J . metamorphic Geol. 1983.
I , 241-262
Contrasting textures in metamorphic and anatectic migmatites :
an example from the Scottish Caledonides
E. L. McLELLAN, Department of Geology and Geophysics, Yale University,
New Haven, Connecticut 06511
Abstract. A method for the quantitative analysis
of the spatial relations of minerals is described.
Dispersed distributions are formed by annealing
and destroyed in post-tectonic migmatuation.
Aggregate distributions characterize solid-state
differentiation, whereas leucosomes formed in
systems of high fluid:rock ratio (in the examples
studied, anatectic melts) show random
distributions.
Quantitative textural analysis can be used to
indicate whether migmatization was post-tectonic
or earlier, though caution is necessary if post-
migmatite cooling is slow or if there is some minor
deformation. More importantly, it can be used to
discriminate melt-present from melt-absent
leucosomes; this is exemplified by a suite of
metamorphic and anatectic migmatites from the
Scottish Caledonides.
The textural evolution of anatexites with
increasing melt percentage is traced. Initial
feldspar porphyroblastesis occurs by Ostwald
ripening via grain boundary melts; subsequently
ophthalmites develop with fabrics and chemistry
inherited from the palaeosome. At greater than
30% melt these inherited fabrics are wholly
destroyed. Deformation prompts segregation
into melanosome and leucosome; resultant leuco-
somes contain no inherited crystals. The scale of
anatectic systems is fixed at the point at which
segregation begins; ophthalmites provide evid-
ence for melt and crystal transfer beyond original
palaeosome boundaries. In contrast, meta-
morphic migmatites are necessarily small-scale
systems because of diffusive constraints, and
melanosomes are invariably produced.
Key-words: anatectic; metamorphic; migma-
tisation; Scottish Caledonides; textures
INTRODUCTION
The bulk of this paper is concerned with
quantitative textural analysis of migmatites, a
02634929/83/0900-0241 $02.00
0 1983 Blackwell Scientific Publications
subject largely neglected since the pioneering
work of Ashworth (1976, 1979) but recently
revived by Karlsson & Wahlgren (1982). Textures
in granites have been the subject of stochastic
modelling by Whitten & Dacey (1974), but the
interpretation of contact patterns in crystalline
rocks is in its infancy (Whitten, Dacey &
Thompson, 1975). The main aim of this study is
to investigate the possibility of distinguishing
textures produced in the solid state from those
formed by crystallization of a melt, and thus to
discriminate melt-present (anatectic or mag-
matic) and melt-absent migmatites. As no experi-
mental data on texture production are presently
available, the approach adopted here is essen-
tially empirical and observational, dealing with
migmatites whose origin is already established.
The effect of post-migmatization modification is
examined and it is shown that textural analysis
combined with thorough petrographic study can
provide valuable genetic information.
Clearly during progressive anatexis there will
be a transition from textures dominated by solid-
state processes to those dominated by the melt
and this will in general complicate the deter-
mination of migmatite origin. Textural variability
within the studied anatectic migmatites is de-
scribed and explained in terms of mechanical
interaction between crystals and melt during
progressive anatexis. The resulting general model
for the physical evolution of anatectic systems
predicts differences in the small-scale geometry of
metamorphic and anatectic migmatites. These
differences are exemplified with respect to the
studied migmatites.
The following nomenclature is used:
Anatexis: development of a melt by in sifu partial
melting.
Leucosome: leucocratic body forming part of a
migmatite; generally rich in quartz and
feldspar(s).
Melanosome: melanocratic body forming part of
a migmatite; rich in mafic minerals.
Mesosome: unmigmatized body occurring in a
migmatite suite.
24 1
 242 E. L. McLellan
Migmatization: formation of a migmatite by any
process.
Neosome: a body produced during migma-
tization.
Ophthalmite: a migmatite characterized by the
occurrence of coarse porphyroblasts.
Heterogeneity of grain size is more marked
than heterogeneity of composition, and
segregation of species is typically very
limited.
Palaeosome: a precursor to a neosome; by
definition palaeosomes ultimately disappear
during migmatization.
Schlieren migmatite: a migmatite in which
melanosomes and/or mesosomes occur as
schlieren (irregular masses with diffuse boun-
daries); leucosome forms more than 50 % of
the migmatite.
Stromatic: a migmatite with a small-scale layered
structure.
INVESTIGATION METHODS
Experimental procedure
Textures are investigated using the line transect
method of Kretz (1969), which measures the
extent to which contact relations between min-
erals deviate from random. A sampling traverse
is laid out with a point counter stage and each
grain contact encountered during the traverse is
recorded. The interval between long traverses
must be greater than the maximum grain size
encountered to avoid repeat encounters of the
same contact. In coarse migmatites this means
that the total number of contacts recorded (n) is
low. This can be corrected for by duplicate
sections, but in practice this was found to be
unnecessary as results are little modified by
increase in n. Results will be biased if samples with
strongly inequidimensional shapes or uneven
grain size distributions are used, and samples
were selected to avoid this where possible. The
effect will be of decreasing importance with
increasing n. One problem is that the number of
degrees of freedom (v) does not depend on the
sample size, so it is not possible to quantify such
bias. Its importance can be reduced by standard-
izing the sample size wherever possible.
. The total number of contacts of each type is
entered as cells in contingency tables. Accessories
are grouped with biotite or, if no biotite is present,
with quartz. This avoids the occurrence of cells
with frequency less than five, which can distort
results. Contingency tables are available upon
request from the author; reasons of space dictate
that the results be presented in the form of x 2 , x 2 / v
values and as observed/expected frequencies for
each contact type.
If a system is random then the probability of a
given event is independent of the preceding event.
Thus, if two mineral species X and Y are
randomly distributed, the probability of passing
from a grain of X to a grain of Y = the overall
possibility of encountering a grain of Y:
where
(total transitions to Y)
x (total transitions from Y)
= total number of transitions
and can be calculated from row and column sums
in the contingency table. The values of xz, where
I
(observed frequency -
x2=E
[ expected frequency)2
expected frequency
are then calculated and compared with the value
of xz predicted for a random distribution with the
appropriate number of degrees of freedom. To
allow comparison between different samples in
which there are different degrees of freedom,
results are normalized to x2/v.
Deviations from random may be expected to be
of two types: (a) regular (dispersed), in which
contacts between grains of different minerals are
more frequent than expected, P ( Y / X ) > P ( Y ) ;or,
(b) aggregare, in which boundaries between
grains of the same species are more frequent than
expected, expressed by the formation of mono-
mineralic clusters, P(Y/X) < P o . These two
distributions can be distinguished by consider-
ation of the ratio of observed to expected
frequency for contacts between like species, e.g. if
this ratio for plagioclase/plagioclase contacts is
very much greater than one, an aggregate
distribution is implied.
Interpretation of dispersed distributions
Like/like contacts are of high energy (Flinn, 1969)
and so will generally be removed by grain growth
if extensive diffusion is possible (Chadwick,
1972). Thus a dispersed distribution (in which
there are very few like/like contacts) is predicted
to characterize rocks which have undergone
extensive solid-state annealing, i.e. most high
grade metamorphic rocks. To test this hy-
pothesis, some examples of granulites (thin
sections from the Harker Collection, Cambridge
 Contrasting textures in migmatites
Table 1. Textural analyses of granulites
Sample Phases n v x2 X’/V
number
84320 P, 0 , C 507 4 52.56 13.4
84322 P, 0 , C 563 4 76.06 19.02
49731 P, 0, C 526 4 50.67 12.67
~-
v = 4; xi,o5= 9.49;
- ~ t =, ~13.28;
, &,05 = 14.86;
L/L = average 0bserved:expectedratio for like/like contacts;L/U = average observed:expectedratio for like/unlike
contacts.
University) were examined with results shown in
Table 1. Dispersed distributions indeed prevail.
Thus, it would be anticipated that palaeosomes
and mesosomes, having long high-temperature
(metamorphic) histories, would have dispersed
distributions. Conversely, a difference in the
textural maturity of mesosomes and neosomes
will support post-annealing migmatization.
Unless migmatization was post-annealing, no
difference need be anticipated between meso-
somes and neosomes as both will have been
affected by the same annealing.
Karlsson & Wahlgren (1982) found no textural
difference between mesosomes and neosomes in
migmatites of unknown age from southwestern
Sweden; however this textural similarity cannot
be the result of post-migmatite annealing, as
mesosomes show not dispersed distributions but
aggregate textures. Thus the mesosomes they
examined were presumably incipiently
migmatized.
Interpretation of aggregate relations
An aggregate distribution is a likely consequence
of growth in the solid state where nucleation will
be strongly controlled by pre-existing crystals; the
lower energy barrier to homogeneous than to
heterogeneous nucleation will favour the develop-
ment of aggregate distributions. Such distri-
butions may, therefore, develop by metamorphic
differentiation S.S. and by precipitation in a solid
matrix from a mobile grain boundary fluid.
Aggregate distributions may also develop by
recrystallization in which one strained grain
breaks down into several new unstrained ones.
This process is likely to be important in coarse-
grained quartz-rich rocks (i.e. most leucosomes)
even at levels of deformation at which tectonic
fabrics are not obviously developed (Ashworth,
1979). This problem is discussed further in the
next section.
243
Observed :expected
P/P o/o c/c L/L P/O P/C c/o L/u
0.43 0.43 0.72 0.53 1.05 1.19 1.26 1.17
0.67 0.34 0.50 0.50 1.35 1.09 1.39 1.55
0.55 0.62 0.63 0.60 1.22 1.25 1.13 1.20
P = plagicclase; 0 = orthopyroxene; C = clinopyroxene;
Interpretation of random relations
A random distribution is likely to develop if
grains nucleate and grow without spatial control
(cf. metamorphic differentiation s.s.). This will
occur if growth occurs in an isotropic medium (a
hydrothermal fluid or a melt). Previous workers
(Whitten et al., 1975, and references therein) have
claimed that some magmatic granites show
Markov relations i.e. are non-random. They do,
however, point out that many of these are deep
seated granites which will have undergone
extensive recrystallization during slow cooling;
the fabrics of many granites are likely to be
metamorphic rather than magmatic (Pitcher,
1979). If nucleation in a melt is essentially
random, then random textures are most likely to
be preserved in a granite which has cooled
relatively quickly. The Lochnagar Newer Granite
is a post-tectonic granite from the study area
(Fig. 1) which was emplaced by cauldron sub-
sidence (Oldershaw, 1974). Such granites are
likely to have cooled quickly (Pitcher, 1979).
Samples from the Lochnagar Granite were
therefore studied in a search for the predicted
random relations. It can be seen (Table 2) that x2
values are compatible with a random origin, but
that there is some tendency toward aggregate
relations amongst the micas. The origin of such
structures is a matter of some debate (Pitcher,
1979; Pitcher & Berger, 1972). Clearly, if such
structures were pronounced it would be difficult
to recognize the originally random magmatic
fabric. The general assumption of random
relations however appears to be valid.
Summary
By comparison of the relative textural maturity
(x2/v) of mesosomes and neosomes it should be
possible to test whether migmatization occurred
post-tectonically or earlier. For post-tectonic
244 E. L. McLellan
migmatites it may then be possible to distinguish
between leucosomes formed by crystallization of
melts or vein fluids (which will generally be
random-textured) and those formed by growth in
a solid matrix (metasomatism or metamorphic
differentiation) which will generally show ag-
gregate relations.
Fig. 1. Locality Map. Migmatites: C = Cowhythe
Gneiss (Ashworth, 1976); D = Deeside Gneisses
(Read, 1955; Sturt el a/., 1977: Ramsay & Sturt,
1979); G = Grampian migmatites (McLellan, 1982a);
S = Strathspey migmatites (Ashworth, 1979). Pre-
tectonic intrusives: 1 = Ben Vuiroch Granite;
2 = Dunfallandy Hill Granite (Bradbury er al., 1976).
Post-tectonic intrusives: L = Lochnagar Granite
(Oldershaw, 1974); G.G.F. =Great Glen Fault;
H.B.F. = Highland Boundary Fault.
These predictions have been tested using a suite
of anatectic and metamorphic migmatites which
lie in the core of the Tay Nappe in the Eastern
Grampians of Scotland. These migmatites are of
undisputed Caledonide age (Johnstone, 1966);
structural evidence indicates that migmatization
occurred syn- and post-D3, D3 being the last
major deformational event in the area. If textural
criteria for migmatite genesis are to be of value
then they must be widely applicable and it is
therefore necessary to examine the possible
modifying effects of post-migmatite cooling and/
or deformation. This forms the subject of the next
section.
APPLICATIONS AND LIMITATIONS
OF THE TECHNIQUE
Tectonic recrystallization
Samples from two Grampian, pre-tectonic (pre-
D3) intrusives, the Ben Vuiroch and Dunfallandy
Hill Granites (Bradbury, Smith & Harris, 1976)
(Fig. 1) whi,ch show various degrees of de-
struction of original texture were selected.
Quantitative analysis of textural relations be-
tween minerals are shown in Table 3. It is
apparent that values of x2 deviate considerably
from random due to aggregate relations and that
the amount of deviation increases with increasing'
(petrographically observed) destruction of orig-
inal textures, i.e. from strained grains to partially
recovered subgrains to recrystallized strain-free
grains. The development of aggregate relations is
clearly attributable to the formation during
recrystallization of new, strain-free grains (Fig. 2)
which increases the number of like/like contacts
without altering the others.
Recalculation of x2 values assuming that
observed frequency = expected frequency for all
contacts between minerals of the same species will
give a minimum value for the x 2 which would be
obtained if subgrain boundaries were ignored in
the traverse. Results, presented in Table 3, show
that x2 values are now compatible with the
hypothesis of a random distribution. That little
distortion is introduced by ignoring like/like
contacts in recalculation is shown by comparison
of x 2 values obtained by this means with those
obtained from a traverse in which subgrain
boundaries were bypassed (Table 3). Obviously,
if this technique were to be applied generally to
the recognition of random relations it would be
easier to ignore subgrain boundaries during the
traverse; this example simply serves to show that
for limited deformation deviations from random
are entirely attributable to recovery. Only when
recrystallization begins (Fig. 2) such that original
strained grains are replaced by new strain-free
grains is it impossible to recognize an original
random fabric and hence an origin involving a
melt.
Thus, whilst distortion will obviously be
introduced if the original magmatic or anatectic
fabric is not wholly random, it seems that a
quantitative textural approach may help in the
recognition of the presence of melts even where
some deformation has occurred.
Retrograde non-tectonic recrystallization
There is some evidence that migmatite textures
Table 2. Textural analyses of granites

