The Importance of Residual Source Material (Restite) in Granite Petrogenesis

The Importance of Residual Source Material (Restite)
in Granite Petrogenesis
by B. W. CHAPPELL, 1 A. J. R. WHITE 2 AND D. WYBORN 3

1
Department of Geology, The Australian National University, Canberra, ACT 2601 Australia
2
Department of Geology, La Trobe University, Bundoora, Victoria 3083 Australia
3
Bureau of Mineral Resources, Canberra, ACT 2601 Australia
(Received 13 November 1985; revised typescript accepted 1 June 1987)
Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

ABSTRACT
There is a great deal of evidence to support the hypothesis that many granites and related volcanic
rocks contain significant amounts of crystals that did not crystallize from a melt. These crystals are
residual from the partial melting of the source rocks (primary restite) or result from the recrystallization
of such residual crystals (secondary restite). Evidence for the presence of restite includes the occurrence
of some rock inclusions that were relatively refractory fragments of the source, and the calcic
plagioclase cores of comparatively uniform composition which occur in many granites. The presence of
an 'old' zircon component in some I-type granites is best explained as being inherited from the source.
Relatively mafic restite-bearing granite magmas can form at geologically reasonable temperatures in
the crust and hence rocks of such composition do not have to be regarded as cumulates. The presence of
more MgO in moderately felsic porphyritic volcanic rocks than can occur in liquids of that bulk
composition at the known temperature of eruption, is best explained by the proposition that the
phenocrysts are restite phases, rather than by the commonly held view that they precipitated from a
melt at a substantially higher temperature. Some plutonic and volcanic rock suites of crustal derivation
are chemically equivalent, but have a composition too mafic to represent melts at crustal temperatures;
this is compelling evidence that those rocks contain a substantial restite component and it shows that
the plutonic rocks are not crystal cumulates. A significant restite component is an elegant way of
overcoming the difficulties of nucleating crystals in unzoned porphyritic granite plutons. Likewise, we
contend that many phenocrysts in volcanic rocks are restite minerals, rather than the result of
nucleation at slow rates of cooling.
Clearly there are granites that are more mafic and restite-free which have formed by the classical
process of precipitation from a melt. However, examples of such granites are rare in the Lachlan Fold
Belt of southeastern Australia. One such case is the Boggy Plain Supersuite and its associated felsic
volcanic rocks. We compare this supersuite with two examples of the more typical restite-bearing case,
the Bullenbalong Supersuite (S-type) and the Glenbog Suite (I-type), and their related volcanic rocks.
Such a comparison shows that there are distinctive textural characteristics of granite suites containing
restite and that the chemical variation trends throughout such suites are much more regular than those
of restite-free suites.
Partial melting in the crust of any rocks containing quartz, feldspars, and H 2 O, either free or
provided by the decomposition of one or more hydrous phases, produces a granitic melt in equilibrium
with restite. Because of the high viscosity of felsic mejts, the meltphasc.might-not-be-extr-acted to form a
body of ^x>mpletely molten" magma. Instead, when the amount of melt becomes sufficiently high, the
rigid restite framework is broken and the whole mass becomes a crystal-rich magma. Because of its
relative buoyancy this magma may move up to intrude as a pluton, or to erupt at the surface. As a result
of the high viscosity and yield strength of the melt, the separation of restite is inhibited, and much of it
may remain to form part of the solidified granite or volcanic rock. According to the restite model, the
composition of a rock formed in this way will be a function of the composition of the melt and restite,
and of the degree of separation between the two. Under favourable circumstances, granite suites with
compositions resulting from restite separation can be used to estimate the compositions of their source
rocks and in this way they provide a window through which we can view the composition of the deeper
crust. In the Lachlan Fold Belt, this has led t o the recognition of the I-type granites derived from older
[Joitmal of Petrology, VoL 28, Part 6. pp 1111 1138. 1987] © Oxford Urmeraly P r e n 1987
1112 B. W CHAPPELL ET AL.
igneous rocks and the S-typc granites produced from sedimentary source rocks that had a composition
influenced by a previous episode of weathering. Within each of these two major groups there are
numerous suites whose compositions reflect compositional variations of the source rocks on a finer
scale.
INTRODUCTION
In this paper we review the evidence for residual source rock material or restite being an
important component of many granite magmas. We discuss the restite model that was
introduced by White & Chappell (1977) to explain the systematic variation in composition of
granites produced from such magmas. The implication of this model for deducing the
composition of the source rocks and hence the nature of the deep crust, is also considered.
Granites of Silurian to Devonian age (Caledonian) are extensively developed in the
Lachlan Fold Belt (LFB) of southeastern Australia, with some Carboniferous occurrences in

the northeastern part of the belt. Granites comprise 61 000 km 2 of a total exposed area of a
little under 300000 km 2 in the LFB (White & Chappell, 1983; Chappell, 1984). Equivalent
volcanic rocks are an important component of the Palaeozoic sequences. The regional
distribution of granites in the eastern part of the LFB is shown in Fig. 1 of White & Chappell
(1983). Wyborn et al. (1981) have also shown the distribution of related volcanic rocks in that
area.
A few of the granites in the LFB are felsic (~ 75 per cent SiO2) and show chemical evidence,
mainly in trace element contents (e.g., very low Sr) of fractional crystallization; more
commonly, other equally felsic granites are thought to represent unfractionated melts with a
composition similar to those formed at minimum melting temperatures. Some of the granites
that are a little more mafic (70-75 per cent SiO2) were formed from unfractionated or weakly
fractionated melts, at slightly higher temperatures, whereas others are thought to have been
produced from a felsic melt containing a small amount of mafic restite. Excluding the
uncommon ( < 1 per cent) and generally very felsic 'A-types' (Collins et al., 1982), the average
SiO 2 content of 848 granites from the LFB is 68-4 per cent (White & Chappell, 1983) and 65
per cent of these granites have SiO 2 contents below 70 per cent. The latter more mafic
granites generally contain restite and have compositions reflecting the degree of separation of
more mafic restite from more felsic melt. More mafic granites derived from a completely
molten magma are rare, but not unknown in the belt. The Boggy Plain pluton of the
Kosciusko Batholith (Wyborn, 1983) is an example of such a body; this is part of the Boggy
Plain Supersuite (Wyborn et al, 1987) in which the mafic members are thought to be
cumulates from a more felsic melt. Much of the evidence for the presence of restite in many
granites and for the restite model is petrographic and chemical and we will assess this
evidence in part by comparing the bulk of the relatively more mafic granites of the LFB with
the less abundant rocks of the Boggy Plain Supersuite.
It is generally accepted that granite magmas are produced by partial melting, usually of the
crust. The product of partial melting is a hydrous melt plus often largely anhydrous restite.
Together these constitute a magma if enough melt is present for the whole to be fluid or
mobile and under these conditions the melt will not be water-saturated. Those magmas that
are completely liquid or restite-free must leave their residue from partial melting at or near
the source as they ascend through the crust. This case seems to be unusual in the LFB, where
more commonly the viscous granite melt carries restite, mostly in the form of single crystals
or small aggregates of crystals. We stress that it is restite of this type which controls the
variation in magma composition according to the restite model. Certain inclusions are
thought to be residues from the source but these are considered to play a minor role in
changing the magma composition.
RESTITE IN GRANITE PETROGENESIS 1113
THE RESTITE MODEL

Historical development of the restite model
Mica-rich inclusions described from granites of the Porto area of Portugal were called
'surmicaceous' by Lacroix (1933). In those granites that have virtually not moved from their
region of origin by partial melting (autochthonous granites), surmicaceous inclusions are
commonly accepted as restite (Didier, 1973).
Bateman et al. (1963) suggested that the mafic hornblende-bearing inclusions in certain
granites of the Sierra Nevada could be "refractory material that was not melted when the
magma was formed". Likewise, van Moort (1966) suggested 'restite' or residue from partial
melting as one possible origin for the mafic inclusions from the Massif du Mont Lozere,
France. The presence of residual material in granites was also considered by Piwinskii &
Wyllie (1968) as a possible explanation of certain experimental observations (see below). It

was first suggested by Chappell (1966) that variation in the amount of refractory source
material could be responsible for the systematic linear chemical variation within certain
granite suites of the New England Batholith of eastern Australia. This suggestion formed the
basis of the restite model presented by White & Chappell (1977) to explain the variation in
many granite suites. The restite model has never become widely accepted, and recently it has
been strongly criticized (e.g., Gemens et al, 1983; Vernon, 1983). It is therefore appropriate at
this time to present the model in some detail, to examine the evidence that can be used to
support it, and to look at its broader implications in granite petrogenesis.
The restite model

White & Chappell (1977) in proposing the restite model pointed out that partial melting in
the crust of any rocks containing quartz, feldspars, and H 2 O, either free or provided by the
decomposition of a hydrous phase, must produce a granitic melt phase in equilibrium with
residual phases (restite). Because the viscosity of the melt is high, the melt phase is not readily
extracted to form a body of completely molten magma. Instead, when the amount of melt
becomes sufficiently high and above the "rheological critical melt percentage" (Arzi, 1978) or
the "critical melt fraction" of van der Molen & Paterson (1979), the rigid restite framework
will be broken and the whole mass will become a crystal-rich or inclusion-rich magma with
lower effective viscosity. Relative to solid source rocks of the same composition, the magma
will be less dense, and its intrinsic buoyancy means that the whole mass may move upwards
to intrude as a pluton, or to erupt. Because of the high viscosity and yield strength of granite
melts, the separation of restite is inhibited and in most, but not all granites some restite
remains and forms part of the solidified granite or volcanic rock (Chappell, 1984). The
progressive separation of melt and restite that does occur during movement of the magma
accounts for the very systematic linear chemical variations found in granites from both single
plutons and groups of plutons, or suites. The process of differentiation amongst members of a
suite is, according to this model, an unmixing process in which the melt phase of the magma
progressively clears itself of restite but does not itself change its composition appreciably.
SYSTEMATIC VARIATION AMONG SOUTHEASTERN AUSTRALIAN

GRANITE AND ASSOCIATED VOLCANIC SUITES
The systematic variation that occurs amongst igneous rocks was well established a century
ago when Judd (1886) introduced the term 'petrographic province' to show that rocks in
different regions have different chemical, mineralogical and petrographic characteristics. To
stress that time as well as space is important to an understanding of the systematics of igneous
rocks, Bowen (1928) introduced the term 'rock association' for rocks associated in the field
and of the same age. The rocks of an association have variable composition, but the variation
is systematic. Recurring associations in comparable environments elsewhere in the world
show that similar genetic processes have prevailed to give rise to rocks of an association e.g.,
the calc-alkaline association. Today, with more precise chemical data, and in many cases
with isotopic control, we can see fine-structure within the variation of associated rocks.
Rocks of a particular association from a particular province e.g., the granites of the LFB, may
be grouped into many petrographic, chemical and genetic units that we call suites. Each suite
shows systematic petrographic and chemical variations as its composition changes. The
granites and related volcanic rocks of the LFB can be subdivided in this way into suites that
are thought to represent different specific source rock compositions. Some suites show

similarities in composition but are different in detail, and we refer to these as supersuites. The
first-order subdivision of these suites and supersuites is into those granites derived from
igneous and sedimentary source rocks, the I- and S-types (Chappell & White, 1974,1984), but
within each of these groups there is a considerable degree of fine-structure, corresponding to
more subtle differences in source rock compositions. Recent work on the Silurian-
Devonian volcanic rocks of the LFB has shown that these can also be divided into suites
based on mineralogical, chemical and field characteristics. Of particular interest is the fact
that many of these volcanic rock suites can be matched closely with plutonic equivalents
(Wyborn et al., 1981; Barron et al., 1982). The volcanic rocks are frequently altered and it is
more difficult to collect fresh samples but the study of carefully selected samples is helpful in
understanding the mineralogy, textures and genesis of the related plutonic rocks. We will
here discuss three groups of granites and their related volcanic rocks, which we will use later
in evaluating the restite model. These are the Boggy Plain Supersuite and Glenbog Suite,
both I-types, and the S-type Bullenbalong Supersuite.
Boggy Plain Supersuite

