JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12
A Review of Shallow, Ore-related Granites:
Textures, Volatiles, and Ore Metals
PHILIP A. CANDELA∗
LABORATORY FOR MINERAL DEPOSITS RESEARCH, DEPARTMENT OF GEOLOGY, UNIVERSITY OF MARYLAND AT
COLLEGE PARK, COLLEGE PARK, MD 20742-4211, USA
RECEIVED JANUARY 1997; ACCEPTED AUGUST 1997
Shallow, texturally complex granites can be associated, spatially
and temporally, with hydrothermal alteration and mineralization.
The formation of a granite-related ore deposit represents the confluence
of many chemical and physical geological factors, including: magma
composition (e.g. concentrations of ore metals, chlorine, and water);
magma oxidation state; the relative timing of crystallization, magma
ascent, magmatic volatile phase exsolution; and the depth of em-
placement. Early magmatic crystallization of ore-metal-rich minerals
coupled with late exsolution of the magmatic volatile phase is
promoted by deeper emplacement levels and lower magmatic water
concentrations, and leads to low efficiencies of removal of compatible
elements from melts into the magmatic volatile phase(s). Water
concentrations in the melt and pressures of crystallization can be
estimated from phase equilibria and textural relationships of horn-
blende in some felsic systems. Physical factors affect the dynamics
of fluid (magma, melt and volatile phase) movement, which in turn
affect the size of zones of alteration and mineralization; granite
textures can record the partial history of some dynamic processes.
Miarolitic cavities are good evidence for magmatic volatile phase
exsolution; both miarolitic cavities and pegmatitic texture exhibit
evidence for external nucleation of crystals. Nucleation-controlled
phenomena, expressed as variations in granite textures, can result
from a multiplicity of causes including undercooling of the melt;
textural elements such as pegmatite, aplite, skeletal crystals, and
graphic textures can form as a result of undercooling. Interconnected
volumes of the magmatic volatile phase allow for advection of
hydrothermal ore fluids through the magma. In shallow granitic
systems, miarolitic cavities can be interconnected, consistent with
this hypothesis.
KEY WORDS: granites; ores; textures; volatiles
∗Telephone: 1-301-405-2783. Fax: 1-301-314-9661.
e-mail: candela@geol.umd.edu
www: http://www.geol.umd.edu/pages/facilities/lmdr/lmdr.html
PAGES 1619–1633 1997
INTRODUCTION
Granite-related ore deposits include porphyry and skarn
type ores, and other hydrothermal deposits that are
spatially and temporally associated with granitic (sensu
lato) rocks. In this paper, I emphasize characteristics
found in porphyry style (Cu–Mo–Au) deposits, and some
related systems; however, my treatment here is not meant
to cover the full range of granitic rocks and mineralization
that make up the granite-related ore systems. The for-
mation of granite-related ore deposits by orthomagmatic–
hydrothermal processes is dependent upon many physical
and chemical magmatic factors, including (but not limited
to) the composition and evolution of the associated
magmas, the rate of magma ascent, the depth and
geometry of magma emplacement, and the local tem-
perature and stress fields at the site of magma em-
placement. Detailed studies of the igneous rocks
associated with ore deposits can therefore yield critical
information regarding the source of metals and ligands
in the ore-generative hydrothermal solutions that are
operative in the formation of granite-related ores. In this
review, I will present a series of narratives, consistent, in
my view, with prevailing field and experimental data on
the orthomagmatic–hydrothermal theory for the genesis
of some granite-related deposits.
Some high-level granites (i.e. those that are emplaced
at depths of ~8 km or less) represent the crystallized
remains of magmas that were associated with subvolcanic,
ore-generative, hydrothermal systems. Therefore, studies
of shallow granites, aimed at understanding magmatic
evolution and volatile phase exsolution at shallow levels,
are necessary if we are to understand ore genesis and
the deep inputs to present-day geothermal systems as
well as evaluate hazards associated with the volcanic
 Oxford University Press 1997
 JOURNAL OF PETROLOGY VOLUME 38
manifestations of these systems. Further, many studies
(e.g. Dilles & Einaudi, 1992) suggest that magmas are
not the sole source of ore metals in some porphyry, skarn
or high-temperature vein type deposits, so the precise
role played by orthomagmatic processes in the genesis
of some deposit types still remains to be elucidated. In
this paper I will review select results from my past
research, clarify what I perceive to be some problems
related to volatiles in magmas, and discuss some of the
problems that remain with regard to magmatic volatiles
and ore metals.