Sample Phases n V x2 x=/v Observed :expected

number - _ _
Q/Q P/P K/K M/M Q/P Q/K Q/M P/K P/M K/M L/L L/U

26615 Q, P, K, M 478 9 14.44 1.60 0.92 0.84 0.94 2.24 1.14 1.07 0.85 1.03 0.86 0.87 0.90 0.96
14738 Q, P, K, M 439 9 8.38 0.93 0.79 0.89 0.97 1.20 1.18 1.12 0.97 0.88 0.95 0.92 0.88 1.00
LG4 Q, P, K, M 528 9 16.62 1.85 1.13 1.06 1.09 2.14 0.93 1.01 0.80 1.01 0.97 0.85 1.09 0.93 r\

~~

v = 9; xi.o5 = 16.92; = 21.67; xi.oo5 = 23.59; Q = quartz; P = plagioclase; M =micas; K = K-feldspar; L/L =average 0bserved:expected ratio for like/likecontacts;
~ t , ~ ~ $
-
L/U = average 0bserved:expected ratio for likelunlike contacts. e
x,
T
2
%.
Table 3. Textural analyses of variously deformed granitic intrusives 3
a;;.
Sample Type Phases n V x2 x2/v Observed :expected
5
2-.-
number
Q/Q P/P K/K M/M Q/P Q/K Q/M P/K P/M KIM "x2 5
140777 S Q, P, K, M 494 9 59.32 6.59 1.19 1.22 2.00 1.47 0.52 0.67 0.79 1.04 1.01 0.71 18.07
140970 S Q, P, K, M 511 9 38.99 4.33 1.13 0.70 1.50 2.57 1.50 0.99 0.61 0.90 1.38 1.08 16.52
14732 S Q, P, K 480 4 38.77 9.94 0.98 1.13 3.60 - 0.97 0.59 - 1.05 - - 12.44
140860 R Q, P, K, M 520 9 105.88 11.76 1.35 1.05 2.10 0.67 0.79 0.46 0.96 1.13 1.67 0.67 63.30
140777 T Q, P, K, M 456 9 11.78 1.31 1.15 1.14 1.10 1.25 0.99 0.87 0.88 1.26 1.02 1.01 9.61

v =4; xi,o5 =9.49; = 13.28; x i . 0 0 5 = 14.86.