This group of rocks has recently been described by Wyborn et al. (1987). The supersuite
consists of more than twenty-five fairly regularly spaced plutonic complexes with a total area
of approximately 2300 km 2 , plus associated felsic volcanics, extending meridionally over a
distance of a little more than 500 km in the LFB. The name derives from the Boggy Plain
pluton, a small (36 km 2 ) body in the Kosciusko Batholith, studied in detail by Wyborn (1983).
Plutons in this suite are often strongly zoned, from gabbro to aplitic granite and are
chemically distinctive, with high Cu, K, Ba, Sr, Rb, Th, U, and light REE relative to other
granite suites in the LFB.
Wyborn et al. (1987) proposed that the primary magmas in the Boggy Plain Supersuite
were liquids of probable potassic andesite composition containing no restite, and that
subsequent compositional variation was controlled by fractional crystallization. Part of the
evidence to support this proposal includes:
(1) Common systematic chemical variation from high temperature (>1000°C) mafic
marginal rocks to lower temperature more felsic rocks in the cores of plutons.
(2) Occurrence of cumulate rocks with varying relative abundances of pyroxenes,
plagioclase and trapped interstitial liquid, resulting in scatter on variation diagrams.
(3) Scarcity of inclusions in the most mafic rocks and absence of any inclusions that could
be regarded as restite.
(4) Absence of plagioclase cores that could be restite (see discussion below).
(5) Curved variation trends for trace elements such as Rb, Th, U, and Ta.
(6) Sharp changes in trends for Ba and Zr related to the incoming of a new liquidus phase
(biotite, K-feldspar, zircon) (see Fig. 3).
(7) Presence of W-Nb-Ta-Bi mineralization in the most fractionated rocks due to a
building up of these elements during prolonged fractional crystallization.
(8) The associated volcanic rocks (Mountain Creek Volcanics, Owen & Wyborn, 1979) are
crystal-poor and correspond in composition to the fractionated liquid lost during the
crystallization of the plutonic cumulates.
The source-rocks for the Boggy Plain Supersuite were thought by Wyborn et al. (1987) to
be gabbros near the base of the crust. The high temperatures reached during partial melting
of that region would be consistent with the known high geothermal gradient in the LFB at
that time. Restite was not incorporated into Boggy Plain Supersuite primary liquids because

of their low viscosities, a result of their high temperature and relatively mafic composition
(high-K andesite). This contrasts with most other granite suites in the LFB which were
derived from more felsic sources at lower temperatures and shallower depths which led to the
production of restite-retentive high viscosity liquids.
Glenbog Suite
This suite is an important component in the Bega Batholith in the eastern part of the LFB
(Beams et al., 1987). The batholith has an outcrop area of 8600 km 2 and consists of more than
130 separate plutons, which can be grouped into many suites. The Glenbog Suite is the most
extensive of these and consists of 12 separate plutons covering an area of 1510 km 2 in a linear
belt 265 km in length along the western edge of the batholith (Fig. 1).
The Glenbog Suite is distinguished from others in the Bega Batholith by its low Na and Sr.
SiO 2 ranges from 6552 to 7379 per cent in 41 analysed samples, and Sr from 153 to
98 p.p.m., in the same two end-member rocks. There is a range of composition for some
individual intrusions, with the variation within an intrusion always parallel to that of the
suite as a whole on a Harker diagram (Fig. 1). The Sr variation relative to SiO 2 is remarkably
linear and there is no systematic spatial variation as both low and high SiO 2 rocks are
scattered along the length of the belt (Fig. 1).
Modally, rocks of the Glenbog Suite range from rare tonalites, through dominant
granodiorites to adamellites (monzogranites), all of which are characterized petrographically
by: (a) prominent, approximately spherical, quartz grains commonly 5 to 10 mm across and
in some cases surrounded by rims consisting of small hornblende crystals; (b) mafic minerals
including hornblende and cummingtonite occurring as ragged crystal aggregates which can
be called clots, that become less abundant in the more felsic rocks; (c) conspicuous subhedral
hornblende prisms which are most abundant in the more mafic rocks; (d) occasional relict
cores of clinopyroxene within hornblende; (e) rare orthopyroxene crystals partly altered to
cummingtonite and surrounded by biotite crystals or in mafic clots surrounded by both
brotite arid~rnagri*etite; and~(f) zoned plagioclase crystals with large cores up to 4 mm but
commonly 2 mm across that are uniform in composition (An 49 _ 34 ) irrespective of whether
the rock is mafic or felsic, although they become less abundant in the more felsic members of
the suite. In rare examples, plagioclase cores may contain inclusions of clinopyroxene and
more rarely quartz; occasionally the reverse is found and plagioclase crystals are included in
the large quartz grains. Accessory minerals are magnetite, ilmenite, allanite, zircon, and
apatite, all five minerals commonly associated with mafic clots; apatite mainly appears as
small prisms in the mafic crystals.
1116 B. W. CHAPPELL ET AL.

65 -
200 100
ppm Sr
FIG. 1. Sketch of twelve plutons of the Glenbog Suite (dark) in the Bega Batholith (crosses) in the Lachlan Fold Belt
of southeastern Australia. On the left is a Harker diagram for Sr in forty one samples of the Glenbog Suite The
ranges in SiO 2 for the different plutons are indicated; these show that both mafic and felsic plutons are distributed
throughout the length of the suite.
Kadoona Dacite—an equivalent volcanic suite

The Kadoona Dacite Member of the Long Flat Flat Volcanics occurs near the northern
end of the Glenbog Suite and it matches fairly closely the distinctive chemistry of that suite
(Wyborn & Chappell, 1986; Wyborn & Owen, 1986). Chemical differences between the
Kadoona Dacite and the Glenbog Suite are small with slightly higher contents of Sr, Ba, Zr,
and Ga in the dacite. The dacite also has a restricted range in SiO 2 content, 64-67 per cent,
near the lower end of the range of SiO 2 values of the Glenbog Suite. A typical sample of
Kadoona Dacite consists of 60 per cent phenocrysts in a fine-grained felsic groundmass with
poorly-preserved eutaxitic structure. Conspicuous are large (commonly 5 mm across)
embayed irregularly shaped quartz crystals similar to the large quartz grains in the Glenbog
Suite rocks. Plagioclase occurs as rectangular crystals commonly 3 mm in their longest
dimension, with a uniform composition (An 48 _ 5J ) except for the rare presence of narrow
inconspicuous sodic rims. The ferromagnesian phases are dominantly orthopyroxene and
pale green to colourless clinopyroxenes commonly with perfect crystal shapes, and lesser
amounts of red-brown biotite. There is only a small amount of hornblende (<x = pale yellow-
brown, ft = brown, y = greenish-brown); both of these hydrous mafic minerals are surrounded
by rims consisting of aggregates of tiny magnetites. Accessory minerals include magnetite,
ilmenite, allanite, apatite, and zircon. Clots (micro-inclusions), consisting of clinopyroxene,
orthopyroxene, plagioclase, ilmenite, and magnetite, are present. Application of geo-
thermometry using two pyroxenes, magnetite-ilmenite and Ti content of biotite all indicate
eruption temperatures around 85O-9OO°C (Wyborn & ChappelL, 1986). Experimental studies
on the Kadoona Dacite (Rushmer et al, 1985) indicate that high temperatures are required
(100O-1100°C) for the plagioclase and pyroxenes to melt These experiments therefore support
a restite origin for such phenocrysts.
Crystallization history of the Glenbog Suite

The parallel study of matching plutonic and volcanic rocks enables us to make several
deductions concerning the crystallization history of the granites of the Glenbog Suite:
(1) The abundance of large crystals (60 per cent by volume) of uniform composition set in a
felsic fine-grained groundmass (Kadoona Dacite) indicates that clinopyroxene, ortho-
pyroxene, plagioclase (An50), and quartz were liquidus phases that co-existed with a near-
haplogranite melt, in agreement with the experimental data of Rushmer et al. (1985).
(2) Since the general consanguinity of the Kadoona Dacite and the Glenbog Suite is
established by their chemistry, and since pyroxenes are the dominant (~75 per cent)
ferromagnesian phase in the dacite, most amphibole and biotite in the plutonic rocks can be
presumed to result from the reaction of pyroxenes with the hydrous felsic melt during slow
cooling. This explains the occurrence of relict clinopyroxenes within the hornblendes and rare
relict orthopyroxenes associated with cummingtonite and biotite in the granites. Even the
subhedral hornblendes may not be the result of direct crystallization from the melt phase; they
could be the result of reaction of early pyroxenes with the melL
(3) The clots of pyroxenes, plagioclase, and magnetite seen in the dacite have presumably
reacted with the felsic melt to produce the clots consisting of ragged amphiboles and biotite
present in the granites.
(4) The uniform and large rectangular-shaped plagioclase crystals in the dacite are identical
in composition, size and overall shape with the cores of the plagioclase in the granites, showing
that reaction between the early plagioclase and felsic melt (continuous reaction of Bowen, 1928)
was virtually non-existent. This is consistent with the views of Johannes (1978) on the sluggish
nature of plagioclase homogenization below 1000°C; this will be discussed below. On the
other hand, the zoned outer rims of the plagioclase in the Glenbog Suite plutonic rocks must be
the product of magmatic crystallization within the magma chambers now represented by the
various plutons.
Bullenbalong Supersuite
The LFB south-west of Canberra contains an important component of mafic and generally
cordierite-rich S-type granites. The Bullenbalong Supersuite is widespread, comprising several
suites of which the most abundant is the Bullenbalong Suite (3100 km2). The most mafic of the
three S-type volcanic suites near Canberra discussed by Wyborn et al. (1981), the Hawkins
Volcanics, is also part of this supersuite. These rocks are exclusively of crustal origia The
possibility that the more mafic rocks might have resulted from the mixing of felsic S-type
crustal melts with melts from deeper in the crust or mantle is excluded because the more mafic
rocks are progressively more peraluminous.
Analyses of 144 granites of the Bullenbalong Supersuite range from moderately felsic to
rather mafic with a range in total FeO from 1 -95 to 5-47 per cent (total FeO is a better indicator
than SiO 2 of compositional spread in S-type granites since mafic granites of this type have a
high quartz content and therefore are high in SiO2 relative to mafic I-type granites).
Petrographically the rocks are granodiorites and tonalites dominated by quartz, plagioclase,
and biotite, with minor K-feldspar, and often altered cordierite. The Hawkins Volcanics are
ash-flow tuffs and lavas that are chemically very similar to the Bullenbalong Supersuite and
comprise porphyritic dacite with about 60 per cent phenocrysts consisting of quartz (30 per
cent), plagioclase (35 per cent), biotite (15 per cent), cordierite (10 per cent), and hypersthene (10
per cent), with almandine garnet as a common accessory phase. The phenocrysts preserve
equilibria established at 500-600 M Pa (5-6 kb) and 800°C (Wyborn et a!., 1981). Broad
plagioclase cores in the granites are similar in composition to weakly zoned (An J0 _ 55 )
phenocrysts in the volcanic rocks, so that the same relations hold between the granites and