GENERAL NARRATIVE ON THE
DEVELOPMENT OF GRANITE-
RELATED ORE SYSTEMS AT THE
MAGMATIC STAGE
The formation of granite-related ore deposits is a by-
product of the irreversible magmatic transfer of heat and
mass from the Earth’s interior to its surface. Commonly,
the rise of magma (aluminosilicate liquid±oxide/silicate
crystals ± sulfide crystals or liquid ± hydrosaline
liquid±vapor bubbles) at epizonal levels leads to either
magmatic volatile phase (MVP) saturation if the system
is initially volatile phase undersaturated, or the further
exsolution of volatiles if the magma is already saturated
with respect to one or more volatile phases. The MVP
is thought to be a critical agent in ore formation because
of the affinity displayed by many ore metals for a volatile
phase relative to rock-forming minerals (see Candela &
Piccoli, 1995), and because of the high fluidity and
buoyancy of the MVP. These physical attributes allow
the volatile phase, in some circumstances, to accumulate
near and above suitable apical structures where it may
precipitate a significant proportion of its sparingly soluble
load. It should be noted that in the Earth’s epizone,
volatile saturation may occur early relative to crys-
tallization progress because of low prevailing load pres-
sure (approximated by the vertical component of stress).
Hence, volatile saturation at epizonal levels, commonly,
is neither late, nor near-solidus; that is, a large amount
of crystallization may occur under water-saturated con-
ditions (if water is the dominant volatile constituent in
the magma). A magma is said to be in a state of volatile
saturation when bubbles of vapor grow by mass transfer
from the melt, or are present in equilibrium with the
magma. The nucleation and growth of bubbles of gas
within the body of a liquid is referred to as boiling.
Boiling occurs when the sum of the vapor pressures of
the components of a liquid, e.g. R(Pivapor) is equal to the
load pressure; boiling can result from a decrease in the
load pressure or an increase in R(Pivapor). A decrease in
load pressure for a given vapor pressure of the dissolved
1620
NUMBER 12 DECEMBER 1997
constituents (at constant temperature and melt com-
position), results in what is referred to as first boiling.
First boiling is fairly common in magmatic systems, and
can occur as a result of magma ascent, or release of
magmatic pressure by mechanical failure of the magma
chamber. As many melts are known to possess con-
centrations of water on the order of 3–6 wt %, some
with CO2 concentrations of up to a few thousand ppm
(Lowenstern, 1994), partial pressures of magmatic gases
can sum to a few hundred MPa. For a melt with ~4 wt %
water and 800 ppm CO2, R(P H2O+P CO2 )~200 MPa (Low-
enstern, 1994). If the magma ascends from a depth of
~12 km, it will saturate with respect to a volatile phase
(first boiling) at ~8 km. Further, as water in a melt is an
important source of the entropy of mixing that stabilizes
felsic melts at temperatures below ~950°C, the decreasing
equilibrium concentration of water that results from first
boiling (but not from isobaric second boiling) can result
in irreversible crystallization. We will return to the ques-
tion of melt crystallization upon rise in the section of this
paper devoted to textures.
Volatile exsolution can also be brought about iso-
barically: if a liquid mixture of volatile and relatively
non-volatile components is cooled below its liquidus and
begin to crystallize phases that are dominantly anhydrous,
the mole fraction of the volatile components in the
mixture, and, concomitantly, R(Pivapor), will increase. In
this case, volatile saturation occurs upon cooling even
though R(Pivapor) tends to decrease with decreasing tem-
perature. This second mode of boiling, which is always
accompanied by crystallization of non-volatile com-
ponents and is isobaric in the limiting idealized case, is
called second boiling. As an example, let us consider the
crystallization of a vapor-undersaturated melt with
2 wt % water and 400 ppm CO2 (corresponding to
~1000 mmol H2O/kg of melt, and ~10 mmol CO2/kg
of melt, respectively) at a constant pressure of 200 MPa.