v = 9; = 16.92; = 21.67; xi,oo5= 23.59.
Q = quartz; P = plagioclase; K = K-feldspar; M = mica; S = subgrains developed; R = recrystallization; T = traverse ignored subgrains.
"x2 = value of x2 calculated by assuming observed = expected frequency for like/like contacts (i.e. leave out subgrains).
h)

R
246 E. L. McLRIlan
Fig. 2. New strain-free grains of K-feldspar produced by recrystallization in sample from Ben Vuiroch Granite.
Scale bar = 2 mm.
can be modified during post-migmatite cooling
even in the absence of tectonic activity. For
example, Ashworth (1979) describes some post-
tectonic migmatites in which there is no macro-
scopic evidence of deformation, but in which
recrystallization of quartz and feldspar has led to
distortion of textural relations. These migmatites
are from Strathspey near the Moine/Dalradian
boundary at stratigraphically and structurally
deep levels (Ashworth, 1979) (Fig. I). In contrast,
in the structurally high-level Cowhythe Gneiss,
post-migmatite textural adjustment is restricted
to undulose extinction in quartz (Ashworth, 1976)
(Fig. 3). Thus, even in post-tectonic migmatites
there may be substantial textural variation, but as
the mode of origin of the Strathspey samples is
indeterminate full interpretation of this variation
is impossible.
However, it is clearly important to determine,
for migmatites formed at the same time and by
the same process, how important such post-
migmatite modification may be. The study area,
lying at greater structural depths than the
Cowhythe Gneiss but likewise containing latest-
Caledonide anatectic migmatites (McLellan,
1982a) provides a suitable test for this. If
retrograde modification increases with structural
depth, Grampian leucosomes would be expected
to show greater variations from random than
Cowhythe leucosomes. First, it is necessary to
determine the range of textures found within the
Grampian leucosomes. All Grampian samples
are from the Flat Belt of the Tay Nappe and are
believed to have identical structural histories.
Whilst some of the Grampian leucosomes show
n o deformation textures, most show incipient
recovery (subgrain formation) in quartz; sub-
grain boundaries can be recognized as such and
avoided and so d o not distort textural analyses.
However, rare examples can be found in which
original quartz grains can no longer be recognized
(Fig. 4) due to substantial recrystallization of
quartz. These leucosomes are otherwise identical
to texturally unmodified rocks, i.e. they are
chemically and mineralogically similar, lack an
S3 fabric and have random feldspar relations.
The difference in texture therefore cannot reflect
differences in the timing of migmatization. The
examples with recrystallized quartz grains are
restricted to the lowest structural level sampled.
 Contrasting textures in migmatites
Fig. 3. Undulose extinction in quartz resulting from development of subgrains during recovery, Cowhythe
Gneiss. Scale bar = 2 mm.
Due to limitation of topography the maximum
structural range sampled was only -700m.
However, results are compatible with the sugges-
tion that the extent of retrograde modification
increases (increasing amounts of recovery fol-
lowed by recrystallization) with increasing struc-
tural depth in the Tay Nappe. Whatever caused
the difference, comparison of the various textures
in Fig. 5 indicates that leucosomes formed at the
same time and by the same process may look very
different texturally, and this may make it difficult
to correlate episodes of migmatization across an
area.
The quantitative and qualitative textures of
various Grampian (low structural levels) and
Cowhythe (high structural levels) leucosomes are
contrasted in Fig. 5. The difference between
Cowhythe and Grampian leucosomes is very
much less than is the variation within Grampian
leucosomes. It is therefore clear that post-
migmatite modification was much more extensive
at the deeper levels of the Grampian examples.
The consistent differences in the extent of
recovery and recrystallization between Cowhythe
and Grampian leucosomes must be related to
247
cover thickness, presumably to more extensive
late deformation at depth (so increasing the strain
energy available to drive recovery and recrystal-
lization) coupled with slower cooling (allowing
more extensive diffusion). Thus the original
textures of the Grampian leucosomes have been
modified, though not destroyed, by recrystal-
lization.
Once recrystallization has occurred further
textural changes in the rock will be controlled by
surface energy distributions and grain growth
may be expected. The relative surface energies of
adjacent grains will once again become important
and it can be predicted that at great depths where
extensive diffusion is possible the aggregate
texture resulting from recrystallization will be
replaced by an annealed texture. Clearly this has
not occurred in the present case. It should be
noted that this process has implications for
identifying the timing of migmatization at deep
structural levels. Thus, if the difference in the
textural maturity of mesosomes and neosomes is
progressively reduced with increasing depth due
to more pervasive annealing, migmatization will
appear to have occurred progressively earlier at
248 E. L . McLellan
Fig. 4. Recrystallized quartz grains in Grampian leucosome. Scale bar = 2 mm.
deeper levels and the textural evidence for its post-
tectonic nature may be obliterated.
Textural reworking us. chemical reworking
Migmatites which are petrographically very
similar to those in the main study area are found
at Cromar, Deeside, some 20-25 miles N (Fig. 1).
The frequent occurrence of andalusite rather than
sillimanite indicates generally lower pressures.
The exact structural position of these migmatites
has been a matter of debate for many pears (see,
for example, Read (1955)) and interpretation of
structural history is rendered very difficult by
intense late mylonitization. Ramsey & Sturt
(1979) and Sturt, Ramsay, Pringle & Teggin
(1977) have, on the basis of pre-Caledonide
( - 710 Ma) ages, re-interpreted this and corre-
latable gneisses (including the Cowhythe Gneiss
of Ashworth (1976)) as pre-Caledonide basement
thrust into its present position pre-D3. On
Ramsay & Sturt’s interpretation, Rb/Sr systems
were not reset during synchronous or subsequent
(D2-D3) amphibolite facies metamorphism.
Krogh & Davis (1973) quote examples in which
resetting at such grades has not occurred.
Textural analysis (Table 4), however, indicates
that there is a difference in textural maturity
between mesosomes and neosomes, suggestive of
post-tectonic (post-D3) migmatization. If these
rocks are pre-Caledonide basement then these
textures must indicate late Caledonide reworking
of pre-Caledonide migmatites. If the Rb/Sr ages
represent the time of the last isotopic homo-
genization, as implied by Sturt et al. (1977), then
it must be possible for migmatization and textural
reworking to occur without any isotopic adjust-
ment. This is clearly very unlikely. I t seems more
plausible that either (a) the rocks are Caledonide
metasediments which underwent substantial loss
of R b relative to Sr during Caledonide migma-
tization, giving spuriously old ages, or (b) the
710 M a ages are produced by partial resetting of
still older ages during Caledonide migmatization,
a process similar to that invoked by Brewer,
Brook & Powell (1979) to explain the production
of -740Ma ages in the reworked Grenville
province of the western Moine.
Thus textural studies, by indicating the scale
and timing of extensive diffusion, can aid the
interpretation of isotopic ages.
 Contrasting textures in migmafites 249

plag

Ln
X Cowhythe leucosomes
c
V
c
+ Grampian leucosomes
C
8 1.60 0 Grampian opthalmites
0)

-
0
0
.-
m
0
-
a I
\
m 1.40
-
0
V
._
m
-a
0
0
0'
._
5 1.20
0 0
c
V
n
X
m
\
0
f 1.00 + o
L

%
n
0 +
0
X
x x 0

0.00

I 1 I I I
I .OQ I.20 1.40 1.60
Observed/expected ratio, quartz/quortz contacts

Fig. 5. Textural contrasts in post-tectonic anatectic migmatites. A = field of Cowhythe (high-level) migmatites;
B = field of structurally deep Grampian migmatites modified by quartz recrystallization; C =field of structurally
less deep Grampian migmatites.