volcanic rocks as described above for the Glenbog Suite and Kadoona Dacite.
Rocks of the Hawkins Volcanics provide very compelling evidence that the more mafic rocks
of the Bullenbalong Supersuite have retained most of their restite. At 800cC magmas with the
composition of the Hawkins Volcanics (SiO2 = 67-3 per cent, MgO = 236 per cent, M^-number
= 48) are 60 per cent crystallized at moderate crustal pressures, as indicated by the
equilibration pressures and phenocryst abundance in the rocks (Wyborn et ai, 1981; Wyborn
& Chappell, 1986). The liquidus of such a rock with only 3-5 per cent H 2 O is likely to be
> 1000°C at 200-700 MPa. Such an estimate is based on a comparison with the experimental
study of an S-type granite composition by Qemens & Wall (1981). The composition that they
studied was more felsic and less magnesian (SiO2 = 70-5 per cent, M g O = 1-54 per cent, Mg-
number = 44) than average Hawkins Volcanics and with 4 per cent H 2 O had a liquidus
temperature of about 925 °C at moderate pressures. Thus it can be seen that for the generation
of complete melts with compositions of the Hawkins Volcanics and, by inference, of the
compositionally identical relatively mafic granites of the Bullenbalong Supersuite, much of the
crust of the LFB would have to be heated to over 1000°C at only moderate pressures. Such a
high temperature is unreasonable and at odds with the geothermal gradients established in
metamorphic rocks and with the vast volumes of S-type magma that were generated in the LFB
(White & Chappell, 1983). A more reasonable model involves heating the source rocks of the
Hawkins Volcanics and the Bullenbalong Supersuite to 800°C and retaining restite in the
magma.
The restite model is equally applicable to the S-type Violet Town Volcanics recently studied
by Qemens & Wall (1984). Mafic minerals in the Violet Town Volcanics are less magnesian than
those in the Hawkins Volcanics e.g., cordierite Mg-number = 58 compared with 65, biotite Mg-
number=40 compared with 55, orthopyroxene Mgr-number = 40 compared with 47, sugges-
ting equilibration with a melt at lower temperatures. We suggest that a magmatic temperature
of a little less than 800 °C is applicable to the Violet Town Volcanics. This also matches
temperatures that we have determined for the most titaniferous Violet Town biotites given in
Clemens & Wall (1984) using the geothermometer formulated by Qirmichael (Luhr et al.,
1984). We also note that the considerably higher temperatures determined by Clemens & Wall
(1984) using the garnet-biotite thermometer recalculate to much lower temperatures when the
TiO 2 content of the biotite is taken into consideration (Indares & Martignole, 1985). A
temperature of a little below 800°C is well below the liquidus of the S-type granite composition
studied by Qemens & Wall (1981) with 4 per cent H 2 O and at moderate pressures, and the
average Violet Town rhyodacite is more mafic (SiO2 = 67-65 per cent) than the experimental
composition. In addition, application of the garnet-orthopyroxene barometer of Harley &
Green (1982) to the analyses of these minerals given by Clemens & Wall (1984) indicates
relatively high /Mow T conditions rather than high T-low P conditions of equilibration
advocated by Clemens & Wall (1984). We therefore consider that a restite origin for much of the
phenocryst volume in the Violet Town Volcanics is consistent with the experimental
observations.
MECHANISMS FOR PRODUCING VARIATION IN GRANITES

The restite model provides a mechanism to account for the variation in composition in
many granite suites. We do not suggest that it is the only cause of such variation. Certainly in
the LFB it appears to have been dominant, and its possible application should be evaluated
elsewhere. Two main alternative mechanisms are suggested at the present time to produce
variation. These are first, fractional crystallization of a mafic parent magma (e.g., Bateman &

Chappell, 1979; McCarthy & Groves, 1979), and second, the mixing of mafic and felsic
magmas (e.g., Reid et al., 1983).
We regard fractional crystallization as a complementary mechanism to restite separation
in producing granite variation, and that process can operate in two ways. In the first case,
relatively mafic high-temperature melts may rid themselves of restite at or near the magma
source and produce a range in rock compositions by fractional crystallization e.g., the Boggy
Plain Supersuite. Second, after restite has been removed from a more felsic melt,
fractionation of that melt may occur, this sometimes happens in the LFB. For example,
McCarthy & Groves (1979) ascribed the variation in granites of the Blue Tier Batholith in the
southern part of the LFB to fractional crystallization of a single parent magma. We consider
that the situation in the Blue Tier is more complex, with a variety of sources producing a
range of more mafic granites in which the dominant cause of variation is restite separation,
and small amounts of very felsic granites within which group variation resulted from
fractional crystallization.
Some of the features of granites that we use to argue for their compositions being
controlled by varying degrees of separation of melt and restite, might also result from the
mixing of mafic and felsic magmas. However, we regard the production of a range of granite
magma compositions by mixing of mafic and felsic material, at least on a wide scale, as
unlikely (White & Chappell, 1985). In the LFB, there are at least ten I-type lithological units,
most of which are probably single plutons, that have an exposed area in excess of 500 km 2
(the largest is 970 km2). Fairly uniform mixing of volumes in excess of 1000 km 3 would be
required in such suites. Even if such large scale mixing were physically possible, observations
that we have made on granites of the LFB would argue against its application in that region,
as follows:
(1) The variation lines for different suites would pass through different end-members at
both the mafic and felsic ends, if all elements are taken into consideration, so that basalt-
granite mixing would involve both different granites and different basalts.
(2) In the case of S-type granite suites, the basalt end-member would have to be
peraluminous since such suites become more peraluminous as they become more mafic.
(3) In some I-type granite suites, e.g., Jindabyne (Hine et al., 1978), elements such as Ni and
Cr are too low in abundance if the linear trend is extrapolated to basaltic compositions.
These data would be consistent, however, with an andesite or basaltic andesite as one end-
member.
(4) The plagioclase cores from many I-type suites are too calcic to be early phenocrysts
from most basalts.
EVIDENCE FOR THE RESTITE MODEL

We now present the evidence in support of the restite model.
Restite inclusions
Granites occurring in a regional-aureole environment may contain large numbers of
refractory metasedimentary inclusions that are identical to the melanosomes in adjacent
migmatites e.g., in the biotite granite of the Trois Seigneurs Massif (Wickham, 1987). Most of
the granites in the LFB have moved away from their source regions and intruded as large
plutons at much higher levels in the crust and these also contain some inclusions that are
unequivocally residual from the source e.g., the cordierite + sillimanite± garnet gneiss
inclusions in some S-type granites (White et al., 1977). These inclusions cannot be matched in
any exposed wall rocks and do not occur in adjacent inclusion-rich I-type granites. Some

restite is therefore certainly present, the question is, how much? In a separate paper we shall
discuss the various types of inclusions that are found in granites and among such inclusions
are what we term 'microgranular inclusions with large crystals' that are common in granites,
particularly the more mafic ones. We consider that type of inclusion to be fragments of restite
containing a melt fraction at a lower concentration than the host magma, below the critical
melt fraction of van der Molen & Paterson (1979). The presence of a melt fraction accounts
for many of the petrographic features of these inclusions that Vernon (1983) has used to infer
an igneous origin for all such inclusions. Such a melt fraction would have been present even in
restite inclusions because of the presence of small proportions of the low temperature melting
components (quartz and feldspars). Although inclusions of this type were important in the
original recognition of restite in many granites (Chappell, 1966,1978), we do not assign them
a major role in the restite model. We stress that the bulk of the restite in granites is carried up
from the source as single crystals or small aggregates and not as large fragments; the latter
though may contribute individual crystals by disaggregation, as it must at and near the
region of generation of the magma, but this is generally not a major source of restite crystals
during the final stages of emplacement of a granite magma.
The fine-grained mafic inclusions in moderately mafic rocks of the Boggy Plain Supersuite
are sparsely distributed and have an abundance not correlated with the colour index of the
host rock. They are not regarded as restite but as accidentally incorporated early precipitates.
They are absent from the most mafic rocks. In contrast, most other relatively mafic granites
of the LFB contain inclusions with an abundance approximately correlated with the colour
index of the host rock, and most of these are regarded as restite.
An important conclusion to be drawn from the occurrence of any restite inclusions in a
granite is that if large pieces of rock can be transported up from the source, then clearly single
crystals with a much lower settling velocity will also be transported. It is these single crystals
that, in our view, constitute most of the restite in granites and related volcanic rocks.
Restite minerals
I-type granites may contain orthopyroxene and clinopyroxene included in other minerals
that have grown from them by reaction during cooling e.g., in the Glenbog Suite (above), and
equivalent volcanic rocks may contain these phases as phenocrysts. S-type granites may
contain cordierite and less commonly almandine garnet, as in the Bullenbalong Supersuite
(above), and the related volcanic rocks contain these phases in addition to orthopyroxene. All
of these minerals can occur as precipitates but some are undoubtedly restite, because if they
are liquidus phases they must, virtually by definition, have been residual phases as well.
Cordierite crystals in the S-type granites of the LFB sometimes consist of inclusion-free rims
surrounding a core containing oriented crystals of sillimanite; such cores are clearly restite
(Fig. 2A). Wyborn et al. (1981) report some cordierites in the Hawkins Volcanics containing
cores with oriented needles of sillimanite + hercynite and garnet cores with sillimanite
inclusions, that they regard as relict from the source rock or restite. Clemens & Wall (1984)
record similar cordierite cores, and rare garnets (in addition to other common garnet types
that they regard as magmatic) in the Violet Town Volcanics. These contain cores with
inclusions of sillimanite ± quartz, and inclusion-free rims which they interpreted as restite or
xenocrystal garnets with magmatic overgrowths. Just as there is evidence that some
inclusions are restite, there is likewise evidence that some minerals are restite, and again the
question is, how much? Two special cases of restite minerals merit separate consideration,
plagioclase cores and 'old' zircons.

