoned biotite or cordierite that is en-riched in
Mg as compared to biotite or cordierite that isin the matrix away
from the garnets. The local chemicalzoning trajectories (Fig. a5)
are virtually flat with respectto Mq strongly indicating that there
has been neithergarnet resorption nor garnet $owth. This type of
zoningis most easily explained as the product of local
retrogradeion exchange between garnet and surrounding
ferromag-nesian minerals where they are in direct contact and
with-out participation ofan interstitial fluid. It is a strong
ar-gument that at peak conditions pervasive fluids were notpresent
in most zone VI rocks. If no aqueous fluid waspresent, then we must
appeal to straight difusion throughthe matrix or localized ion
exchange at grain-to-graincontacts. The latter was first observed
and studied in cen-tral Massachusetts by P. C. Hess (1969, l97l),
and hisobservations have been repeated countless times.
The interpretation of these kinds of garnets in terms
ofgeothermometry is apparently much simpler. The ho-mogeneous
garnet composition can be compared with thematrix biotite
composition away from the garnet to es-timate peak conditions (Fig.
46). The garnet rim com-
1799
Fig. 49. Thin-section sketch showing garnet-bearing regionsclose
to diopside and preservation of anorthite + calcite awayfrom
diopside. From Berry (1989).
position and the adjacent biotite or cordierite composi-tion can
be used to estimate the temperature of lastretrograde ion exchange.
The only question might bewhether some other process intervened
between the peakmetamorphism and the retrograde ion exchange that
couldhave effectively changed the typical matrix biotite
com-position without also producing concentric zoning in thegarnet.
In my opinion this is unlikely.
Conclusion concerning garnet zoning
It seems fairly clear that at temperatures below about500 "C,
most garnet zoning can be interpreted in termsof simple growth
zoning. At higher temperatures, how-ever, all sorts
ofcontradictions are produced by diffusion,which commonly makes it
very difficult to select the cor-rect mineral compositions for
quantitative geothermom-etry and geobarometry. If all of the
conditions and con-siderations outlined here are correctly
accounted for, andone happened to be confronted with appropriate
rocks,then it may be possible to construct detailed P-7 pathssuch
as those ofSpear et al. (1990).
Pnlsn RELATToNS rN AWOLLASTONITE.BEARING MARBLE
In the next section of this paper, I want to show howdetailed
petrographic observations in a single thin sectionled to a new
conception for a petrogenetic grid. This con-ception comes from the
Ph.D. thesis of Henry Berry (1989,l99l) and concerns one aspect
ofhis interpretation ofthefirst documented occurrence of regional
metamorphicwollastonite from the granulite-facies in New
England,near Sturbridge, in the center of Zone Yl of Figure 2.This
section is based on a small highlight of a more com-prehensive
study that Berry is preparing for separate pub-lication.
Petrography
The rock is an even-grained, foliated, white quartzosemarble
with brightly reflecting elongate grains of white
ROBINSON: EYE OF THE PETROGRAPHER. MIND OF THE PETROLOGIST
o 5mm
-
r 800
Fig. 50. Photomicrograph showing abundant garnet coronasbetween
anorthite and calcite close to large diopside grain inlower right
and absence of such coronas between anorthite andcalcite in upper
left. Dark grains are sphene. Width of view, 3mm approximately.
Photograph provided by H.N. Berry.
wollastonite (Fig. a7). The rock consists mainly of
quartz,calcite, anorthite (An98), wollastonite, calcic
scapolite(-Me86), ferroan diopside [Fe/(Fe + Mg) :
0.40-0.45],grossular-rich garnet (Gross88-90), and sphene.
Locallywollastonite appears to be concentrated along
veinlikesurfaces that may have allowed concentration of
earlymetamorphic fluids.