Table 4a. Textural analyses of Deeside neosomes

Sample Phases n v x2 x2/v Observed :expected
numbers
Q/Q P/P M/M L/L Q/P Q/M P/M L/u
141004 Q, P, M 249 4 8.29 2.07 1.18 1.14 1.83 1.38 0.89 0.75 0.96 0.87
141005 Q, P, M 256 4 15.53 3.88 0.97 0.86 2.20 1.31 1.19 0.85 0.75 0.93

Table 4b. Textural analyses of Deeside mesosomes

Sample Phases n v x2 x2/v Observed:expected
number
Q/Q P/P M/M L/L Q/P Q/M P/M L/u
141008 Q, P, M 472 4 43.36 10.84 0.59 0.64 0.51 0.58 1.17 1.30 1.15 1.21
141004 Q , P, M 498 4 94.66 23.67 0.31 0.57 0.44 0.44 1.39 1.41 1.03 1.27
-
v = 4; xi,o5= 9.49; xi = 13.28;,, = w 8 6 ; Q = quartz; P = plagioclase; M = mica; L/L = average
0bserved:expected ratio for like/like contacts; L/U = average 0bserved:expected ratio for like/unlike contacts.
250 E. L. M c k l l a n
Summary
The evidence in the above sections indicates that
post-migmatite modification is, in extreme cases,
capable of destroying evidence of both the timing
and the mechanism of migmatization. Certainly
analysis of a few samples from a limited area is
likely to lead to misleading results. If, however, a
variety of samples from a range of structural
depths is available then the role of post-migmatite
modification can be assessed. Used cautiously,
therefore, quantitative textural analyses can
potentially provide information on migmatite
genesis. Such textural studies may help constrain
the interpretation of isotopic ‘ages’.
DESCRIPTION OF TAY NAPPE
MIGMATITES
The studied migmatites lie at low structural levels
of the Tay Nappe and are at kyanite and
sillimanite grade. Two distinct groups of migma-
tites can be recognized, both of which formed by
closed-system processes.
Stromatic and flecky migmatites (Group A
migmatites). Leucosomes have widely variable
compositions which are far removed from poss-
ible cotectic compositions; the locations of
stromatic leucosomes are controlled by stress-
induced potential gradients. Evidence (further
detailed in McLellan, 1982a) therefore suggests
that these formed by metamorphic differentiation.
Ophthalmitic and schlieren migmatites
(Group B migmatites). Leucosomes of schlieren
migmatites show a narrow range of compositions
which are a close approximation to the cotectic
composition in the Ab-An-Si0,-H,O system.
Scatter of compositions away from the cotectic in
ophthalmitic types is related to inheritance of
non-melted material from the palaeosome
(McLellan, unpublished data). Strong normal
zoning in plagioclase and euhedral zone boun-
daries in leucosomes are more readily explained
on an anatectic than a metamorphic hypothesis.
The involvement of high percentages of melt is
also supported by the development of cross-
cutting relations and of highly disharmonic folds
with widely varying axial plane orientations
interpreted as forming by a flow of melt during
deformation (McLellan, 1982b). Thus these
Group B migmatites are believed to be anatectic.
Further evidence in support of this will be
presented elsewhere (Mckllan, unpublished
data).
Structural evidence indicates that migma-
tization occurred syn- and post-D3, D3 being the
last major deformation event in the area. Most
stromatic leucosomes of Group A migmatites are
parallel to S3 (where this is developed by
transposition on the limbs of F3 folds) and wrap
boudins carrying truncated SI-S2 fabrics. Such
leucosomes follow Sl-S2 round the hinges of F3
folds. These relations are compatible with either:
(i) post-D2, pre-D3 migmatization, leucosomes
developing on an S2 fabric and being transported
and boudinaged along with it; or, (ii) post-D3
migmatization mimetic on pre-existing fabrics
(S3 in fold limbs and S1-S2 in fold hinges). The
frequent occurrence of grain size differences
between coarse leucosomes and finer mesosomes
suggests that migmatization was not followed by
any penetrative deformational-annealing events.
Whilst it is not possible unambiguously to
determine the time of migmatization, in at least
some cases final crystallization must have occur-
red post-D3.
Rare stromatic leucosomes of Group A
migmatites develop in the axial planes of F3 folds
and carry an S3 grain shape fabric; these are
believed to have formed syn-D3.
Ophthalmitic migmatites of Group B are
generally schistosity-bounded, but may locally
crosscut both S3 and the fabric in GroupA
migmatites (Fig. 6). Schlieren migmatites of
Group B may have a very weak S3 fabric formed
by parallelism of muscovite crystals; they de-
scribe complex F3 folds (McLellan, 1982b).
Again structural relations are not diagnostic of
the time of initiation of migmatization (melting)
but final crystallization must have occurred post-
D3.
Group A migmatites
Leucosomes consist of quartz and plagioclase, in
variable proportions, and leucosome grain size
(average -5mm) is very much greater than
generally small (stroma -
mesosome grain size (- 2 mm). Leucosomes are
10-20 mm across,
lenses up to 30mm in length) and have a
geometric resemblance both to quartz pods and
veins and to crenulation cleavage seen at lower
grades. Melanosomes are invariably developed;
they are of comparable extent to leucosomes, are
of similar grain size to mesosomes and have a
strong planar fabric parallel to the mesosome/
leucosome contact. The boundary between meso-
some and melanosome is arbitrarily fixed at a
content of l0-15% felsic material. Biotite is the
dominant mica; muscovite is very common in
metapelitic types, and hornblende in metatuffs.
Epidote, sphene, garnet and tourmaline are
 Contrasting textures in migmatites
Fig. 6. Ophthalmitic migmatite of Group B cross-cutting both S3 and the fabric of a Group A migmatite.
- 5 rnrn
Leucosome
Melanosome
Mesosome
Fig. 7. General features of Group A migmatites.
Note grain size contrast between leucosome and
(melanosome + mesosome), but similarity of aspect
ratios throughout. Plagioclase in the leucosome
shows aggregate relations.
locally accessory. Figure 7 illustrates the general
features of these migmatities.
Group B migmatites
These are ophthalmitic or schlieren types. True
leucosomes are distinct in grain size (up to 20 mm)
and texture from associated non-leucosomes,
25 I
while melanosomes are thin and irregularly
developed. Where melanosomes do occur they
have an indistinct foliation parallel to leucosome/
mesosome contacts. They are generally coarser
(up to 10mm) than associated mesosomes, and
melanosome mica is typically less elongate than
mesosome mica. Ophthalmites (grain size up to
15 mm) grade into mesosomes by an increase in
the preferred orientation of micas and a decrease
in feldspar grain size.
In metapelite types, biotite is generally sub-
ordinate to muscovite. This latter occurs in two
generations: primary matrix muscovite of high
length to breadth ratio occurring in mesosomes
and some ophthalmites, and coarser secondary
muscovite (which is more equant and euhedral) in
leucosomes, melanosomes and some ophthal-
mites. All micas are grouped together for textural
analysis; garnet and tourmaline are common and
are grouped with micas. Feldspar porphyroblasts
in ophthalmites may contain small (- 1 mm)
rounded inclusions of quartz and/or micas; such
inclusions were disregarded in textural analysis.
In metatuffs biotite is invariably present, with
or without hornblende. Epidote, garnet and
sphene are common; apatite is rare. Discrete
252 E. L. McLxllan
grains of both magnetite and ilmenite occur
sporadically. These accessories are grouped with
biotite for textural analysis.
Mesosomes in both metapelites and meta-
tuffs are of small grain size (-3mm) and have
an annealed texture. Foliation is pronounced,
marked both by grain alignment and by inequi-
dimensional crystals. All mineral species are of a
comparable size.
The following types of neosome are
recognized :
True ophthalmites. Feldspar porphyroblasts
are only slightly coarser than other species;
muscovite (in metapelites) or biotite accessories
is very coarse, euhedral and dispersed throughout
-
the coarse quartzofelspathic matrix. Grain size is
15 mm, and there is no trace of foliation.
Figure 8b shows the general features of this type.
Semi-ophthalmites. This type grades into true
-
ophthalmites by an increase in the grain size of all
species; average grain size is 12 mm, but is very
variable. Plagioclase is typically coarser than
0 mesosomes typically have dispersed distributions
10 rnrn
H differences.
10 rnrn
Key
Fig. 8. Sketches from photomicrographs. (a) Semi-
ophthalmite. Mesosome with development of coarse
plagioclase porphyroblasts. (b) True ophthalmite.
Note well-rounded plagioclase.
other minerals and may be wrapped by micaceous
stringers which may show some parallelism of
alignment. In metapelites these stringers contain
primary muscovite and may be overgrown by
secondary muscovite. This type (Fig. 8a) is
transitional between true ophthalmites and meso-
somes and shows extreme variation.
-
Discrete leucosomes and melanosomes. Crystal
size in leucosomes is generally coarser ( 25 mm)
than in ophthalmites. Melanosomes are coarse,
lack good foliation and commonly contain 5 %
plagioclase. Metapelite leucosomes contain
-
coarse, secondary muscovite which may show
some alignment. Figure 9 illustrates some of these
features.
RESULTS AND INTERPRETATIONS
Relative textural maturity of palaeosomes and
neosomes
Quantitative textural analyses of mesosomes,
leucosomes and ophthalmites are given in
Tables 5-7 and Fig. 10. It can be seen that
(compare average observed to expected ratios for
like/like and likelunlike contacts), attributed to
prolonged solid state annealing. Leucosomes and
ophthalmites have lower xz values and do not
show dispersed distributions, which is interpreted
as being due to destruction of annealed textures
during D3 migmatization. Thus there is a
difference in the textural maturity of mesosomes
and neosomes, confirming that migmatization
was post-tectonic.
Melt-present us. melt-absent migmatites
According to the discussion above, leucosomes
formed in the solid state by metamorphic
differentiation should show aggregate relations,
while those formed by anatexis should show
random relations. As Group A migmatites are
known to have formed by metamorphic
differentiation and Group B by anatexis, quanti-
tative textural analysis should show these
Group A migmatites
Group A leucosomes are small, and thus very few
measurements can be made; strictly, therefore,
textural analyses are not statistically respectable.
The inadequate textural data of Table 6a indicate
low xz values, compatible with crystallization
being post-tectonic and with some tendency to
 Contrasting textures in migmatites 253