After 50% crystallization, the concentrations of water
and carbon dioxide have doubled, yielding, as before,
R(Pivapor)~200 MPa at ~800–900°C. Under these con-
ditions, the partial pressures of both water and carbon
dioxide are on the order of 100 MPa (Lowenstern, 1994),
and therefore the mole fractions of water and carbon
dioxide in the vapor phase are sub-equal. If no carbon
dioxide were present, the melt would not have saturated
with respect to a volatile phase until the water con-
centration had risen to ~6 wt % (at 67% crystallization
of the melt), showing that volatile saturation occurred
much earlier from any given aliquot of melt because of
an initial concentration of carbon dioxide of only 400
ppm. However, the amount of vapor that can be evolved
from a melt with only 800 ppm CO2 is severely limited,
as <0·1 wt % CO2 is present in the melt. It should be
noted that the solidus temperature does not increase
during second boiling. In fact, in a melt with a few
 CANDELA A REVIEW OF SHALLOW, ORE-RELATED GRANITES
hundred ppm carbon dioxide, the activity of water in-
creases with progressive exsolution of a CO2–H2O MVP
during second boiling, and the solidus temperature of
the magma actually drops during volatile exsolution. In
the first boiling case, the melt would saturate with an
H2O–CO2-rich volatile phase at 200 MPa vs 100 MPa
for the CO2-free system.
Given this background, we can consider the following
general scenario, which sets up the balance of the paper.
A vapor-undersaturated magma, rising in the crust, sat-
urates with respect to a volatile phase by first boiling.
The magma may of course reach the surface and erupt;
on the other hand, the magma ascent rate may be slowed
by external far-field stresses or by an increase in viscosity
because of crystallization or bubble production [however,
see Dingwell (1997)]. The change in viscosity of a magma
upon rise, as a result of crystallization, will be equal to
the rate of change in the viscosity of the magma as it
crystallizes, times the ratio of the rate of crystallization
relative to the rate at which the melt is rising. As magma
ascent slows, some crystallization will occur because
of irreversible heat loss to the magma’s cold epizonal
surroundings, and volatile exsolution will be driven, in-
creasingly, by crystallization (second boiling) rather than
by decompression (first boiling); we note, however, that
the adiabatic crystallization that occurs upon ascent,
driven by first boiling, will also further promote volatile
exsolution. Ponding of the increasingly viscous magma
may be aided by tectonic factors. In many compressional
settings, local dilational pull-aparts provide space for
magmatic intrusion. It is not clear whether internal
magmatic factors (e.g. the balance of buoyancy vs viscous
drag forces) or the rates of brittle strain of the crust, are
the limiting controls on the rate of magma ascent; most
likely, one or the other factor will be dominant. In cases
where the crustal strain rate is limiting, then magma may
rise only as rapidly as brittle deformation occurs, and
magma ascent and emplacement will be punctuated by
periods of complete crystallization of melt. Such a pluton
would be assembled over a period of time, and the
volatile phase saturation scenario outlined here would be
a quasi-periodic process (Candela & Blevin, 1995a), with
magmas sometimes emplaced in concentric fashion, and
at other times emplaced adjacent to previous magma
pulses (Vigneresse & Bouchez, 1997). Hence, whether a
magmatic–hydrothermal system is produced as a result
of a single pulse or multiple pulses of magma, and
whether the multiple pulses of magma produce a single
pluton or multiple intrusive phases, will be a complex
function of the rate of magma generation, upper-crustal
strain rate, and the time-integrated physical properties
of the magma (Hanson & Glazner, 1995). Depending
upon the mode and rate of magma emplacement, a
pluton may represent a single batch of magma, or it may
represent a time-integrated chamber, with the magma
1621
present at any time corresponding to only a small portion
of the total inferred pluton volume. The geometry of
volatile egress, therefore, in relation to the present-day
geometry of a given pluton, will depend upon the rate
of strain at the site of emplacement, and the rate of
magma supply.
In summary, we might expect a magma, after some
differentiation at depth, to experience volatile exsolution,
and further crystallization, upon ascent. As the magma
is emplaced, near-isobaric volatile exsolution will com-
mence. During the phase changes occurring at each
stage, ore metals and other magmatic constituents will
partition among the melt, mineral and volatile phases,
in accord with the appropriate partitioning and solubility
equilibria (as modified by kinetic constraints when ir-
reversibility of the magmatic processes is considered).
Of course, melts richer in volatile components can be
expected to reach MVP saturation earlier (early relative
to crystallization progress) than melts poorer in volatiles;
further, earlier volatile phase saturation is also expected
for hotter melts transported to shallower levels. As we
shall see, the timing of volatile saturation is critical
in our understanding of ore genesis according to the
orthomagmatic–hydrothermal model. In the next few
sections of this paper, I will discuss a number of factors
related to the magmatic stage of development of granite–
ore systems, including crystal–melt equilibria for ore
metals, water concentrations in the melt, and pressure
of emplacement. I will then proceed to discuss textures
in high-level granites and the processes of MVP egress,
followed by a narrative on the hydrothermal aspects of
the MVP exsolution process.