Plagioclase cores
In all granites in suites which we consider to have compositions controlled by varying
degrees of restite separation, there are plagioclase crystals with calcic cores that are uniform
or almost uniform in composition, commonly surrounded by a multitude of oscillatory zones
and followed outwards by normal zoning (Fig. 2B). Most cores have embayed margins, or are
rounded, indicating partial resorption. These cores are considered to be restite phases
because of their uniform composition, contrasting in this regard with the zoned outer
margins. Such cores are rare or absent in those suites thought to have contained very little or
no restite such as the Boggy Plain Supersuite (Fig. 2C). Plagioclase precipitated from granite
melts is zoned because it is a particularly difficult mineral to re-equilibrate with a melt at
temperatures less than 1000 °C (Johannes, 1978). Johannes states that "reaction rates are very
sluggish below 900 °C and that even geological time spans are necessary to reach
equilibrium". We interpret the uniform plagioclase cores as having formed during the slow
prograde ultrametamorphism (melting) of the granite source rocks. Cores of this type have a
rather constant composition in all members of a particular granite suite irrespective of
whether the rocks containing them are relatively felsic or mafic. In the Jindabyne Suite (Hine
et al, 1978), the plagioclase cores are all near An 80 even though the members of the suite range
from hornblende-rich gabbroic-diorites with 53 per cent SiO 2 to biotite granodiorites
containing 68 per cent SiO 2 . In that suite changes occur in the abundance of such crystals and
the size and composition of their outer zones.
Presence of'old1 zircon

There are two well documented cases of the occurrence of old zircon of similar age to Nd
model-ages. These are in the Newer Scottish granites (Pidgeon& Aftalion, 1978; Hamilton et
al, 1980) and in the Berridale Batholith of southeastern Australia (McCulloch & Chappell,
1982; Williams et al, 1983).
Pidgeon & Aftalion (1978) found that most of the Newer granites from the Grampian and
Northern Highlands of Scotland contained inherited 'old 1 zircon. Thus the hornblende-
bearing (I-type) Foyers granite, for example, provided zircon fractions that define an upper
limit on concordia in excess of 1000 Ma. In the same pluton, Hamilton et al (1980) reported
eNd values for six samples in the range —63 to —8-9, with Nd model ages (chondritic mantle)
of 860-1190 Ma; these ages would be ~ 200 Ma older using a depleted mantle model.
Pidgeon & Aftalion (1978) state that the trend of the discordia for the Foyers granites clearly
demonstrates the presence of a component of old inherited zircon.
Williams et al. (1983) studied zircon U-Pb systems in four I-type plutons in the Berridale
Batholith. Two of these plutons are felsic (72 and 76 per cent SiO 2 in the samples studied) and
1 an
B. W. CHAPPELL ET AL.
1122
RESTITE IN GRANITE PETROGENESIS ] U3
"IP

FIG. 2. (A) Cordierite crystal in the Minnegans Adamcllite from the Kosciusko Batholith, with a core (considered to
be restite) crowded with sillimanite needles and a clear outer rim of cordierite considered to have crystallized directly
from the melt phase. Plane polarized light. (B) Complexly zoned and twinned plagioclase in the Jindabyne Tonalite
from the Kosciusko Batholith. The slightly altered core is very calcic, is uniform in composition and is considered to
be a restite phase, whereas the outer zones have crystallized from the melt phase of the magma. Crossed polars. (C)
Plagioclase crystals in a granodiorite from the Boggy Plain pluton in the Kosciusko Batholith. In contrast to those
crystals in the Jindabyne Tonalite, these have good shapes, less complex zoning and no restite cores. Crossed polars.
(D) 'Clot' of ferromagnesian minerals in the Yalgatta Granodiorite of the Glenbong Suite. Biotites predominate,
hornblendes are irregularly shaped and have high relief, and magnetites are black. Clots are interpreted as 'made-
over' or secondary restite.
they may show a small amount of inheritance of an old zircon component. The other two
plutons are more mafic (~65 per cent SiO2) and contain hornblende; for both of these
Williams et al. (1983) reported a zircon component with an age much greater than the
emplacement age. The Tara Granodiorite, emplaced at 412 Ma, shows marked inheritance of
a component at least 820 ±50 Ma old. The slightly older Finister Granodiorite (~440 Ma)
contains inherited zircon modelled at 1220± 110 Ma.
Compston & Chappell (1979) showed that there is a strong correlation between initial
87
Sr/ 86 Sr (at ~ 400 Ma) and calculated source Rb/Sr for the more mafic I-type granites of the
Berridale and Kosciusko Batholiths. This corresponds to a common time of separation of the
source rocks from the mantle of ~ 1500 Ma. McCulloch & Chappell (1982) confirmed that
the source rock age of the mafic I-type granites was much older than the emplacement ages.
They found a range of Nd model ages (depleted mantle) ranging from 1030 Ma for the Tara
pluton, to 1420 Ma for Finister and 1440 Ma for the Merumbago Granodiorite, a pluton
regarded as consanguinous with the Finister Granodiorite. These values of the Nd ages are
model-dependent, but there is good general agreement with the zircon inheritance ages of
WiHiams-et ah (-1-983).
We consider that 'old' zircon such as that reported by Pidgeon & Aftalion (1978) and
Williams et al. (1983) represents solid material brought from the source, and must be taken to
show that at least some restite is carried up by the rising magma. The alternative possible
explanation is that the zircons result from accidental contamination. We reject the latter
possibility, at least in the general case, since other mineralogical and chemical evidence for
contamination is not seen. We consider that these old zircons in I-type granites are derived
from chemically primitive, but old and therefore isotopically evolved, crust of the type
1124 B W. C H A P P E L L ET AL.
advocated by Compston & Chappell (1979) and Chappell (1984), rather than from the
metasedimentary end-member of a series of mixed sources as would be implied by the model
of Gray (1984) for the granites of the LFB. However, in both models the old zircons would be
residual from the melting of the source rocks, which is the point relevant to the present
argument.
Textures
Textural evidence, and comparison with related volcanic rocks (see above), suggests that
much of the hornblende and biotite in granites forms by reaction of earlier pyroxenes, which
may have been restite, with the melt. This type of evidence is strengthened by comparison of
rocks thought to contain restite with rocks thought to have formed by precipitation from a
melt. For example, in rocks of the Tuolumne pluton of the Sierra Nevada (Bateman &

Chappell, 1979) and the Boggy Plain Supersuite (Wyborn et al., 1987), well-shaped crystals
are relatively common (Fig. 2C). Wyborn (1983) also reported plagioclase crystals
poikilitically enclosed in orthopyroxene and hornblende in the more mafic rocks of the
Boggy Plain pluton; these mafic minerals undoubtedly crystallized from the melt. These
textures contrast strongly with the more common textures in granites of the LFB where mafic
minerals in particular, often occur in ragged aggregates (Fig. 2D).
Linear chemical variations

Strong linear correlations for all elements would be expected, and are observed, in those
granite suites whose chemical variation can be ascribed to variation in the degree of
separation of restite from melt; sometimes the linear correlations are extremely high e.g., for
TiO 2 in the Moonbi Suite (Chappell, 1978). The opposite situation does not apply in suites
that have undergone fractional crystallization since that process may produce linear or close
to linear trends for many elements, although generally not for all elements. Clemens et al.
(1983) point out that near linear chemical trends are exhibited by granitoid-volcanic suites
which, in their opinion, show no evidence for the presence of appreciable restite. The point is,
though, that suites in which restite unmixing has occurred must show linear variation for
every element. A good example of nonlinear variation is provided by Ba and Zr in the Boggy
Plain pluton (Fig. 3) where there is a marked inflection corresponding to the appearance of
new liquidus minerals carrying those elements. Again, for that pluton, Sr shows considerable
scatter in the more mafic half of the composition range (Fig. 3), caused by variations in the
proportions of plagioclase and pyroxene in the more mafic cumulates. For the Tuolumne
pluton which is thought to have been produced from a largely molten magma (Bateman &
Chappell, 1979), Sr shows a very scattered variation (Compston & Chappell, 1979), again in
contrast to the regular variation generally observed for that element in suites containing
restite (see Fig. 1). However, for Tuolumne, the situation may be complex as Kistler et al.
(1986) have shown the presence of two separate source magma batches to be responsible for
some of the variation in that pluton.
The regular variation for Sr in the Glenbog Suite in Fig. 1 is remarkable when it is
considered that those data come from 12 separate plutons extending over a distance of
265 km. For each pluton in the Glenbog Suite for which there are several analyses (the
maximum is 10 analyses for the Glenbog Granodiorite), the trend for that pluton is
indistinguishable from the trend for the whole suite. This is readily explained by the restite
model on the assumption that the rocks come from a uniform source. It would be difficult to
account for this variation if each pluton was separately undergoing fractional crystallization,
particularly since the Sr partition coefficients for plagioclase are strongly dependent on
1000 -
800 °° ° V
O o
D
m 600
E : • • S
-
Q_
_ o ^
Q_ 400
O °
_ o
200
°BO
150 ° °°o
L °£ ° °°
E 100 ~ o

Q_ % o
Q_ 08
50 -
O O
O
1 . . . . 1 . . . . 1 1 . . . . 1
°o °° o oo
° *%*
»
600 o t»
o o o o
L
LD
°° ° O
400 o°
E
Q_
Q_
°°oo
o
200
o
1 . . . . 1 . . . . I , . . . 1 . . . . 1 . . . . t
50 55 60 65 70 75
X SiO2
FIG. 3. Harker diagrams for Ba, Zr, and Sr in the Boggy Plain pluton of the Kosciusko Batholith in southeastern
Australia. The marked inflection in the plots for Ba and Zr corresponds to the appearance or disappearance of
liquidus minerals carrying those elements. The scatter in Sr contents at low SiO 2 values is due to variations in the
proportions of plagioclase and pyroxene in individual samples. The trends on this figure are evidence for a largely
liquid magma and the control of rock compositions by fractional crystallization, in contrast to those suites in which
separation of rcstite controls the compositions and generates linear variations.
temperature (Drake & Weill, 1975) and the rates of cooling of the plutons must have varied as
a result of their different sizes, which range from 8 to 335 km 2 in area at the present level of
exposure.
77ie problem of high temperatures if the rocks were mafic liquids

Sometimes unrealistically high temperatures are required to produce the more mafic
granites from liquids of the same composition. When such granites are widespread e.g., the
mafic S-type granites in the Kosciusko Batholith of the LFB, this problem is alleviated if
these are not complete melt compositions. Piwinskii & Wyllie (1968) stated that "The high
temperatures required for complete fusion of the granodiorites and tonalites, even with
excess water vapour present, suggest to us that rocks of this composition are more likely to be
emplaced as magmas containing suspended crystals . . . than as completely liquid magmas".
This problem is also resolved if the mafic rocks are cumulates; this is discussed in the next
paragraph.
1126 B W. CHAPPELL ET AL.
The chemical equivalence of some plutonic and volcanic suites