Fig. 9. Leucosome, Group B migmatite. Note well-rounded plagioclase and sharp, euhedral boundaries of com-
positional zones. Scale bar = 4 mm.

aggregate relations for all species. In view of the
apparently post-tectonic origin these aggregate
relations are unlikely to be due to recrystal-
lization, but may be due to epitaxial nucleation in
a solid state differentiation process.
Plagioclase crystals in Group A leucosomes,
whilst coarser than mesosome plagioclase, have
aspect ratios (1ength:width)comparable to meso-
some feldspar (Table 8a). This may suggest that
at the low fluid:rock ratios anticipated for
metamorphic differentiation growth was epitaxial
on pre-existing feldspar. Growth of plagioclase in
a solid matrix will be constrained both by the
imposed matrix shape and, as neosomes subse-
quent to development of an anisotropic fabric, by
anisotropic fluid movement. Inequant crystals

Table 5. Textural analyses of Grampian mesosomes

Sample Phases n V x2 x2/v Observed :expected

number
P/M L/u
~

140742 407 32.88 8.22 1.22 0.76 0.83 0.94 0.74 0.89 1.51 1.05
140743 544 54.11 13.53 0.68 0.48 0.49 0.55 1.13 1.21 1.28 1.21
140782 511 80.05 20.01 0.35 0.48 0.51 0.45 1.26 1.35 1.23 1.20
140789 537 78.60 19.65 0.54 0.37 0.52 0.48 1.23 1.18 1.38 1.26
140954 808 82.23 20.56 1.55 0.83 0.67 1.02 1.09 1.03 1.05 1.06
141098 424 72.51 18.13 0.45 0.45 0.43 0.44 1.22 1.29 1.33 1.28
141198 464 85.98 21.49 1.47 0.56 0.80 0.94 0.33 1.42 1.49 1.08
38005 498 31.19 7.79 1.33 0.43 0.95 0.90 0.37 1.07 1.23 0.89

-’ v = 4; = 9.49; ~t,~,
= 13.28; xt.oo5 = 14.86;-Q = quartz; P = plagiociase; M = mica;
L/L= average 0bserved:expected ratio for iike/like contacts; L/U = average 0bserved:expected ratio for like/
unlike contacts.
254 E. L. McLellan
Table 68. Textural analyses of Group A leucosomes

Sample Phases n V x2 x2/v Observed :expected

number
Q/Q PIP M/M Q/P Q/M P/M
140731 Q, P, M 119 4 8.92 2.23 1.13 1.40 3.00 0.81 0.69 1.16
140902 Q, P 80 1 6.71 6.71 1.19 1.45 - 0.70 - -
140923 Q,P 181 1 11.47 11.47 1.16 1.25 - 0.73 - -
140951 Q,P, M 251 4 14.38 3.59 1.22 1.23 2.32 0.83 0.81 0.92
141038 Q,P, M 430 4 3.92 0.98 1.09 0.98 1.23 0.99 0.84 1.04
141055 Q,P, M 80 4 4.25 1.07 1.11 1.05 4.00 0.95 0.69 1.00
141178 Q,P 499 1 21.39 21.39 1.08 1.51 - 0.79 - -

v = 1; x 2 o, = 3.84; x2 = 6.63; x2 ool = 7.88.

v = 4; x!:ol = 9.94; x!:ol = 13.28; x!:ool = 14.86.
Q = quartz; P = plagioclase; M = micas.

Table 6b. Textural analyses of Group B leucosomes

Sample Phases n V x2 x2/v 0bserved:expected

number
Q/Q PIP Q/P

140743 Q,P 500 1 0.49 0.49 1.00 0.94 1.04

141053 Q,P 500 1 3.39 3.39 1.01 1.50 0.92
141042 Q, P 497 1 5.74 5.74 1.02 1.59 0.89
141155 Q, P 499 1 11.58 11.58 1.05 1.45 0.85
141180 Q,P 500 1 0.21 0.21 0.98 1.00 0.98

Table 7. Textural analyses of ophthalmites

Sample Phases n 0bserved:expected

number
M/M L/L Q/P Q/M P/M L/u
140716* 41 7 4 9.15 2.29 1.11 1.41 1.21 1.24
0.85 0.90
0.95 0.90
140720; 498 4 13.33 3.33 1.20 1.00 1.24 1.04 0.93
1.14 0.97 0.78
140774 496 4 4.13 1.03 1.27 1.00 1.03 1.10
1.03 0.97
0.92 0.96
140869 499 4 49.64 12.41 1.23 1.27 1.67 1.39 0.84
1.09 0.86 0.66
140873 493 4 12.12 3.03 1.11 0.89 1.23 1.08
1.16 0.98
0.97 0.81
140897 499 4 19.94 4.99 1.07 0.98 0.91 0.99
0.92 0.89
0.91 0.84
141052 500 4 7.29 1.83 0.78 0.95 0.86 0.86 0.98
1.06 1.06 1.15
141 153 512 4 15.82 3.96 1.01 0.87 1.05 0.981.42 0.96
1.05 0.42
141 192* 498 4 11.22 2.81 0.87 0.84 0.44 0.72
0.94 1.09
1.13 1.19
-
v = 4; xi,os = 9.49; xi,ol= 13.28; xi.oo,=&86; Q = quartz; P = plagioclase; M =mica; L/L = average
0bserved:expected ratio for likejike contacts; L/U = average 0bserved:expected ratio for likelunlike contacts.
Indicates ophthalmite gradational to mesosome.
 Contrasting textures in migmatites 255

1.6 - rn

8 0

- *0 .
I
1.4

In
c
0
0
*
*
C .A0
1.2.
clr
- A
2-
n A
0' 1.0.
._ * *
*
c
e
U /
./---.,
* A '
8 1 y
8,