WATER CONCENTRATIONS, AND
PRESSURE OF EMPLACEMENT
Initial water concentrations with special
reference to the role of hornblende
The initial water concentration of a melt can only be
defined for a given initial melt composition. In many
cases, this is taken to be the composition of either the
granitic rock in question or some primitive member of
the suite. It should be noted also that the concentration
of water in the melt phase is different from the water
concentration in the magma. A magma with 2 wt %
water and 50% crystals, and which is volatile phase
undersaturated, contains a melt phase with 4 wt % water.
The composition of the bulk magma may correspond to
the composition of a true partial melt, a mixed melt,
melt plus restite, or a fractionated melt; postulating the
water content and bulk composition of inferred parental
magmas is, therefore, a difficult task.
Hornblende is a hydrous mineral that is present in
many tonalitic to granodioritic I-type igneous rocks. It is
 JOURNAL OF PETROLOGY VOLUME 38
not common in granites (sensu stricto) and the very highly
evolved and ‘specialized’ granites associated with many
lithophile metal deposits do not contain hornblende (how-
ever, it does occur as a phenocryst phase in some rhy-
olites). By definition, hornblende does not occur in S-
type granites (White & Chappell, 1977). For these systems,
studies of melt inclusions in phenocrysts (Thomas, 1994)
or water concentrations in cordierite (Carrington & Har-
ley, 1996) are yielding promising results with regard to
the estimation of water concentrations at the time of
trapping of the inclusions, and at the ‘water closure
temperature’ of cordierite, respectively. However, how
these water concentrations can be related to a particular
stage of crystallization and melt evolution still must be
addressed.
The use of hornblende as an indicator of water con-
centrations in melts, and as a geobarometer, has not
been discussed in a magmatic–hydrothermal context.
Given that both new phase equilibrium data on horn-
blende stability and new formulations of the amphibole
geobarometer have recently been published, the balance
of this section will be devoted to a discussion of horn-
blende. The presence of hornblende can yield information
on the minimum water concentration in melts with which
it is in equilibrium; below a certain fugacity of water
(for a given bulk magma composition and at a given
temperature and pressure), hornblende will decompose.
The limiting dehydration reaction (shown schematically)
for supersolidus hornblende stability, according to Ruth-
erford (1993), is:
melt + hornblende→low-Ca pyroxene +
plagioclase + Fe–Ti oxides.
According to experiments on the Mount St Helens dacite
(62–64 wt % SiO2) by Rutherford & Devine (1988), the
minimum water concentration in the melt needed to
stabilize hornblende at T=750°C, partial pressure of
water equal to total pressure, and f (O2)=NNO (nickel–
nickel oxide) + 1, is on the order of 4–4·5 wt %, or
water vapor pressures of 100–140 MPa. This yields a
minimum fugacity of water for hornblende stability, in
a magmatic bulk composition corresponding to Mount
St Helens dacite, of ~95 MPa (i.e. partial pressure of
water of 120 MPa and fugacity coefficient of ~0·8) at
750°C.
At about 900°C and a total pressure of 220 MPa,
Rutherford & Devine (1988) reported that hornblende is
destabilized in the presence of a CO2–H2O vapor at
mole fractions of water in the vapor less than ~0·6. That
is, hornblende was destabilized at partial pressures of
water below ~220 × 0·6≈130 MPa, or f (H2O)~120
MPa, given a fugacity coefficient for pure water ~0·9,
and assuming ideal mixing in the vapor at these tem-
peratures and pressures.
1622
NUMBER 12 DECEMBER 1997
In summary, hornblende stability is limited to fugacities
of water greater than ~95–120 MPa at temperatures of
750–900°C, at least for dacitic magmas at total pressures
on the order of 100–200 MPa, and for the oxygen
fugacities that prevail in many arc and I-type granite
magmas.