An unusually detailed petrographic study of a singlethin section
by Berry indicated different local chemicalenvironments that
require treatment in different chemicalsystems (Fig. a8). In the
first place, the rock can be di-vided into regions containing
qtartz + calcite (mainly
Fig. 51. Thin-section sketch in wollastonite-bearing regionwith
abundant diopside, showing garnet rim between anorthiteand
wollastonite. Sketch provided by H.N. Berry.
iL,
ROBINSON: EYE OF THE PETROGRAPHER. MIND OF THE PETROLOGIST
eN(Fe,Mg)O
s i
g,
* . .
_ t. d . '
Ar2o3
Fig. 52. The chemical tetrahedron SiO,-CaO-AlrOr-(Fe,Mg)O,
showing the compositions of quartz (QZ), anorthite
(AN),wollastonite (wO), diopside (DI), calcite (CC), and the
garnetsolid solution grossular (GR) to almandine-pyrope (ALM,
PY).From Berry (1989).
plagioclase + calcite regions of Fig. 48) and those con-taining
qtartz + wollastonite (wollastonite regions of Fig.48). Clearly
conditions of the calcite + qvartz: wollas-tonite reaction had been
reached in some parts ofthe rockbut not in others, suggesting a
very locally variable XCO,of the fluid may have obtained at the
peak of metamor-phism.
Secondly, the rock can be divided into regions whereplagioclase
and calcite are in equilibrium and there is nogarnet, and other
regions where plagioclase and calciteare separated by abundant
garnet. The regions of abun-dant garnet are invariably close to the
widely scatteredgrains of ferroan diopside. This is illustrated in
Figure 49and more dramatically in Figure 50 where garnet hasgrown
at the expense of anorthite + calcite only in thevicinity of
diopside. The necessary participation of fer-roan diopside in
garnet-forming reactions is shown by theanalysed garnet
composition: grossular 88-90, alman-dine 5, pyrope 0.5-0.7,
spessartine 1.3-1.8, atdradite2.T-3.4. An analogous reaction has
taken place in wollaston-ite-bearing regions near diopside, where
wollastonite andanorthite are not in contact and anorthite is
rimmed bygarnet (Fig. 5l). These critical textural observations
em-phasized the role of the diopside component in all
thegarnet-producing reactions in this rock and led to
consid-eration of equilibria in a more complex system and
thederivation ofa specific new petrogenetic grid.
Chemography
The consideration of calcite-quartz-wollastonite equi-libria
requires SiO,-CaO-H,O-COr, the addition of an-orthite and grossular
reactions requires the addition of
sio2
a
-
CaO
ALM. PY
'ff,
(Fe,Mg)O Azog
QZPROJECTION
@
CaO-AlrO,
ANORTHITE PROJECTIONFig. 53. Quartz projection of phase
compositions onto the
base ofthe tetrahedron ofFigure 52 and anorthite projection
ofthe same compositions onto the plane SiOr-CaO-(Fe, Mg)O. Inthe
anorthite projection ALM, PY garnets plot at negative valuesofCaO.
From Berry (1989).
AlrO., and the addition of ferroan diopside and the realgarnet
compositions requires the addition of (Fe,Mg)O,here lumped for
simplicity as one component. The re-sulting chemical tetrahedron is
shown in Figure 52. Allthe phases plot at point compositions except
garnet, whichhas a theoretically complete range from pure grossular
topure almandine-pyrope. When projected from the quartzapex onto
the base of the tetrahedron, the phase com-positions appear as
shown in Figure 53 (top). A morehelpful projection is from
anorthite onto the plane SiOr-CaO-(Fe,Mg)O (Fig. 53 bottom), which
already containsthe phases qtrarlz, wollastonite, calcite, and
diopside.Garnet projects as a line below the
diopside-wollastonitejoin. The grossular end of the series projects
positivelybetween calcite and wollastonite, but the
almandine-py-rope end projects to negative values ofCaO well
outsidethe triangle ofprojection. The big advantage ofthis
pro-jection is that it allows direct observation of the
specialgarnet compositions that are coplanar with other phasesin
the tetrahedron, and it is used in all discussion thatfollows.