*/ 'd
L
0 ( A

<
n

W
0.8 .
\
.T--
I X
X

Cow hy t he
L> Ieucosomea
m A
In
n
0

0.6 .
X

X X
X
X
0.4 '-
X

0.2 0.4 0.6 0.8 I .o 1.2 I .4 I .6

Observed/expected ratio, q u a r t r / q u o r t z contacts

X Grampion mesosomes
0 Group A leucosornes
8 Group B leucosornes
* Ophtholmites
A Gronites
A Deformed gronites
0 Deeaside leucosornes

Fig. 10. Summary diagram of observed:expected ratios for quartz/quartz and plagioclase/plagioclase contacts
for Grampian, Deeside and Cowhythe migmatites, together with various intrusives from the same area.

such as those figured by Vernon (1975) are
therefore expected to develop.
Group B migmatites
Leucosomes (Table 6b) and ophthalmites
(Table 7) in Group B migmatites have random
distributions; where in Table 6a like/like contacts
are much more frequent than expected, and like/
unlike contacts less frequent, in Table 6b like/like
contacts are about as frequent as like/unlike
contacts (though there is some tendency to
aggregation of plagioclase). These relations are
interpreted as forming by growth in a free fluid,
i.e. a hydrothermal vein or melt. Whilst textural
analysis alone does not discriminate between
these possibilities other features of these migma-
tites (closed system origin, plagioclase chemistry,
 256 E. L. McLullan
Table 8a. Aspect ratios of plagklase, Group A DISCUSSION
migmatites
~ Textural evidence, though not unambiguous, can
Sample Type Mean S.D. be used to constrain possible migmatite-forming
processes. Thus, anatectic leucosomes tend to
140711 Mesosome 1.688 0.462 show random relations, whilst metamorphic
140839 Mesosome 1.693 0.278
leucosomes show aggregate relations. However
140843 Mesosome 1.703 0.384
140848 Mesosome 1.621 0.287 the distinction between these types is less clear
141178 Mesosome 1.877 0.652 than one might expect, due to a noticeable
140711 Leucosome 1.702 0.489 tendency for self-association of plagioclase in
140729 Leucosome 1.624 0.322 anatexites. If this tendency is due entirely to
140839 Leucosome 1.528 0.349 recrystallization it would be expected that
140893 Leucosome 1.501 0.315 modification of quartz relations will be at least as
140897 Leucosome 1.588 0.341
marked as that of plagioclase (see, for example,
Fig. 1 of Ashworth, 1979); this is not so and the
cause of the plagioclase self-association remains
Table 8b. Aspect ratios of plagioclase, Group B unknown. Clearly where several processes are
migmatites superimposed, as here, the interpretation of
textural relations of isolated samples is very
Sample Type Mean S.D. difficult.
The textural differences between anatectic or
141154 Mesosome 2.003 0.496 igneous and metamorphic migmatites result from
140720 Ophthalmite 1.289 0.177
140869 Ophthalmite 1.234 0.114 the differing environments of nucleation and
140873 Ophthalmite 1.212 0.181 growth available in a solid and in a melt. Clearly
141I47 Ophthalmite 1.265 0.189 during progressive anatexis there will be a
141179 Ophthalmite 1.294 0.258 transition between these two environments as
141042C Leucosome 1.151 0.154 melt percentage increases. This is likely to
141042F Leucosome 1.223 0.128 complicate the genetic interpretation of migma-
1410426 Leucosome 1.148 0.089
141154 Leucosome 1.197 0.166 tite textures. The great variability of textures seen
in Group B migmatites and discussed below
illustrates this transition very well.

cotectic compositions, irregular rather than

fracturecontrolled geometry of leucosomes)
indicate the latter. Thus textural analysis
discriminates metamorphic from anatectic
leucosomes.
In leucosomes of Group B migmatites plagio-
clase is well-rounded (Table 8b). This is true both
of outer morphology and of the boundaries of
inner zones (Fig. 9); idiomorphic zone boun-
daries are common even in crystals whose outer
shape has been determined by late stage mutual
interference. As the high aspect ratios of Group A
neosomes are an inevitable consequence of
metamorphic differentiation, the very low aspect
ratios of Group B neosomes provide further
evidence for a non-metamorphic origin of these
migmatites. Such rounding is attributed to
diffusioncontrolled growth in an anatectic melt.
The low water content anticipated for such melts
will tend to produce high viscosity, so lowering
diffusion rates and favouring diffusion rather
than interfacecontrolled growth (cJthe experi-
mental results of Dowty (1980) on melts of very
much higher water contents).
TEXTURE EVOLUTION IN ANATECTIC
MELTS
During the initial stages of melting texture
changes very little, although it is petrographically
obvious that aggregate relations may develop in
the micas. Recrystallization of feldspar to coarser
grain size occurs (Fig. 11). Johannes & Gupta
(1982) also note an increase in the grain size of
felsic minerals during incipient melting. Grain
size distributions (Fig. 11) suggest that originally
larger grains grow at the expense of smaller. The
process is therefore an Ostwald ripening in which
high energy small crystals dissolve in a grain
boundary partial melt network, this dissolved
material subsequently re-precipitating around the
originally coarser, low-energy grains. Feldspar
growth is apparently achieved by passive dis-
placement of other species without attendant
recrystallization (Fig. 12), whence the develop-
ment of mica aggregates. Thus feldspar
porphyroblastesis in mesosomes may begin once
a grain boundary melt network is established;
 Contrasting textures in migmatites 251
401- dispersed distributions inherited from the palaeo-
some; clearly these will be destroyed at greater
than 30% melting.
Plagioclase compositions in ophthalmites pro-
vide further support for relative motion of grains
i.e. advanced melting (Fig. I3a). Within any one
0 3 6 9 12 15 18 21 24 27 30 sample, the edges of plagioclase grains are of
approximately the same composition; this must
indicate the imposition of a fixed potential over
2o r b
large distances, presumably in the presence of an
extensive melt. However, core compositions of
I 1 contiguous crystals vary widely; consequently
0 3 6 9 12 15 18 21 24 27 30 strongly normally-zoned crystals can be found
-
3
0- adjacent to strongly reversely-zoned crystals.
L