The more mafic S-type granite compositions in the LFB contain up to 33 per cent MgO
and 5-8 per cent total FeO in cordierite-rich rocks which must come from the crust. By
comparison with S-type volcanic rock compositions (Wyborn et al., 1981) and experimental
data (Clemens & Wall, 1981), it can be seen that these compositions cannot represent liquids
at any reasonable temperature. They must be either cumulates from a more felsic liquid, or a
mixture of melt and restite. The latter is favoured by the common case in which both plutonic
and volcanic rocks have the same composition, as it is not possible to visualize thick ash
sequences (up to 2 km thick) having a cumulate chemistry. This is especially so if
complementary volcanic rocks representing the more fractionated liquids, which would be
more likely to erupt, are absent from the volcanic pile. Wyborn et al. (1981) showed that both
the average of and range in composition of the Hawkins Volcanics (23 analyses) is close to

and slightly more mafic than that of the mafic S-type granites of what we here call the
Bullenbalong Supersuite. Likewise, the Kadoona Dacite, which contains 22 per cent MgO
and which erupted at ~900°C, too low a temperature for such a composition to be a water-
undersaturated melt at moderate crustal depths (above), has a composition similar to that of
the widely developed granites of the Glenbog Suite. In contrast, the Mountain Creek
Volcanics are more felsic than associated plutonic rocks of the Boggy Plain Supersuite to
which they seem to be related by fractional crystallization (Wyborn et al., 1987). The plutons
in that association are the high temperature cumulates and the volcanics represent the
fractionated liquid that escaped from the cumulate pile.
MgO content of moderately felsic lavas

Some volcanic rocks of known eruption temperature contain more MgO than can occur in
liquids at those temperatures at any reasonable pressure and water content. For example, the
May 18,1980, Mount St Helens pumice described by Rutherford et al. (1985) contains 60 per
cent glass plus phenocrysts and erupted at ~ 930°C with a melt containing 057 per cent MgO
and a bulk composition of 1-70 per cent MgO. A total melt with these compositions would
occur at ~ 1100°C, using the experimental data of Rutherford et al. (1985). Those authors
also show that on an anhydrous basis the composition of glass inclusions trapped in
plagioclase is identical to that of the matrix glass between the phenocrysts, indicating that the
melt did not undergo significant fractionation after the melt was trapped in the plagioclase. It
is therefore difficult to argue that the glass composition was produced by fractional
crystallization of a more MgO-rich higher temperature liquid. As an alternative, we suggest
that the Mount St Helens pumice represents a melt plus restite crystals, this being in accord
with the experimental data. It is also consistent with the petrographic and chemical data of
Kuntz et al. (1981) and Lipman et al. (1981). The data from those authors show that the
compositional variation in the Mount St Helens magma chamber can be related to an
increase in the abundance of phenocrysts downwards in the chamber with no variation in
the liquid composition. Another example is the Kadoona Dacite, discussed above, which is
estimated to have erupted at ~900°C and contains 2-2 per cent MgO compared with < 1
per cent MgO that should occur in a water-undersaturated melt at those temperatures.
Difficulty of nucleating crystals within a pluton

Unzoned porphyritic granite plutons, if originally completely molten, must have nucleated
crystals throughout their volume. Sparks et al. (1984) have pointed out that "the margins of a
chamber are a much more probable site for nucleation in a magma chamber than internally
within homogeneous magma". Those authors list four ways in which nuclei for such
phenocrysts could originate: (a) by erosion of crystals nucleated and grown on the margins of
a chamber; (b) by mixing between magmas which results in a supersaturated melt; (c) by
nucleation at double-diffusive interfaces where undercooling is large; and (d) by transport of
restite crystals from the zone of partial melting. The first three mechanisms are unlikely to be
important in an unzoned and relatively uniform pluton. A good example of the fourth
mechanism is the Moonbi pluton of the New England Batholith, eastern Australia (Chappell,
1978). This is a fairly large body (250 km 2 ), with a moderate range in composition (SiO2 from
65 to 71 per cent) in which the variation in the abundance of different elements is generally
highly correlated. The pluton is not zoned and the SiO 2 content within the total range is not
predictable in any given place. This pluton is porphyritic in K-feldspar, and possesses many

of the features that we would ascribe to the presence of significant amounts of restite. It gives
every appearance of having undergone little if any fractionation near its present level in the
crust and this is consistent with the growth of crystals on widely dispersed restite nuclei. This
argument would also apply to porphyritic volcanic rocks such as the Kadoona Dacite and
Hawkins Volcanics, which from their plutonic equivalents are thought to have come from
fairly homogeneous magma chambers. The small compositional variations in these volcanic
rocks are attributable to differences in abundances of phenocrysts, with the most phenocryst-
rich rocks being the most mafic. We thus challenge the traditional view (e.g., Vernon, 1983)
that most phenocrysts in volcanic rocks result from homogeneous nucleation at small
degrees of undercooling. We contend that this sort of nucleation may be rare. Vernon (1983)
also states that "In the extreme application of the restite hypothesis, phenocrysts in volcanic
rocks have been regarded as modified restite. . . . This interpretation, if accepted, alters the
traditional view of volcanic rock microstructures". This is hardly an argument against the
restite model!
Difficulty of separating granite melts from their solid matrix

We have previously commented that the reason for restite being an important component
of many granites is because of the relatively high viscosities of siliceous melts (White &
Chappell, 1977; Compston & Chappell, 1979). Highly viscous granite melts might therefore
be expected to have difficulty in separating from their solid matrix in the zone of partial
melting and the whole mass, that is melt plus solid material, would then move as a fluid
magma. This has recently been analysed rigorously by McKenzie (1985) who has shown that
the ability of a melt to separate from the residual crystals depends principally on the viscosity
of the melt. He considers the case of a 10 km thick layer of wet migmatite, formed by the
partial melting of the crust, and shows that for an initial melt fraction of 10 per cent and an
average grainsize of 1 mm, melt and solid matrix would undergo only limited separation over
geologically long time periods. McKenzie (1985) states that "acid melts will only be able to
move with respect to.their matrices if their viscosity is less than about 1O2 Pas, a condition
which is only likely to be satisfied if the melts are very rich in volatiles". This is in complete
contrast to mantle melts (basaltic magmas), with relatively low viscosity, that can easily
separate from their matrices, even when present only in very small fractions.
Distribution, of apatite as inclusions in other minerals

In those granites thought by us to contain significant amounts of restite, apatite occurs as
inclusions that are not uniformly distributed in other minerals. Chappell (1966) studied this
1128 B. W. CHAPPELL ET AL.
in analysed K-feldspars, biotite, and hornblende in granites from the Moonbi district
(Chappell, 1978) and found that apatite occurs as inclusions in all mineral phases but is
concentrated in biotite and hornblende by a factor of twenty or more relative to K-feldspar.
These data are consistent with phosphorus always having been saturated in the melt at low
concentrations. Apatite was on the liquidus and must therefore have been a restite phase; we
suggest that it was a primary restite phase which was incorporated in the biotite and
hornblende crystals as the primary restite pyroxenes reacted with the magma. A survey of 52
analyses of biotite from granites, made by classical methods prior to 1966, showed that in
almost all cases appreciable calcium was reported, up to a maximum of 2-60 per cent (Larsen
& Draisin, 1948). We are aware of only one case in which this was recognized and a correction
applied; it was done by Turner (1899) in a biotite granite from Yosemite Valley. These data
suggest that apatite is often concentrated in biotite (and hornblende), which if our
interpretation of its occurrence there is correct, implies that restite is an important

component of many granite magmas.
Concluding statement
We have listed ten lines of evidence in support of the restite model. These are not all of
equal weight in our general argument and many of these could individually be amenable to
alternative explanations. However, as a body of evidence we consider that they establish
the validity of the restite model, and its importance in the petrogenesis of many
granites.
RECOGNITION OF RESTITE
We do not contend that all of the compositional variations in granite suites result from
varying degrees of separation of melt and restite. However, we think that many and perhaps
most suites that include more mafic rocks (SiO2 < 70 per cent) do vary in composition for that
reason, so that relatively mafic rocks in such suites contain more restite than the associated
more felsic rocks. Even very felsic granites (~ 73-75 per cent SiO2) may contain small
amounts of restite.
In those granite suites that show a wide range in composition, the total pattern of the
chemical variation will indicate whether or not restite unmixing has played a role in
producing the compositions of rocks in the suite. Such indications should be confirmed by
the observation of restite minerals such as plagioclase cores, and textures (see Fig. 2). If
chemical data are not available or if the range in composition is small, then the presence of
rather uniform calcic plagioclase cores is the best single line of evidence for the presence of a
restite component. Other primary restite minerals will generally be destroyed during slow
cooling of the granite at higher levels in the crust, but restite plagioclase is preserved as cores
because it is effectively removed from the system by zoning and it equilibrates very slowly at
the temperatures of crystallization of granite plutons. Associated with such cores one would
expect to find clots of ferromagnesian minerals formed when anhydrous primary restite
pyroxenes move into the stability field of hydrous minerals as the activity of water in the
melt increases. In the absence of plagioclase cores, such clots might not themselves be
diagnostic of restite but they are certainly a conspicuous feature of those granites known to
contain appreciable contents of restite.
IMPLICATIONS OF THE RESTITE MODEL

Compositions of source, melt and restite
For those suites to which the restite model is applicable, severe constraints can be placed
not only on the composition of the melt phase, but also on the composition of the source
material from which the granites were derived; these granites "image their sources"
(Chappell, 1979). The composition of granite source, melt and restite must lie on the
compositional trend defined by the granites and the relationship between the three
compositions is shown diagrammatically in Fig. 4, this being a simple variation diagram of
an element against SiO 2 for five granites of a hypothetical suite. The granite compositions,
shown as dots, have a perfect linear trend. This means that if the differentiation results from
restite removal (restite fractionation), then the composition of the melt must be at sample
1 or between that sample and L, the latter being the limiting point on the SiO2-axis. If restite

can be recognized petrographically then it will be most abundant in sample 5, less common in
4, etc. and least common, or absent, in sample 1.
When selecting granite for chemical analysis, inclusions of all types are discarded and we
have not used the composition of lithic inclusions in any restite modelling. The restite that
gives rise to trends within certain granite suites is represented by clots of ferromagnesian
minerals commonly recognizable in the field, and generally overlooked, and the cores of
crystals only clearly seen under the microscope.
Composition of the primary melt

If the composition of a suite is entirely controlled by the fractionation of restite, then the
composition of the melt for all elements will lie at a SiO 2 content between the most felsic
rock containing restite and the limiting case of an element intersecting the SiO2-axis at a
slightly higher SiO2-content (Fig. 4). This is the situation for what we call 'minimum-melt-
75
FIG. 4. Diagram representing the partial melting of a source rock S to produce restite R and liquid L. The magma at
its source will consist of R + L and as it moves away, varying degrees of separation of those two end-members can
generate a range of magma and rock compositions, illustrated by points 1 to 5.
1130 B. W. CHAPPELL ET AL
composition' granites (below) in which the trends for elements such as MgO and P 2 O 5
intersect the SiO2-axis near a haplogranite composition. For those granites formed from
higher temperature restite-bearing melts, the element trends intersect the SiO2-axis at
values greater than the haplogranite composition and the composition of the melt cannot
be determined so precisely. In each case, the primary melt composition can only be
determined if it has not been modified by fractional crystallization following on the removal
of restite. Chappell & White (1984) refer to the restite fractionation regime, which we have
been discussing; this can be followed by a fractional crystallization regime in which the
composition of the melt is modified by the removal of precipitated crystals. In some suites
e.g., the Boggy Plain Supersuite of the LFB, the fractional crystallization regime is entered
early in the magmatic history, but more commonly in the granite suites of the LFB, it is
encountered only late in the magmatic history, if at all. In that case, the granites are felsic so
that quartz and feldspars are virtually the only minerals crystallizing in appreciable

amounts and there will be little change in the major element composition of the resultant
granite. However, some trace elements will be strongly fractionated; for example Sr, Ba, and
Eu decrease rapidly in the melt since these elements are strongly partitioned into the solid
phases, and Rb increases in abundance in the melt which it favours relative to the feldspars.
This leads to rapid changes in the Rb/Sr and Rb/Ba ratios and such changes are indicative
of fractional crystallization from the melt. In the granites of the LFB, Rb contents greater
than 250 p.p.m. and Sr and Ba less than 70 and 200 p.p.m., respectively, are characteristic of
magmatic systems where such fractional crystallization has occurred.
Minimum-melt and non-minimum-melt compositions