ROBINSON: EYE OF THE PETROGRAPHER. MIND OF THE PETROLOGIST t 8 0
l
QZ+Ct+GAR
QZ+CAR+CO2
DI+AN+qZ+CC AR+CO2
Fig. 54. Anorthite projections showing three chemographiesof the
equilibrium DI + AN + QZ + CC + GAR + CO' (re-action 14 of Berry,
1989) depending on the equilibrium com-position ofgarnet. Other
numbers indicate continuous reactionsdescribed in detail by H.N.
Berry. From Berry (1989).
A new petrogenetic grid
In Figure 54 are illustrated the chemographic relationsof the
equilibrium quartz + calcite * anorthite + diop-side * garnet
(reaction l4), all obviously within the sta-bility of calcite +
qtrartz. If garnet is less grossular thanthe
quartz-diopside-anorthite plane, as in reaction l4A,then the join
diopside-anorthite gives way to the assem-blage quartz + calcite +
garnet, consuming CO', and isthus a garnet-forming reaction that
would take place withdecreasing f. If the garnet composition is
slightly moregrossular than the quartz-diopside-anorthite plane, as
inreaction l4B, then the plane diopside-anorthite-calcitegives way
to quartz + garnet * minor CO, and is a gar-net-forming reaction
that would take place with increas-ing Z. If the garnet composition
is more grossular thanthe diopside-calcite-anorthite plane, as in
reaction l4C,then all the phases quartz-calcite-diopside-anorthite
giveway to garnet + a large amount of COr. The limitingsituation
would be the reaction where quartz-calcite-an-orthite gives way to
grossular + CO, without participa-tion of diopside.
Of course all of these segments are part of the samecontinuous
reaction, which must change its P-I slopedrastically in the
presence of CO'-bearing fluid, and eachsegment is separated from
the next by an invariant pointcharacterized by a particular
coplanarity of phases. Be-tween reaction l4A and reaction l4B,
invariant point
+AN
REACTION
DI+ AN +CP2
REACNON
DI+AN+CCsio2-2Al2o3
REACTION
GAR+ CO2
-
I 802
>q: r.r. ral.o
T '-bOc
Fig. 55. Plot ofreaction coefrcients versus XG ofgarnet
forreaction 14 of H.N. Berry. Points I67c, I83, and Il00c
representgarnet compositions at invariant points produced by
collineari-ties ofgarnet composition with other phases. From Berry
(1989).
I67c represents the copianarity of
quartz-diopside-anor-thite-grossular 67. Between reaction l48 and
reaction l4C,invariant point I83 represents the coplanarity of
diop-side-calcite-anorthite. Invariant point I I 00 represents
thelimiting coplanarity of quartz-calcite-anorthite with
gros-sular. The varied stoichiometry of reaction 14 with vari-able
grossular content ofgarnet is illustrated in Figure 55,showing how
CO, changes dramatically from a reactant
ROBINSON: EYE OF THE PETROGRAPHER, MIND OF THE PETROLOGIST
for garnet compositions below grossular 67 to a productthat
becomes increasingly voluminous to a maximum forpure grossular. In
the vicinity of 167 , the CO, coemcientis vanishing and there is no
AS of devolatilization, andhence the reaction is primarily pressure
dependent.
From the above considerations, Berry constructed theP-?"loop
shown in Figure 56 (left). At each of the threeinvariant points the
curve of reaction 14 is tangent to asecond univariant curve. At 167
this is the fluid-indepen-dent reaction grossular 67 * quartz:
diopside + anor-thite, which has a metastable extension to lower
temper-ature. This reaction is critical in subsequent
discussions.At I83 the second univariant curve is the
decarbonationreaction diopside + calcite + anorthite: grossular 83
+qtrartz + COr, which has a metastable extension to higherpressure.