k l
0 3 6 9 12 15 18 21 24 27 30
30
20 [I
0 3 6 9 12 15 18 21 24 27 30
Grain size, mrn
Fig. 11. Grain size distributions in (a) mesosomes;
(b) mesosomes with some plagioclase porphyroblas-
tesis; (c) ophthalmites; and (d) leucosomes.
according to Busch, Schneider & Mehnert (1974)
this may occur at very low percentages of melt.
As the degree of melting increases, a rheo-
logically critical melt percentage (henceforward
RCMP) is reached at which grains can move
freely without mutual interference; this occurs at
-30% (Arzi, 1978) and will obviously lead to
destruction of the gneissic texture. Ophthalmites
grade petrographically from types with a faint
gneissose foliation defined by mica stands wrap-
ping the feldspar porphyroblasts to wholly
random types. A random texture may be expected
to develop only if there is no mutual interference
of grains, i.e. at melt percentages greater than
RCMP. Thus samples such as 140869 (Table 7)
which show greater deviation from random than
would be predicted are presumably samples
which contained less melt than RCMP. Some of
these ophthalmites e.g. 141 192 show observed/
expected ratios for plagioclase/plagioclase con-
tacts of less than one (Table 7) and have aspect
ratios intermediate between those in mesosomes
and leucosomes. These are interpreted as relict
C
There is n o consistent relation between core
composition and grain size such as could be
attributed to nucleation of successive genera-
I( tions. This suggests that in ophthalmites core
compositions are not related to each other by the
process of migmatization; migmatization was
d only effective in imposing a uniform composition
on grain edges, in this case by crystallization of
dispersed melt. Core composition variations are
thus pre-migmatitic, probably metasedimentary.
The occurrence of crystals from different
metasedimentary layers in close proximity but
not in planar geometry suggests that there was
sufficient melt to disrupt metasedimentary layer-
ing and allow relative motion of crystals. If
several adjacent mesosomes are molten then
transfer across original boundaries is to be
anticipated. Ophthalmite bodies can be found
which greatly exceed the size of the layers from
which they are supposed to form; this, and the
contiguity of crystals inherited from originally
separate layers, indicates that such transfer has
occurred. Indeed the whole melt-crystal mush
appears to have been able to move, as evidenced
by local crosscutting relations (Fig. 6). Thus
ophthalmites retain some traces of the fabric and
chemistry of their precursor gneisses, although
their macroscopic appearance is predominantly
due to mobility resulting from the presence of
melt.
Quantitatively there is little difference in
randomness (x2/v) between leucosomes and
ophthalmites, but leucosomes wholly lack disper-
sed relations. Plagioclase compositions provide
support for higher melt: solid ratios in leuco-
somes than in ophthalmites (Fig. 13b). There is a
clear correlation of core composition with grain
size, such that cores of coarser grains are more
anorthitic than those of smaller grains, whereas
there is little variation in the composition of grain
edges within a leucosome. This presumably
reflects nucleation and growth of successive
258 E. L. McLellan
Fig. 12. Photomicrograph showing development of coarse plagioclase porphyroblast associated with displace-
ment of mica. Scale bar = 2 mm.
generations of successively more albite-rich
plagioclase as a cooling melt crystallizes. There is
no evidence for inherited grains. Thus, in contrast
to ophthalmites, leucosomes are entirely the
products of melt crystallization and contain no
inherited features.
Thus textural analysis indicates that with
increasing melting leucosomes develop via an
ophthalmitic stage from gneissose mesosomes.
The gneissic fabric inherited is progressively
destroyed by the free growth of crystals in a
relatively large volume of melt.
Above the RCMP crystals can move and grow
freely, but there need be no relative, segregative
movement of crystals from melt. Such segre-
gation has occurred at some stage, as evidenced
by the occurrence of discrete leucosomes and
melanosomes and by the removal of inherited
grains from the leucosomes, but its exact
mechanism must remain cause for speculation.
In the absence of deformation, segregation of
melt occurs in response to a hydrostatic pressure
gradient and can be modelled using either Stoke's
Law (at high melt percentages) or Darcy's Law (at
low melt fractions). Melt segregation occurs
relatively easily in the mantle because uniformity
of composition ensures even distribution of melt,
formingcontinuous networks over a considerable
depth of range (i.e. considerable differences in
hydrostatic pressure exist to drive segregation).
The same may not be true of anatexis in highly
heterogeneous crustal sediments where con-
tinuity of melts d a y be restricted to less than one
metre vertical distance. Taking pgncirr = 2.6 g cm
(Wenk & Wenk, 1969) and pme,t=2.4gcm-3
(albite melt; Kushiro, 1978) then Ap is equivalent
to a pressure gradient of 20 bar/km, which for a
melt network lOcm thick is a pressure difference
-
of only 2 x 10- bars. Such stresses will give creep
rates 10-25cm-' and melt segregation in the
absence of deformation will be very limited.
Various authors (Shaw, 1969; Weertman, 1968)
have suggested that shear stress will promote
segregation. This is only true above the RCMP as
below it melt transfer will again be limited by
compensating creep in the solid matrix. No
quantitative data are available on rates of melt
extraction above the RCMP.
Segregation leads t o the formation of a
leucosome with melt contents very much greater
 Contrasting textures in migmatites 259