Using the methods for deducing the melt composition outlined above, viz the intersection
on variation diagrams of various refractory components such as P 2 O 5 and MgO with the
SiO 2 -axis and the presence or absence of petrographically recognizable restite material in the
high-Si rocks of a suite, two end-member types of restite-controlled suites may be recognized.
The variation diagrams of P 2 Oj in two such cases are shown in Fig. 5. In the first case of the
Moruya Suite (Griffin et al., 1978) the linear trend for P 2 O 3 intersects the SiO2-axis near 77
per cent SiO 2 , i.e. at or near to a haplogranite composition. A similar situation holds for
MgO. The melt must contain some very small amount of P 2 O 5 and MgO and so its
composition will lie between about 76 per cent SiO 2 and the SiO 2 content of the most SiO 2 -
rich sample that still contains restite. We refer to such suites as minimum-melt-composition
suites. This is not the water-saturated 'minimum melting temperature' of Tuttle & Bowen
(1958) since the melt phase of the magma at the site of partial melting will be undersaturated
in water so that there is still a large interval between the formation temperature and the
water-saturated solidus at the site of crystallization. Minimum-melt-composition suites do
not form at the same temperature, but rather over a range of temperatures; the actual
temperature will depend on the H 2 O content of the melt.
In contrast, the variation of P 2 O 5 with SiO 2 in the Jindabyne suite (Hine et al, 1978)
depicted in Fig. 5 is parallel to the SiO2-axis. If this variation results from restite
fractionation, then the source, the melt, and the restite must contain the same, or very similar,
amounts of P 2 Oj. In this case P 2 O 5 has a much higher solubility in the melt phase. In suites
of this kind, the most felsic rocks of the suite have lower SiO 2 contents and the melt
components of the suites have higher P, Ca, Fe, Mg, and some trace elements such as Cr and
Ni, relative to minimum-melt-composition suites (White & Chappell, 1977). We refer to these
suites as non-minimum-melt-composition suites. For these suites it is thought that the
temperature of melting of the source rock had to be raised to a higher level than for
minimum-melt-composition suites, before sufficient melt phase (~ 40 per cent) was produced
60

0. 15
m
o
0. 10 OD 0 O O O
0. 05
J i ndabyne
55 60 65 70
7. SiO2
FIG. 5. Harker diagrams for P 2 O 3 in the Moruya and Jindabyne suites in the Lachlan Fold Belt. The variation in
composition in these suites is generated by varying degrees of separation of restite from more siliceous melt. In the
Moruya Suite the melt was siliceous (> 75 per cent SiO2) and low in P 2 O 5 ; the higher-temperature of melting in the
Jindabyne Suite produced a more mafic melt (> 68 per cent SiO2) containing close to the same P 2 O 5 content as the
bulk restite.
to convert the mass into a fluid magma (Arzi, 1978; Compston & Chappell, 1979; van der
Molen & Paterson, 1979). This means that minimum-melt-composition and non-minimum-
melt-composition melts are derived from sources of different composition. In the latter case
there is a deficiency of one of the haplogranite components quartz, K-feldspar or plagioclase
in the source, which was therefore depleted before sufficient melt was formed, with the result
that melting continued to higher temperatures. This difference could be in the availability of a
volatile component such as H 2 O which will be a factor in determining the amount of melt
formed during partial melting. In an extreme case, high temperature liquids of relatively low
viscosity may be produced if the source rocks are more mafic and this can lead to the
production of restite-free magmas e.g., the Boggy Plain Supersuite.
Composition of the primary restite

The composition of the primary restite will lie on the extrapolation of the variation to
lower SiO 2 values in Fig. 4. Its composition is not given by those of inclusions thought to be
restite, because: (a) these probably represent more refractory parts of the source, which
therefore produced insufficient melt to become a magma; and (b) their compositions are
invariably modified by reaction with the melt. Compston & Chappell (1979) modelled the
composition of primary restite using the disappearance of a normative mineral such as quartz
when the compositions were extrapolated to lower SiO 2 values. These model restite
compositions are limiting low-SiO2 values.
1132 B. W. C H A P P E L L ET AL.
Primary and secondary restite

The primary restite produced by partial melting in the crust may contain some biotite or
hornblende, but the dominant ferromagnesian restite phases will normally be two pyroxenes
for I-type sources and orthopyroxene ± cordierite ± garnet ± biotite for S-type sources. These
anhydrous phases are produced by the vapour-absent breakdown of hydrous minerals.
Biotite ± hornblende could be present in the primary restite, a case shown for example by the
occurrence of abundant inclusions of monazite in the biotite crystals of some S-type granites.
Anhydrous primary restite minerals are often preserved in volcanic rocks as large crystals
(phenocrysts), but in plutonic rocks they are largely or completely modified. In the latter case,
during slow cooling there is an increased activity of H 2 O within the melt, leading to a reversal
of the earlier reactions, but at a lower pressure so that the changes are not exactly reversed.
For this reason, biotite rarely with orthopyroxene inclusions, hornblende sometimes with

clinopyroxene cores (I-types) and various sheet silicates replacing cordierite (S-types) are
found in granites as secondary restite. Unzoned plagioclase cores in granites are thought to be
unmodified, although partly resorbed, primary restite.
Restite quartz
Quartz must be a restite phase at the source in all minimum-melt-composition granites
that formed in equilibrium with that mineral plus feldspars. Thus it may be a restite phase, at
the source, for many I-type granites and it is probably a restite phase for all S-type granites,
even though the latter are normally non-minimum-melt-composition granites. In almost any
rising granite magma quartz is resorbed but much will persist to high levels, so that restite
quartz is probably widely distributed in granites but its identification is generally difficult.
Restite quartz in volcanic rocks occurs as partly corroded phenocrysts (Wyborn & Chappell,
1986), but likewise any precipitated phenocrysts will be resorbed. In the Glenbog Suite, the
large lumps of quartz up to 1 cm across that are evenly scattered throughout the rock and
surrounded by hornblende rims are clearly an early phase. In that case, the occurrence of
those crystals in granites occurring over a distance of some 250 km is not consistent with their
derivation from contamination by sedimentary material, nor by mixing of a haplogranite and
a basalt, and we interpret them as restite. A large proportion of the quartz in mafic S-type
granites is likely to be restite but not readily identifiable as such. The modal abundance of
quartz in S-type granites either stays at about the same value or increases slightly as the rocks
become more mafic. Thus the most mafic S-type granites contain about 40 per cent quartz
and 25 per cent biotite and such rocks would have formed from magma containing
approximately 40 per cent melt and 60 per cent restite. About one-third of the melt would
have been a quartz component, equivalent to 12 per cent in the total-rock, so that 70 per cent
of the quartz would be of restite origin. Most mafic S-type granites also contain fragments of
milky quartz commonly up to 5 cm across, and in extreme cases elongate and up to 25 cm in
length, that are clearly restite and which White et al. (1977) have interpreted as fragments of
vein quartz from the source rocks.
Composition of the source

The composition of the source rock must also lie on the trend in Fig. 4, if the variation is
controlled by restite separation from melt. In a general way, the granites themselves will
reflect the composition of the source, or form an image of their source e.g., the low Na 2 O
content of S-type granites in the LFB reflects the low content of sodium in their sedimentary
source rocks. More specifically, attempts can be made to determine the actual source rock
composition, and there are two ways of doing this. First, Compston & Chappell (1979)
calculated model melt and model restite compositions in non-minimum-melt-composition

suites and blended these in appropriate proportions e.g., in a ratio 2:3 if 40 per cent melt is
considered appropriate for the transition from rigid restite framework to fluid magma. Such a
calculation for minimum-melt-composition suites gives minimum values for the SiO 2
content of the source since a melt of minimum-melt-composition must be in equilibrium
with quartz and two feldspars and therefore must be more siliceous than the model restite.
The second method was suggested by Chappell (1984) to overcome the problems of
modelling minimum-melt-composition suites, and the following is taken directly from that
publication. If F is the fraction of melt that marks the transition from rigid restite framework
to fluid magma, then F is the minimum amount of melt that the magma can contain and (\-F)
is the maximum amount of restite. If the magma moves up and differentiates to form a pluton
of variable composition then the most mafic rock in that pluton will contain the maximum

amount of restite, again (1-F), and the minimum amount of melt, F. Thus in a granite suite the
most mafic possible composition (most restite-rich) will be the same as the source
composition, within the limits of the model. Thus in any suite, the most mafic granite
composition can be taken as the model source composition; this will be close to the actual
source composition if exposure allows the full range of chemical variation in a given suite to
be seen.
This second method of estimating source compositions clearly cannot be used if mafic
rocks are not present, but in our experience in the LFB it can be used sufficiently often to give
a useful estimate of many source rock compositions.
Implications in the interpretation of granite suites

For those suites to which the restite model can be applied, the implications are that
different suites correspond in a specific way to different source rock compositions and much
of the variety between granite suites is the result of sources having different compositions.
Groupings of suites, or supersuites, correspond to sources which are similar but not identical
in composition. The first-order subdivision of suites in the LFB, into I-types and S-types,
reflects the differences in source rocks that had or had not been weathered. Within the S-type
group, there are differences of a second-order that correspond to different degrees of
weathering of the source rocks (Chappell, 1984).
Occurrence of incompatible elements in magmas

Another implication of the restite model is that some elements may be present at a
significant concentration in granite magmas despite their very low solubility in granite melts.
For example, much of the light rare-earth elements in S-type granites can be present in restite
monazite, so that the more mafic rocks will be relatively LREE-enriched. Again, it is possible
that the solubility of Sn in oxidized I-type melts is low, so that much of the Sn found on
analysis of such a rock was present in the magma in restite minerals such as biotite and
sphene rather than in the melt.
Width of some contact aureoles