At Il00 the second reaction is the decarbon-ation reaction calcite
+ anorthite + qvarlz: grossular100 + COr. In this case the reaction
14 curve comes intangent to the other reaction at an end point, and
thereare no metastable extensions. To me, the resulting dia-gram
far exceeds the worst nightmares encountered inJ. B. Thompson's
phase equilibrium course thirty yearsago. My student Berry had only
to experience this byproxy. Figure 56 (righ| shows the same
equilibria butwith the addition ofgarnet composition isopleths for
var-ious continuous reactions within the region of stability
ofgarnet + quartz-calcite-anorthite.
QZ + CC + AN + Ca-poorer GAR > Ca-r icher GAR + CO2
iri;e/Gr (67) GAR + QZ
l 8l ol +l 6
6'g
/:
J /o l
N
+
o
r@
Az
+cc
+o
CC + D l + AN + GAR > Gr (83) - r i chor GAR + CO2
T +
Fig. 56. P-I diagrams in presence of COr-bearing fluid, showing
relations of reaction 14 of H.N. Berry to other reactionsinvolving
garnet in the chemical system of Figure 54. Diagram to left shows
intersections of reaction 14 with other reactionsinvolving
collinearities ofgarnet compositions with other phases. Hachures on
reaction 14 indicate the three segments: l4A (triplehachures), 14B
(double hachures), and l4C (single hachures) demarcated by
invariant points I67c and I83. Diagram to right showsisopleths
ofgarnet composition for various reactions within the srability
field ofgarnet. Adapted from Berry (1989).
-
m\ /
D l + A N
REACTION 168:
Dl+ AN + WO GAR + QZ
Fig. 57. Anorthite projections showing two chemographiesof the
equilibrium DI + AN + QZ + WO + GAR (reaction 16of Berry, 1989)
depending on equilibrium composition of garnet.From Berry
(1989).
An analogous but simpler reaction sequence in the fieldof
stability of quartz-wollastonite instead of quartz-cal-cite is
illustrated in Figure 57. Ifgarnet has less grossularthan the
quartz-diopside-anorthite plane as in reaction16A, then the join
diopside-anorthite gives way to theassemblage quartz + wollastonite
+ gamet, a fluid-in-dependent reaction that would take place with
increasingZ. If the garnet composition is more grossular than
thequartz-diopside-anorthite plane, as in reaction l68, thenthe
plane diopside-wollastonite-anorthite gives way toqvartz-garnet, a
fluid-independent reaction that wouldtake place with decreasing Z.
The limiting situation wouldbe the reaction where
wollastonite-anorthite gives way togrossular-quartz with decreasing
Twithout participationof diopside.
Reactions I 64, and l6B are parts ofa single continuousreaction
marked by two invariant points, as shown inFigure 58 (top). At each
of the invariant points the curveof reaction I 6 is tangent to a
second univariant curve. AtI67w this is the fluid-independent
reaction grossular 67+ quartz: diopside + anorthite, which has a
metastableextension to higher temperature. This reaction is the
sameone that connects at lower temperature toI67c in Figure56. At
Il00 the second reaction is the fluid-independentreaction
wollastonite + anorthite : grossular + quartz.In this case the
reaction 16 curve comes in tangent to theother reaction at an end
point, and there are no metasta-ble extensions. Figure 58 (bottom)
shows the same equi-libria but with the addition of garnet
composition iso-pleths for two continuous equilibria within the
region ofstability of garnet * quartz-wollastonite-anorthite. All
ofthe equilibria in Figure 58 are fluid independent.
The equilibria in Figure 56, with quartz-calcite stable,and
those in Figure 58, with wollastonite stable, are crudemirror
images of each other, reflected across the fluid-
T +
GAR + QZ > WO +-AN + Ca-poorer GAR
Fig. 58. P-Z diagrams showing relations of reaction 16 ofH.N.