a .4

-33.7

33.2

30.7 An content, grain centre

32.3 An content, grain edge

-
10 mm

Fig. 13. Comparison of plagioclase composition relations in (a) ophthalmites and (b) leucosomes. Drawn from
photomicrographs.Compositional data obtained on an EDS microprobe (University of Cambridge).
260 E. L. McLellan
than the R C M P and a melanosome with melt
percentage very much less than the RCMP. It is
likely that at least some of the plagioclase crystals
inherited from the mesosome and found in
ophthalmites are segregated along with the mafic
minerals into the melanosome; the presence of
plagioclase in melanosomes has already been
noted. At melt percentages less than the R C M P
further melt extraction requires that there be
creep in the solids and this is likely to be very slow;
further transfer of melt out of the melanosome is
likely to be very inefficient. Thus some melt may
remain in the melanosome and crystallize in siru.
This model therefore predicts similarity of
plagioclase edge composition between melano-
some and leucosome; such similarity has pre-
viously been used as an argument against anatexis
(Misch, 1968; Amit & Eyal, 1976). Once a
melanosome develops it will act as a barrier to any
further movement of material. Thus, in anatectic
systems, neosome scale is fixed at the point at
which melt transfer ceases, i.e. at the point at
which melanosome/leucosome segregation occurs.
4
H
I cm
@ @ plaqiocloseinherited
from paloeosomes
Fig. 14. Schematic evolution of Grampian Group B
migmatites.
1. Original paragneiss. Layers A, B, and C are of
different composition.
2. Incipient melting and plagioclase porphyroblas-
tesis. There is a greater percentage of melt in
layers B than in layers C, but A is unmelted.
3. Melting is occurring syn-D3. Melt percentage is
greater than in 2. The Occurrence of deformation
prompts segregation of leucosomes and
melanosomes.
4. Melting is occurring post-D3. The percentage of
melt has increased to greater than the RCMP and
ophthalmites form in which the meltcrystal mix
has become mobile. Original palaeosome boun-
daries are unrecognizable. Locally mesosomes
may be fractured and these fractures infilled by
meltcrystal mix, resulting in agmatites.
In the Dalradian examples, the size and
chemistry of neosomes indicates that they are
systems of partial melts from more than one
palaeosome; melanosome/leucosome segregation
-
occurred late (i.e. at large system size-some
leucosomes are 5 x mesosome size) o r not at all
(ophthalmites). Structural relations described
earlier support a model of initial syn-D3 melting
in schlieren types, continued melting post-D3
leading to the formation of ophthalmites; both
types crystallized post-D3. The proposed evol-
ution of the Dalradian anatexites is shown
schematically in Fig. 14. In contrast, Johannes &
Gupta (1982) found layer-by-layer trans-
formation of paragenesis into migmatite, i.e.
neosome scale = palaeosome scale; in this case
melanosome/leucosome segregation occurred at
an early stage. If such segregation is indeed due to
deformation, then this apparent difference in the
scale of migmatization is due to a difference in the
tectonic histories of the two regions and is not an
initial feature of the migmatite-forming process.
The geometry and scale of anatectic migmatites is
therefore unpredictable.
COMPARATIVE GEOMETRY OF
ANATECTIC AND METAMORPHIC
MIGMATITES
In the above section a general model for the
physical evolution of a n anatectic migmatite is
described. Key features of this model are:
(i) that segregation into leucosomes and
melanosomes will only occur if mechanical
separation of melt-solid mixtures can occur;
discrete melanosomes and leucosomes may not
form and those which d o will have a very irregular
geometry which will reflect the mechanism of
segregation.
(ii) system size in anatectic leucosomes may be
large or small depending upon the relative timing
of melt formation and of melt segregation (system
deformation).
In contrast, consideration of the mechanics of
mass transfer in metamorphic differen tiation
suggests that such migmatites will invariably
develop regular melanosomes. Metamorphic
differentiation proceeds by transfer of material
from regions of high potential to regions of low
potential, thus leucosomes grow by continued
crystallization at sites of low potential coupled
with passive displacement of adjacent material
(micas, etc.). Thus metamorphic melanosomes
are formed essentially passively by chemical
extraction of felsic components; their formation
is a necessary consequence of migmatization. In
 Contrasting textures in migmatites
contrast with Grampian anatectic melanosomes,
Grampian metamorphic melanosomes are well-
developed and have good foliation. They are
comparable in grain size and grain shape to
mesosomes (cf. anatectic melanosomes) and
grade into them by an increase in the amount of
felsic material. This gradational nature is attri-
buted to passive extraction of leucosome
material.
A small-scale geometry is a necessary conse-
quence of migmatization by metamorphic
differentiation. As the thickness of melanosome,
across which leucosome components must be
transported increases, the potential gradient
driving such transport falls and diffusion is
consequently slowed. Thus metamorphic
differentiation is a self-limiting process and will
always generate small-scale structwes; no such
predictions can be made for the geometry of
anatectic systems.
Outside the migmatite area metasedimentary
-
layers averaged 1 cm in thickness; within the
metamorphic migmatites the distance from the
centre of one leucosome to the centre of the
adjacent leucosome is very variable but averages
5 1 cm. This comparability indicates that in the
metamorphic migmatites melanosome/leucosome
segregation occurs on a scale which is the same as
that of the palaeosome. Stromatic metamorphic
migmatites are not mdre heterogeneous than their
parent rocks, and this is in contrast to anatectic
migmatites. Thus, in the Grampian example,
system size in the metamorphic migmatites is
very much smaller than that in the anatectic
migmatites.
CONCLUSIONS
Quantitative textural analysis of a migmatite can
be of great value in interpreting the formation of a
migmatite provided that post-migmatite cooling
was relatively rapid and that there was little or no
post-migmatite deformation. For post-tectonic
migmatites the method can potentially dis-
criminate melt-present from melt-absent leuco-
somes. Partial melting of a mesosome leads
initially to feldspar porphyroblastesis which
begins when a grain boundary partial melt
network is established; as melt percentage in-
creases the inherited fabric is progressively
destroyed and inherited grains are overprinted by
the melt. Above -30% melt random textures
develop and the melt-crystal mixture may become
mobile. The separation of a melt-crystal mixture
into leucosome and melanosome probably re-
quires deformation. Inherited plagioclase together
26 1
with mafics are displaced to the melanosome
and the neosome system size becomes fixed. In
metamorphic systems system size is necessarily
small, and melanosome formation results from
chemical extraction of felsics with passive displace-
ment of mafics in front of a growing leucosome.
ACKNOWLEDGEMENTS
The research upon which this paper is based was
carried out in the Department of Earth Sciences,
Cambridge, and technical support received there
is gratefully acknowledged. Research was sup-
ported by a Research Studentship from the
Natural Environment Research Council. The
paper was written during tenure of a NATO/
NERC Overseas Research Fellowship in the
Department of Geology and Geophysics, Yale
University. I also wish to express my thanks to all
those of my colleagues who took time to criticize
my ideas, in particular Dr J.R. Ashworth who
first introduced me to the techniques used. Any
inaccuracies of interpretation or presentation are,
of course, my own.
REFERENCES
Amit, 0. & Eyal, Y., 1976. The genesis of Wadi
Magrish migmatites (N.E. Sinai). Contrib. Mineral.
Petrol., 59, 95-1 10.
Arzi, A. A,, 1978. Critical phenomena in the rheology of
partially melted rocks. Tectonophysics, 44, 173-1 84.
Ashworth, J. R., 1976. Petrogenesis of migmatites in
the Huntly-Portsoy area, northeast Scotland.
Mineralog. Mag., 40,661-682.
Ashworth, J. R., 1979. Textural and mineralogical
evolution of migmatites. In The Caledonides of the
British Isles-Reviewed (ed. Harris, A. L., Holland,
C. H. & Leake, 9. E.). Spec. Publ. geol. SOC.London,
8, 357-361.
Bradbury, H. J., Smith, R.A. & Hams, A. L., 1976.
‘Older Granites’ as time-markers in Dalradian
evolution. J. geol. SOC. London, 132, 677-684.
Brewer, M. S., Brook, M. & Powell, D., 1979. Datingof
the tectono-metamorphic history of the southwestern
Moine, Scotland. In The Caledonides of the British
Isles-Reviewed (ed. Harris, A. L., Holland, C. H. &
Leake, B.E.). Spec. Publ. geol. Soc. London, 8,
129-137.
Bush, W., Schneider, G . & Mehnert, K.R., 1974.
Initial melting at grain boundaries, Part 11: Melting
in rocks of granodiorite, quartzodioritic and tonalitic
composition. Neues Jahrb. Miner. Mh., 1974,
345-370.
Dowty, E., 1980. Crystal growth and nucleation theory
and the numerical simulation of igneous crystal-
lisation. In Physics of Magmatic Processes (ed.
Hargraves, R. 9.). pp. 419-485. Princeton University
Press, Princeton.
262 E. L. McLellan
Flinn, D., 1969. Grain contacts in crystalline rocks.
Lithos, 3, 361-370.
Johannes, W. & Gupta, L.N., 1982. Origin and
evolution of a migmatite. Contrib. Mineral. Petrol.,
79, 114-123.
Johnstone, G. S., 1966. The Grampian Highlands.
British Regional Geology (3rd edition), H.M.S.O.,
London.
Karlsson, G. & Wahlgren, C.H., 1982. A statistical
investigation of grain contacts in a migmatite. Neues
Jahrb. Miner. Mh., 1982, 348-360.
Kretz, R., 1969. On the spatial distribution ofcrystals in
rocks. Lirhos, 2, 3 9 4 6 .
Krogh, T. E. & Davis, G . L., 1973. The eflect of regional
metamorphism on U-Pb systems in zircon and a
comparison with Rb-Sr systems in the same whole
rock and its constituent minerals. Carnegie Inst.
Wash. Yb., 1972, 601-610.
Kushiro, I., 1978. Viscosity and structural changes of
albite (NaAISi,O,) melt at high pressures. Earth
planet. Sci. h t t . , 41, 87-90.
McLellan, E. L., 1982a. Barrouian migmatites and the
rhermal history o j the Eastern Grampians. Unpubl.
Ph. D. thesis, Cambridge.
McLellan, E. L., 1982b. Problems of structural analysis
in migmatite terrains. In Migmatires, Melting and
Metamorphism (ed. Atherton, M. P. & Gribble,
C. D.), pp. 299-302. Shiva Publishing, Nantwich,
England.
Misch, P., 1968. Plagioclase compositions and non-
anatectic origin of migmatitic gneisses in the
Northern Cascades of Washington State. Contrib.
Mineral. Petrol., 17, 1-70.
Oldershaw, W., 1974. The Lochnagar granitic ring
complex, Aberdeenshire. Scott. J. Geol., 10, 297-
309.
Received 17 Ma!* 1983: revision accepted 28 Julj. 1983.
Pitcher, W.S., 1979. The nature, ascent and emplace-
ment of granitic magmas. J. geol. SOC.London, 136,
627-662.
Pitcher, W.S. & Berger, A.R., 1972. The Geology OJ
Donegal: a Study OJ Granite Emplacement and
Cnroyfing. 435 pp. Wiley-Interscience. London.
Ramsay, D. M. & Sturt, B.A., 1979. The status of the
Banff Nappe. In The Caledonides OJ the Brirish
Isles--Reuiewed(ed. Harris, A. L., Holland, C. H. &
Leake, B.E.). Spec. Publ. Geol. Soc. London, 8,
145-1 5 1.
Read, H. H., 1955. The Banff Nappe: an interpretation
of the structure of the Dalradian rocks of north-east
Scotland. Proc. Geol. Assoc., 66, 1-29.
Shaw, H.R., 1969. Rheology of basalt in the melting
range. J. Petrol., 10, 510-535.
Sturt, B.A., Ramsay, D. M., Pringle, I. R. & Teggin,
D. E., 1977. Precambrian gneisses in the Dalradian
sequence of northeast Scotland. J. geol. SOC.London,
134,4144.
Vernon, R. H., 1975. Metamorphic Processes:
Reoctions cmcl Microstructure, Developmen/. 247 pp.
Wiley. New York.
Weertman, J., 1968. Coalescence of magma pockets
into large pools in the upper mantle. Bull. geol. SOC.
Am., 83, 3531-3532.
Wenk, E. & Wenk, H.-R., 1969. Physical constants of
Alpine rocks (density, porosity, specific heat, thermal
diffusivity and conductivity). Schweiz. Min. Petrog.
Mill., 49, 343-357.
Whitten, E.H. & Dacey, M.F., 1974. On the
significance of certain Markovian features of granite
textures. J. Petrol., 16, 429-453.
Whitten, E. H., Dacey, M. F. & Thompson, K.D.,
1975. Markovian grain relationships of a Grenville
granulite. Am. J. Sci., 275, 1164-1 182.