The presence of restite in more mafic granite magmas may account for some apparent
anomalies in the width of contact aureoles in the LFB. While it must be recognized that the
factors that determine the width of an aureole may be complex, the relative temperatures of
different granite plutons must be an important parameter. In the Kosciusko region of the
1134 B. W. CHAPPELL ETAL.
LFB, very minor aureoles are consistently found around some quite mafic granites e.g., the
mafic S-type granites, which is consistent with relatively high crystallinities and low
temperatures in a magma made up of felsic melt and mafic restite components. Thus, Owen &
Wyborn (1979) who indicated the width of recognizable contact metamorphic effects around
plutons on the Tantangara 1:100000 Geological Map, were unable to recognize any
metamorphism around the McLaughlins Flat Granodiorite, a rather mafic member of the
Bullenbalong Suite. In contrast, the Bugtown Tonalite, which in places is a little less than one
kilometre from the McLaughlins Flat body, is surrounded by a conspicuous aureole, varying
from 800 m to 1-5 km in width. Bugtown is now recognized as the northernmost
representative of the Jindabyne Suite (Hine et al., 1978), discussed above as a non-minimum-
melt-composition suite, formed a higher temperature melt containing varying proportions of
restite. Likewise, the Boggy Plain pluton, thought to have been virtually restite-free for other

reasons, and hence to have formed from a comparatively high temperature melt, has a well
developed aureole which can be recognized up to 1-5 km from the contact of similar country
rocks.
Granite tectonics
The restite model implies that many granite magmas move away from their source en
masse rather than by a process of early separation of melt from solid material. In such a case,
the whole source volume moves upward, in contrast to a situation in which melt moves up
through solids residual from melting that are moving down. This then represents a classic
diapiric mechanism in which fluid material moves upward relative to the surrounding rocks
because of that fluidity combined with relatively low density. Such movement decreases the
gravitational potential and will continue until the magma either erupts or 'quenches'. The
latter situation for a granite magma means that crystallization has proceeded to the extent
that solid material makes up more than 60 per cent of the magma, i.e. the amount of melt
becomes less than the critical melt fraction. We suggest that the term 'granite tectonics' be
used to refer to the structural effects of the movement of large bodies of granite magma in this
way, by analogy with the term 'salt tectonics' applied with reference to dynamic salt
structures (e.g., Jackson & Talbot, 1986). The movement of large bodies of granite magma,
probably with volumes of several thousand cubic kilometres for the largest plutons, through
distances of 15 km or more in the crust, could be expected to produce effects outside that
pluton analogous to those associated with salt diapirs. Such external structures could include
marginal synclines or sinks around the pluton that extend down into the deep crust to
compensate for the lateral movement of magma into the granite diapir. On a larger scale, the
possibility that some large basins marginal to granite batholiths may have been initiated by
the composite effect of many separate diapirs, should be examined.
PARTIAL MELTING SCENARIOS

There is little doubt that most granites are products of partial melting. Most people agree
that the melt is produced by vapour-absent melting where the volatile phase, normally H 2 O,
is provided for the melt by the breakdown of biotite or hornblende (e.g., Burnham, 1979).
Considering the case of I-type granite sources, the source rock is here considered to be a
hydrated mafic to intermediate rock such as basalt (gabbro) or andesite (diorite) with the
assemblage amphibole + biotite -I- plagioclase + quartz + magnetite ± K-feldspar. Most partial
melting occurs when amphibole (and biotite) breaks down in the presence of quartz
and feldspars. After partial melting, the assemblage will be granite melt undersaturated with
volatiles together with restite phases—plagioclase + clinopyroxene + orthopyroxene + mag-

netite. In the presence of excess quartz and feldspars, the activity of H 2 O or other volatiles
controls the amount of melt phase and hence quartz and K-feldspar can also occur as restite
phases. Since hornblende and biotite have widely varying compositions which break down at
varying temperatures, these two phases may also be present as restite. By definition, restite
must be present during partial melting, although its precise nature is debatable. However,
after this stage there are several possible scenarios.
Restite retention hypothesis

We suggest that during partial melting the melt and restite phases closely approach
equilibrium, so that the residual phases are relatively uniform in composition; if they are

isolated by melt they will have good crystal shapes and may be texturally indistinguishable
from precipitated crystals. Provided sufficient components to form melt are present (quartz,
feldspars, and H 2 O) and heat is continually introduced, melting will continue until the rigid
restite framework is broken and mechanical instability occurs with resultant production of
magma. This will occur when 30-50 per cent of melt is present and the system passes from
solid to suspension-like behaviour. This profound change in character of the partly melted
rock is illustrated by the experiments of van der Molen & Paterson (1979). That smaller
amounts of melt are insufficient for mobility is also indicated by field studies on migmatites in
which partial melts amounting to about 10 per cent of the volume can persist through several
phases of deformation without loss of fold coherency (Fleming & White, 1984). In this
scenario, when the amount of melt reaches about 40 per cent, the whole mass moves upward
because of its buoyancy. When there are heterogeneities in the source composition so that
there is less than the critical amount of melt present to form a magma in some places, these
heterogeneities can remain and appear as restite inclusions. Some separation of restite, both
as single crystals and inclusions, can occur during the ascent of the magma and it is this restite
fractionation that leads to the geochemical variation within a suite of granites derived from a
particular melting event. The activity of volatiles such as H 2 O would normally increase
during ascent so that the restite, particularly quartz, is likely to be partly absorbed during this
stage and no fractional crystallization will occur. As the magma cools, reaction of the
pyroxenes with the melt produces hornblende and biotite as soon as the magma enters the
stability field of these minerals. The same reactions will simultaneously occur within restite
inclusions. In some cases, all restite will be removed and the magma will then pass from the
restite fractionation regime to the fractional crystallization regime (Chappell & White, 1984).
Alternative scenarios
Those who argue against the restite model and advocate fractional crystallization must
accept the hypothesis of complete or almost complete separation of melt from restite, at or
near the site of partial melting, in spite of the high viscosity of granite liquid compared with
basalt liquidand the difficulty of separating low melt-fractions ofvlscous liquid (McKenzie,
1985). We accept that such a separation sometimes happens, as in the case of more mafic
melts, at higher temperatures e.g., the Boggy Plain Supersuite (Wyborn et al., 1987), but this is
not the general case for the Lachlan Fold Belt. After complete separation, the melt must
partly crystallize, with the liquidus phases being the same as those of the restite. This occurs
despite the problems of homogeneous nucleation (Sparks et al. 1984). In the case of the
Kadoona Dacite, there must have been 60 per cent crystallization of pyroxenes, plagioclase,
and quartz before the hornblende and biotite stability fields were reached. All of these early-
1136 B. W. C H A P P E L L ET AL.
formed crystals were retained in the magma, and crystallization must have occurred under
equilibrium conditions and extremely slowly for the plagioclase crystals to equilibrate. Any
fractional crystallization that occurred must have been of such a nature as to be able to
produce good linear correlations for all elements. Also, according to this scenario, basalt
must have mingled with the granite melt at some stage, to explain the mafic inclusions, and
moreover this basalt must have incorporated phenocrysts from the granite magma such as
the large quartz crystals in inclusions in the Glenbog Suite.
Another scenario could be that granitic melt separated completely at the partial melting
stage, and then mixed with basalt to generate the linear chemical trends. With this scheme,
the pyroxenes and plagioclase in the volcanic equivalent Kadoona Dacite would be
phenocrysts from the basalt and the quartz crystals must have been phenocrysts from a
granitic melt, which must therefore have partly crystallized before mixing. The absence of K-
feldspar and more albitic plagioclase phenocrysts indicates that the mixed magma reacted

and incorporated all granitic crystalline phases except quartz in the melt.
The low probabilities associated with aspects of these alternative scenarios lends further
strong support to the hypothesis of restite retention and the restite model.
CONCLUSIONS
We conclude that the restite retention hypothesis is simple and readily explains all of the
facts deduced from petrography (including fabric studies), mineralogy, chemistry, and
experimental studies, and is more in keeping with what is known about the physics of
magmas than alternative hypotheses. We suggests that the restite model, involving varying
degrees of separation of the restite components from the melt phase of the magma, explains
the petrographic and chemical variation within and between many granite suites. We think
that the simplest way of explaining certain microgranular inclusions is that they are restite
derived from the source region of partial melting. A melt fraction was present in these
inclusions at a lower concentration than in the host magma, which explains the petrographic
features of these inclusions. Using presently available evidence, we suggest that, in most cases,
lithic inclusions of any type have little effect on the chemical variation of granite suites.
ACKNOWLEDGEMENTS
This study has benefited greatly from discussions with many people. This paper has been
improved after the cdmments of two reviewers and of Tony Ewart and Steve Wickham.
Financial support has been received from the Australian Research Grants Scheme to BWC
and AJRW for studies of granites in the Lachlan Fold Belt.
REFERENCES
Arzi, A. A., 1978 Critical phenomena in the rheology of partially melted rocks. Tectonophysics, 44, 173-84.
Barron, L. M., Scheibner, E., & Suppel, D. W., 1982. The Mount Hope Group and its comagmatic granites on the
Mount Allen 1:100000 sheet. New South Wales. NSW Geol. Surv. Quart. Notes 47, 1-17.
Bateman, P. C , & Chappell, B. W., 1979. Crystallization, fractionation and solidification of the Tuolumne intrusive
series, Yoscmite National Park, California. Bull. geol. Soc. Am. 90, 465-82.
Clark, L. D., Huber, N. K., Moore, J. G , & Rinehart, C. D., 1963. The Sierra Nevada Batholith: a synthesis of
recent work across the central part. Prof. Pap. U.S. Geol. Surv. 414D.
Beams, S. D., Chappell, B. W., & White, A. J. R., 1987. The Bega Batholith: a petrographically and chemically
asymmetrical plutonic complex of Devonian age in the Lachlan Fold Belt, southeastern Australia (in prep.)
Bowen, N. L., 1928. The Evolution of the Igneous Rocks. Princeton. Princeton University Press.
Burnham, C. W., 1979. Magmas and hydrothermal fluids. In: Barnes, H. L (ed.) Geochemistry ofHydrothermal Ore
Deposits. New York: Wiley & Sons, 71-136.
R E S T I T E IN G R A N I T E P E T R O G E N E S I S 1137
Chappell, B. W., 1966. Petrogenesis of the granites at Moonbi, New South Wales. Ph.D. thesis. The Australian
National University.
1978. Granitoids from the Moonbi district, New England Batholith, eastern Australia. J. geol. Soc. Ausl. 25,
267-83.
1979. Granites as images of their source rocks. Geol. Soc. Am. Prog. Abs. 11, 400.
1984. Source rocks of I- and S-type granites in the Lachlan Fold Belt, southeastern Australia Phil. Trans. R.
Soc. Lond. A310, 693-707.
White, A. J. R., 1974. Two contrasting granite types. Pacific Geol. 8, 1 7 3 ^ .
1984.1- and S-type granites in the Lachlan Fold Belt, southeastern Australia. In Xu Keqin & Tu Guanchj
(eds) Geology of Granites and Their Metallogenic Relations. Beijing; Science Press, 87-101.
Clemens, J. D., Wall, V. J., & Clarke, B. D., 1983. Models for granitoid evolution and source compositions: restite—
R.I.P.? Geol. Soc. Aust. Abs. 9, 178-79.
1981. Original and crystallization of some peraluminous (S-type) granitic magmas. Can. Miner. 19,
111-31.
1984. Origin and evolution of a peraluminous silicic ignimbnte suite: The Violet Town Volcanics. Contr.
Miner. Petrol. 88, 354-71.