Berry to other reactions involving garnet in the chemicalsystem of
Figure 57. All equilibria are fluid independent. Dia-gram at top
shows intersections ofreaction 16 with other reac-tions involving
collinearities of garnet compositions with otherphases. Hachures on
reaction 16 indicate the two segments: 16A(square hachures) and I
68 (triangular hachures) demarcated byinvariant points I67w.
Diagram at bottom shows isopleths ofgarnet composition for
reactions within the stability field ofgar-net. Adapted from Berry
(1989).
dependent reaction quartz + calcite : wollastonite * COr.These
are shown in Figure 59, in which fluid-dependentand
fluid-independent equilibria are keyed differently'Most of the
equilibria in Figure 56 are fluid dependent;those in Figure 58 are
all fluid independent, but theyshare the two fluid-independent
equilibria connectingl6Tcto I67w and ll00c to Il00w. At low xco,
(Fig. 59A),the fluid-dependent equilibria are at low Z. As XCO,
in-creases, they and the related invariant points slide tohigher
temperature along the fl uid-independent equilibria(Fig. 59B, C,
and D), progressively shrinking the stabilityfield of pure
grossular in the center and the field of gros-
sular-rich garnet on either side. Eventually the two in-variant
points Il00c and Il00w merge along the calcite+ quartz :
wollastonite * CO, reaction (Fig. 59C), andat higher values of XCO,
(Fig. 59D), progressively lessand less grossular garnet can be in
equilibrium with quartz-
ROBINSON: EYE OF THE PETROGRAPHER, MIND OF THE PETROLOGIST
1803
REACTION 164:
W O + G A R + Q Z
No+
E
o
i
N
l ( ,
p l ,z
E
o
-
lozw
'l /
/
1804 ROBINSON: EYE OF THE PETROGRAPHER. MIND OF THE
PETROLOGIST
G R O S S U L A R
:/ /
:/ "r**2
GROSSULAR
IroocIroow
Iroocw
/D . /
/Fig. 59. Composite P-7 diagrams of phase relations of DI-
AN-QZ-CC-WO-GAR-CO, in Figures 56 and 58 separated bythe
equilibrium CC + QZ : WO + CO,. Fluid-dependent re-actions (dashed)
slide toward higher temperature against fluidindependent reactions
(solid) as X.o, increases from A to D.Region of end-member
grossular stability is marked in the cen-ter, and areas of
stability of less grossular garnet are indicatedby stipple. Adapted
from Berry (1989).
calcite-wollastonite-anorthite-diopside. Further, as
XCO,increases, the total length of the combined reactions 14and 16
that govern the stability ofgarnet with anorthite-diopside and
either quartz-calcite or wollastonite de-creases.
Petrogenetic interpretation
Berry's petrogenetic interpretation of these rocks, basedon the
petrographic features and the petrogenetic grid de-veloped above,
is illustrated in Figure 60. The rocks at-tained the quartz +
calcite: wollastonite reaction at rel-atively low XCOr, and once
encountered, the XCO, ofthe fluid increased. Slightly later in this
history, a fluid-independent reaction was encountered that
produced
Fig. 60. P-?rdiagram to illustrate P-Z-fluid evolution
ofwol-lastonite marble. Isopleths I illustrate CC + QZ: WO + CO,for
various X.o,. Reaction 2 is AN + WO : GR. Reaction 4'is the
reaction CC + PLAG : SCAP. Reaction 15 is the contin-uous
fluid-independent reaction WO + AN * Ca-poorer GAR: Ca-richer GAR +
QZ. Reaction l6b defines the limit of GARstability with DI + AN +
QZ + WO as in Figures 58 and 59.From Berry (1989).