Collins, W. J., Beams, S. D., White, A. J. R., & Chappell, B. W., 1982. Nature and origin of A-type granites with
particular reference to southeastern Australia, Ibid. 80, 189-200.
Compston, W., & Chappell, B. W., 1979. Sr-isotope evolution of granitoid source rocks. In: McElhinny, M. W. (ed.)
The Earth: Its Origin, Structure and Evolution. London. Academic Press, 377—426.
Didier, J., 1973. Granites and their enclaves. Amsterdam: Elsevier.
Drake, M. J., & Weill, D. F., 1975. Partition of Sr, Ba, Ca, Y, Eu 2 + , Eu 3 + , and other REE between plagioclase
feldspar and magmatic liquid: an experimental study. Geochim. cosmochim. Acta, 39, 689-712.
Fleming, P. D., & White, A. J. R., 1984. Relationships between deformation and partial melting in the Palmer
migmatites, South Australia. Aust. J. Earth Sci. 31, 351-60.
Gray, C. M., 1984. An isotopic mixing model for the origin of granitic rocks in southeastern Australia. Earth planet
Sci. Lett. 70, 47-60.
Griffin, T. J., White, A. J. R., & Chappell, B. W., 1978. The Moruya Batholith and geochemical contrasts between the
Moruya and Jindabyne Suites. J. geol. Soc. Aust. 25, 235-47.
Hamilton, P. J., O'Nions, R. K., & Pankhurst, R. J., 1980. Isotopic evidence for the provenance of some Caledonian
granites. Nature, 287, 279-84.
Harley, S., & Green, D. H., 1982. Garnet-orthopyroxene barometry for granulite and peridotite. Ibid. 300,697-701.
Hine, R., Williams, I. S., Chappell, B. W., & White, A. J. R., 1978. Contrasts between I- and S-typc granitoids of the
Kosciusko'Batholith. J geol. Soc. Aust. 25, 219-34.
Indares, A., & Martignole, J., 1985. Biotite-garnet thermometry in the granulite fades: the influence of Ti and Al in
biotite. Am. Miner. 70, 272-8.
Jackson, M. P. A., & Talbot, C. J., 1986. External shapes, strain rates, and dynamics of salt structures. Bull. geol. Soc.
Am. 97, 305-23.
Johannes, W., 1978. Melting of plagioclase in the system Ab-An-H 2 Oand Qz-Ab-An-H 2 O at Plho = 5 kbars, an
equilibrium problem. Contr. Miner. Petrol. 66, 295-303.
Judd, J. W., 1886. On the gabbros, dolerites and basalts of Tertiary age in Scotland and Ireland. J geol. Soc. Lond.
42, 49-97.
Kistler, R. W., Chappell, B. W., Peck, D. L., & Bateman, P. C , 1986. Isotopic variation in the Tuolumne Intrusive
Suite, central Sierra Nevada, California. Contr. Miner. Petrol. 94, 205-20.
Kuntz, M. A., Rowley, P. D., MacLeod, N. S., Reynolds, R. L., McBroome, L. A., Kaplan, A M., & Lidke, D J., 1981.
Petrography and particle-size distribution of pyroclastic-flow, ash-cloud, and surge deposits. In:
Lipman, P. W. & Mullin, D. R. (eds) The 1980 Eruptions of Mount St Helens, Washington, Prof. Pap U.S. Geol.
Surv. 1250, 525-39.
Lacroix, A., 1933. Sur quelques granites des environs de Porto. An. Fac. Cienc Univ. Porto, 18, 35-68.
Larsen, E. S., & Draisin, W. M., 1948. Composition of minerals in the rocks of the Southern California batholith.
Rep. Int. Geol. Congr. X VUl Session, 2, 66-79.
Lipman, P. W., Norton, D. R., Taggart, J. E, Brandt, E. L., & Engleman, E. E., 1981. Compositional variations in
1980 magmatic deposits. In: Lipman, P. W. & Mullin, D R. (eds) The 1980 Eruptions of Mount St. Helens,
Washington, Prof. Pap. U.S. Geol. Surv 1250, 631-40.
Luhr, J. F., Carmichael, 1. S. E, & Varekamp, J. C , 1984. The 1982 eruptions of El Oiichon volcano, Chiopas,
Mexico: mineralogy and petrology of the anhydrite-bearing pumices. J. Volcanol. geotherm. Res 23, 69-108.
McCarthy, T. S., & Groves, D. I., 1979. The Blue Tier Batholith, north eastern Tasmania. Contr. Miner. Petrol. 71,
193-209.
McCulloch, M. T., & Chappell, B. W., 1982. Nd isotopic characteristics of S- and I-type granites. Earth planet Sci
Lett. 58, 51-64.
McKenzie, D., 1985. The extraction of magma from the crust and mantle. Ibid. 74, 81-91.
Montel, J. M., 1985. Is monazite guilty? Experimental determination of Ce-monazite solubility in Na 2 O-K. 2 O-
SiO 2 -Al 2 O 3 melts. Terra Cognita, 5, 330.
Owen, M., & Wyborn, D., 1979. Geology and geochemistry of the Tantangara and Brindabella 1:100000 sheet
areas. Bur. Miner. Resow. Geol. Geophys. Aust. Bull. 204, 52 pp.
1138 B. W. C H A P P E L L ET AL
Pidgeon, R. T., & Aftahon, M., 1978. Cogenetic and inherited zircon U-Pb systems in granites: Palaeozoic granites
of Scotland and England. In. Bowes, D R , & Leake, B. E (eds) Cruslal Evolution in Northwestern Britain and
Adjacent Regions, Geol. J. Special Issue, 10, 183—220.
Piwinskii, A. J., & Wyllie, P. J., 1968. Experimental studies of igneous rock series- a zoned pluton in the Wallowa
batholith, Oregon. J. Geol. 76, 205-34.
Reid, J. B , Evans, O. C , & Fates, D. G., 1983. Magma mixing in granitic rocks of the central Sierra Nevada,
California. Earth planet Set Lett. 66, 243-61.
Rushmer, T., Holloway, J R., & Wyborn, D., 1985. Melting relations in a classic Australian I-typedaciteat lOkbars
EOS, 66, 1130.
Rutherford, M J, Sigurdsson, H., Carey, S., & Davis, A., 1985. The May 18, 1980, eruption of Mount St. Helens 1
Melt composition and experimental phase equilibria. J geophys. Res. 90, 2929-47.
Sparks, R. S J., Huppert, H. E., & Turner, J S., 1984. The fluid dynamics of evolving magma chambers. Phil. Trans
R. Soc. Land. A310, 511-34.
Turner, H. W., 1899. Some rock-forming biotites and amphiboles. Am. J. Sci. 157, 294-8.
Tuttle, O. F., & Bowen, N. L., 1958. Origin of granite in the light of experimental studies in the system
NaAISi 3 0 8 -KAlSi 3 O 8 -SiO 2 -H 2 0. Geol. Soc. Am. Mem. 74.
van der Molen, I., & Paterson, M. S., 1979 Experimental deformation of partially melted granite. Contr. Miner.

Petrol. 70, 299-318.
van Moort, J C, 1966. Les roches cnstallophylliennes des Cevennes et les roches plutoniques du Mont Lozere. Ann.
Fac. Sci. Univ. Clermont-Ferrand, 31, 272 pp.
Vernon, R. H., 1983. Restite, xenohths and microgranitoid enclaves in granites J. Proc. R.Soc N S.W. 116,77-103
White, A. J R., & Chappell, B. W., 1977. Ultrametamorphism and granitoid genesis. Tectonophysics, 43, 7-22.
1983. Granitoid types and their distribution in the Lachlan Fold Belt, southeastern Australia. In:
Roddick, J. A. (ed.) Circum-Pacific Plutonic Terranes, Geol. Soc. Am. Mem. 159, 21-34.
1985. Chemical evidence for different sources of granite magmas. Geol. Soc. Am. Prog. Abs 17, 749.
Williams, I. S., & Chappell, B. W., 1977. Geology of the Berndale 1:100000 Sheet (8625). Sydney: Geol. Surv.
NSW.
Wickham, S. M., 1987 The segregation of granite magmas. J. geol Soc. Land., 144, 281-97.
Williams, I. S., Compston, W., & Chappell, B. W., 1983. Zircon and monazite U-Pb systems and the histories of
I-type magmas, Berridale Batholith, Australia. J. Petrology, 24, 76-97.
Wyborn, D., 1983. Fractionation processes in the Boggy Plain zoned pluton. Ph.D. thesis. The Australian National
University.
Chappell, B. W., 1986. The petrogenetic significance of chemically related plutonic and volcanic rock units.
Geol. Mag. 123, 619-28.
Johnston, R. M., 1981. Three S-type volcanic suites from the Lachlan Fold Belt, southeast Australia. J.
geophys Res. 86, 10335-^8.
Owen, M, 1986. Araiuen, New South Wales— I • 100000 map commentary. Canberra: Bur. Miner. Resour. Geol.
Geophys. Aust.
Turner, B. S., & Chappell, B. W., 1987. The Boggy Plain Supersuite—a distinctive belt of I-type igneous rocks of
potential economic significance in the Lachlan Fold Belt. Aust. J. Earth Sci. 34, 21—43.

The Importance of Residual Source Material (Restite) in Granite Petrogenesis

Uploaded by

Copyright:

Available Formats

The Importance of Residual Source Material (Restite) in Granite Petrogenesis

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Importance of Residual Source Material (Restite) in Granite Petrogenesis

Uploaded by

Copyright:

Available Formats

The Importance of Residual Source Material (Restite)

by B. W. CHAPPELL, 1 A. J. R. WHITE 2 AND D. WYBORN 3

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

THE RESTITE MODEL

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

The restite model

SYSTEMATIC VARIATION AMONG SOUTHEASTERN AUSTRALIAN

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Boggy Plain Supersuite

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Kadoona Dacite—an equivalent volcanic suite

Crystallization history of the Glenbog Suite

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

MECHANISMS FOR PRODUCING VARIATION IN GRANITES

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

EVIDENCE FOR THE RESTITE MODEL

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Presence of'old1 zircon

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Linear chemical variations

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

77ie problem of high temperatures if the rocks were mafic liquids

The chemical equivalence of some plutonic and volcanic suites

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

MgO content of moderately felsic lavas

Difficulty of nucleating crystals within a pluton

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Difficulty of separating granite melts from their solid matrix

Distribution, of apatite as inclusions in other minerals

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

IMPLICATIONS OF THE RESTITE MODEL

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Composition of the primary melt

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Minimum-melt and non-minimum-melt compositions

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Composition of the primary restite

Primary and secondary restite

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Composition of the source

calculated model melt and model restite compositions in non-minimum-melt-composition

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Implications in the interpretation of granite suites

Occurrence of incompatible elements in magmas

Width of some contact aureoles

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

PARTIAL MELTING SCENARIOS

volatiles together with restite phases—plagioclase + clinopyroxene + orthopyroxene + mag-

Restite retention hypothesis

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

Downloaded from http://petrology.oxfordjournals.org/ at Duke University on May 19, 2012

You might also like