scapolite from plagioclase * calcite, which removedpractically
all of the albite component from plagioclase.Following a curye of
increasing T and P while maintain-ing an increasing XCO, consistent
with the quartz * cal-cite : wollastonite equilibrium, the rock
eventuallyreached a pressure where calcite + wollastonite +
anor-thite + diopside reacted to form garnet + quartz. Thiswould
have been for a composition of grossular 83 alongthe fluid
independent reaction l68, which would in effectalso be at invariant
point I83cw in the terminology ofFigures 59C and 59D. Further
reaction allowed the garnetto become slightly more grossular-rich,
to about grossular90, but beyond this it appears that either
compressionand heating ceased or reaction rates due to fluid loss
orslow diffusion did not permit garnet of higher grossularcontent
to form. The necessary counterclockwise P-Zpathfor this sequence of
events independently supports theearly part of the P-T path claimed
for this region on thebasis of assemblages in aluminous rocks
(Schumacher etal., 1989; Robinson et al., 1989).
Summary
In the example given above, the role of the diopsidecomponent in
all garnet-producing reactions is empha-sized, and this leads to
consideration of equilibria in amore complex system and the
derivation of a specific newpetrogenetic grid. In the
interpretation of this rock, onecan see that it is not just the
phases present in the rock
14A
I
c.
K ,roo
- -- REACTION r5|SOP|.€THNUMSERS SI'loW GARNETcoMPosrIoN
- ---'- REACTION 1 CONTOURNUMBERS SHOW FLUIDCOMPOSITloN
I67c
-
ROBINSON: EYE OF THE PETROGRAPHER. MIND OF THE PETROLOGIST I
805
Fig. 62. Eclogite boudins (dark smudges) in
metamorphosedsandstones ofthe Seve nappe, south slope ofGrapesvare,
Swed-ish Caledonides. Photograph by the author.
tifully developed with the growth of coarse metamorphictextures.
Nearby, however, neither the eclogite nor itscountry rock may show
any remaining vestiges of its eclo-gite-facies history, being
entirely overprinted by amphib-olite-facies assemblages.
Eclogitized diabase dikes in the Grapesvare area'Swedish
Caledonides
My first example is from rocks of the Seve nappe inthe
Grapesvare area of the Swedish Caledonides (Fig. 6l)described by
Andr6asson et al. (1985), Santallier (1988),and Msrk et al. (1988)
and visited personally by me inAugust 1990. The rocks in question
are metamorphosedProterozoic sandstones and subordinate shales
injectedbv diabase dikes. believed to have formed in an exten-
Fig. 61. Geologic index map of the Scandinavian Caledon-ides
showing eclogite localities discussed in this paper. Adaptedfrom a
figure provided by P.-G. Andr6asson.
and their compositions that are important but also thespatial
and textural relations of the minerals that explainthe reactions
involved and their probable sequence.
Mnr,luonpHrc HrsroRy UNDERSTooD oNLyIN TERMS OF FEATURES OF
LARGER ROCK BODIES
In the last segment of this paper, I consider petrologyon a
larger scale. In many places, rock history can onlybe charted
through study of the complete geologic settingof a large outcrop or
even an entire mountain. Particu-larly spectacular are
well-preserved pyroxene granulites,gabbros, and diabases in the
Scandinavian Caledonides(Fig. 6l) that are locally transformed
along shear zonesinto eclogite and amphibolite. Here a host of rock
com-positions all appear to have been subjected to eclogite-facies
conditions, but the exact record they give of thiswas apparently
severely dependent on very local stressand fluid availability. In
some locations, the original pro-toliths, such as coarse-grained
Proterozoic gabbros, py-roxene granulites, and diabases, appear to
have been to-tally protected or only slightly affected by
Caledonianeclogite-facies overprinting. Elsewhere, eclogite is
beau-
Fig. 63. Photogeologic map of distribution of metamor-phosed
diabase layers and boudins at Grapesvare. Boudins innorthern
two-thirds of map area are eclogite; those to south (in-dicated D)
are diabase with garnet coronas and omphacitic py-roxene rims.
Modified from a figure provided by P.-G. Andr6as-son.
-
l 806
Fig. 64. Outcrop views of eclogite boudins at Grapesvare.Top:
Lower contact of large boudin with enveloping countryrock below.
This boudin could not be seen on August 7,1990,because ofsnow cover
from the previous winter. Bottom: Partialviews of two small
boudins. Light colored area in center of bou-din is garnet-rich
eclogite; darker ends are amphibolitized. Pho-tographs provided by
P.-G. Andr6asson.
sional environment along the margin of Baltica. In thisthey
closely resemble the typical rocks of the structurallylower Siirv
nappe and differ from the latter mainly inbeing involved in more
intense metamorphism. A typicalview of the field area is shown in
Figure 62.The exposureis about 800/0, intemrpted only by a few
areas of moss,small bogs, and lakes. The lighter-colored areas are
highlyfolded and lineated metamorphosed feldspathic sand-stone and
minor shale with rare calc-silicate. The darkeroval areas are large
boudins of metamorphosed diabaseup to about 15 m thick. The
detailed distribution of maficboudins over an area of about 8 x l0
km is shown inFigure 63. Some of the boudins occur in rows and
areinterpreted to be dismembered portions of individualdikes.
Within the map area, there are no obvious strati-graphic or
tectonic contacts, though, ofcourse, there is ahigh degree
ofoverall strain.
In the northern two-thirds of the map area, all of themafic
boudins are eclogite, usually with a sheath of ret-rograde
amphibolite of variable thickness (Fig. 64). Eclo-gite is well
preserved in some very small boudins (Fig.64, bottom), whereas
other large boudins are extensively
ROBINSON: EYE OF THE PETROGRAPHER. MIND OF THE PETROLOGIST
Fig. 65. Photomicrograph of eclogite from Grapesvare,showing
zoned garnets about I mm in diameter in a matrix ofquartz,
omphacite, hornblende, and dark fine-grained symplec-tite of
pyroxene and plagioclase after omphacite. Provided byP.-G.
Andr6asson.
G A R N E T
E P I D O T E \
/ / I n c r e a s i n g P
Fig. 66. Petrographic detail ofGrapesvare eclogite. Top:
De-tailed photomicrograph of gamet at extinction under
crossednicols separated from fresh omphacite by a zone
ofplagioclase-pyroxene symplectite. Bottom: Sketch of the same
contact re-gion. Epidote from an original low-grade assemblage
wasoyerg.rown by garnet, with progressively increasing Xr" and
withomphacite inclusions Jd31 that were in contact with matrix
om-phacite Jd35. During retrogression some omphacite broke downto
pyroxene-plagioclase symplectite and was replaced locally
orelsewhere completely by amphibole. Provided by P.-G.
An-dr6asson.
\ .\ J d
i a "
,j
/ r . :
-
Fig. 67. Large boudin of metamorphosed diabase from thesouthern
part ofFigure 63. Note two geologists for scale to left.Photograph
by the author.
amphibolitized. The typical eclogite consists of small eu-hedral
garnets in a matrix of well-preserved omphacite,of retrograde
pyroxene-plagioclase symplectite, and ofsecondary amphibole (Fig.
65). There is no preservationof original igneous texture except for
rare light-coloredpatches that may be relict plagioclase
phenocrysts.
A detailed study area along a garnet-omphacite contactis shown
in Figure 66. The core of the garnet contains
Fig. 68. Photomicrographs of metamorphosed diabase. Top:Ophitic
texture of plagioclase laths with abundant coronas ofgarnet at
extinction under crossed nicols. Bottom: Plagioclaselaths with
epidote needles separated from fine-grained areas ofpyroxene and
amphibole by coronas of high-relief garnet. Pro-vided by P.-G.
Andr6asson.
1 807
Fig. 69. Photomicrograph of margin of original igneous py-roxene
in metamorphosed diabase boudin. Rim area is suffirsedwith exsolved
ilmenite, and remaining pyroxene in rim area con-tains up to 8o/o
jadeite content. Provided by P.-G. Andr6asson.
epidote believed to have been included during progradegarnet
growth. The garnet shows apparent prograde growthzoning from Xr, :
0.35 to 0.40, and the garnet containsomphacite inclusions with
composition Jd3l as com-pared to Jd35 in the coarse matrix
omphacite. Omphacitebreakdown to symplectite of less sodic pyroxene
* pla-gioclase is localized near the garnet contact. Closest tothe
garnet, this consists of pyroxene Jd25 and An20; atthe symplectite
edge farthest from garnet, it consists ofpyroxene Jdl I and An25.
Very commonly the pyroxeneof this reaction symplectite is replaced
by hornblende.Santallier (1988) has estimated conditions of
eclogite for-mation at about 700 "C and 19 kbar. Detailed studies
ofNd-Sm mineral isochrons in the eclogites (Mork et al.,1988)
suggest that the eclogite-facies metamorphism tookplace at 500 Ma
in the so-called Finnmarkian event, wellbefore the Scandian final
emplacement of the thrust sheets.
In approximately the southern one-third of the maparea all of
the mafic boudins are recognizable texturallyas diabase, with a
fine ophitic texture of former igneousplagioclase and, locally,
plagioclase phenocrysts recogniz-able in hand specimen. The form of
the boudins (Fig. 67)and the character of the enclosing rocks is
the same. Inthin sections (Fig. 68) there seems to be evidence for
twometamorphic events, an early greenschist-facies meta-morphism
with formation of epidote in plagioclase anda later high-pressure
event with the formation of delicategarnet coronas between
plagioclase and mafic minerals inthe matrix. Some original Ti-rich
calcic igneous pyroxeneis still preserved (Fig. 69) but has been
attacked margin-ally with exsolution of Fe-Ti oxides and shows an
in-crease in jadeite component up to 8o/0. It appears thatthese
rocks were just beginning the process of eclogiteformation but had
gone through very much less of theprocess as compared with similar
former dikes just a ki-lometer or two away to the north on the same
mountain.What was it that caused these dramatic differences
inreaction progress within the confines of a small area ln asingle
thrust sheet? Andr6asson and coworkers are now
ROBINSON: EYE OF THE PETROGRAPHER, MIND OF THE PETROLOGIST
-
I 808
Fig. 70. Pavement outcrop at Vinddoladalen, Trollheim,Norway,
showing progressive development of eclogite along ashear zone in
coarse ophitic gabbro (shown to left). Gray areasare dominated by
garnet, which tends to form coronas aroundlarge remaining
plagioclases. Photograph by the author.
striving to understand the meaning of these differencesby
studying mafic boudins in a transitional position mid-way up the
slope.
Eclogitized Proterozoic gabbro, Vinddoladalen,western Trollheim,
Norway
My second example comes from exposures of Protero-zoic basement
rock about 100 m below a basement-cov-er contact in Vinddsladalen,
western Trollheim (Fig. 6l),studied by Tsrudbakken ( I 98 I ) and
visited by me in I 98 l.Here, there is an extensive sequence
ofCaledonide thrustsheets, and both basement and tectonic coverhave
been very complexly refolded. Within this basementthere is a boudin
of coarse gabbro and metamorphosedgabbro about I 80 m thick. Within
the core of the boudin,the gabbro appears pristine, with large
plagioclase lathscommonly retaining finely exsolved opaque dust.
Locallywithin the gabbro, there are shear zones, several ofwhichare
well exposed on polished pavements (Fig. 70). Alongthe margins of
the shear zones the gabbro shows garnetcoronas around plagioclase,
and inward the rock changesto a gneissic rock with some relict
plagioclase and thento an eclogite gneiss. The outer margins of the
gabbro andeclogite against Proterozoic gneissic country rock are
ev-erywhere converted into gamet amphibolite or even gar-net-free
amphibolite, in some locations apparently pro-duced by hydration
directly from gabb