We can consider a hypothetical example to see how
these data may be used to estimate the initial magmatic
water concentration. The estimate of water con-
centrations in granitic rocks is fraught with many diffi-
culties, and this example is probably close to a best case
scenario. If a magma series, believed to be related by
fractional crystallization, has a porphyritic granite of
roughly dacitic composition (to conform to the ex-
periments of Rutherford and coworkers) with ~25%
phenocrysts, including a few percent hornblende, then
one might infer that hornblende crystallized during the
first 25% of the crystallization history of the system. This
suggests that the melt had a minimum initial con-
centration of water equal to 3–3·5 wt % near its liquidus
(yielding on the order of 4–4·6 wt % at 25% crys-
tallization). As the temperature of hornblende crys-
tallization in this case would be on the order of 900°C,
the higher figures for these water concentrations would
probably be more accurate; i.e. the best estimate for a
minimum initial water concentration would be 4±0·5
wt %. The presence of hornblende, and an estimate of
when it began to crystallize, is probably the best method,
currently, for the estimation of initial water con-
centrations in granitic systems.
Hornblende barometry
Methods to determine the pressure of crystallization of
hornblende have recently been summarized and discussed
by Anderson (1996) and Ague (1997). The essence of the
technique relies on the fact that the aluminum con-
centration in amphibole increases with increasing pres-
sure in the presence of a low-variance granitic phase
assemblage. The aluminum-bearing exchange com-
ponents responsible for the variation in the aluminum
concentration in amphibole appear to be the edenite
(NaAlSi–1) and Tschermak (Al2Mg–1Si–1) exchanges, which
operate on the additive tremolite component (the par-
gasite component comprises equal proportions of these
three components). Various calibrations for the Al-in-
hornblende geobarometer are available, but the interested
reader is referred to the papers by Anderson and Ague,
and references therein. In all but Ague’s method, the
barometer depends upon the presence of a large number
of mineral and fluid phases: quartz, plagioclase, K-
feldspar, biotite, hornblende, sphene, Fe–Ti
oxides + melt + vapor. However, Ague has identified
an equilibrium among the mineral phase components:
 CANDELA A REVIEW OF SHALLOW, ORE-RELATED GRANITES
Tremolite + Phlogopite + 2Anorthite + 2Albite=
2Pargasite + 6Quartz + K-feldspar (1)
which he used to explain the Al-in-hornblende barometric
effect. It should be noted that this equilibrium involves
no melt, no volatile species, and requires the presence of
only five minerals. The equilibrium is solved along with
the amphibole–plagioclase geothermometer to yield tem-
perature and pressure. Ague’s method shows promise,
and yields pressures similar to those from the previous
calibrations discussed by Anderson (1996), but with the
added benefit of not requiring the presence of as many
phases. However, before the amphibole barometer can
be used, some issues regarding the stability of hornblende
need to be clarified. The preceding discussion on the
stability of hornblende as a function of water fugacity
illustrates some of the problems related to the lower
pressure limit of hornblende stability, and the following
discussion considers, qualitatively, the limited range of
temperatures over which biotite and hornblende may
coexist.
As N. L. Bowen noted in his now famous ‘reaction
series’ early in this century, hornblende can react with
melt to yield biotite. According to Abbott (1981) and
Vythnal et al. (1991), ferromagnesian mineral-bearing
assemblages above the solidus evolve, with decreasing
temperature, from (shown schematically): melt + hbl,
to melt + hbl + bt (an even reaction involving direct
precipitation of biotite and hornblende), to the odd
reaction (melt + hbl→bt), and finally to melt + bt. I
express the equilibrium along the odd reaction curve as
5K-feldspar + 2H2O + 6Al2O3(melt) + 3Tremolite=
5Phlogopite + 6Anorthite + 12Quartz (2)
where mineral names represent phase components in the
melt or in the appropriate minerals. For a given activity
of SiO2, hornblende reacts to biotite down temperature
(i.e. water is on tremolite side of the equilibrium), and
as the activity ratio K-feldspar/anorthite increases. The
alumina activity in this expression represents the activity
of Al2O3 over and above that necessary to describe the
compositional variations in a subaluminous melt with
feldspar stoichiometry (Williams et al., 1997); the activity
of Al2O3 increases from zero as the aluminum saturation
index (ASI) of the melt increases beyond unity. The
crystallization of hornblende, at least at low pressures,
causes the ASI of the melt to rise because low-pressure
hornblende is low in aluminum. Generally, the as-
semblage hornblende + biotite occurs over only a portion
of the temperature range between the temperature of
appearance of hornblende and the solidus. Temperatures
and pressures estimated from hornblende geo-
thermometers and geobarometers may reflect the limited
temperatures of coexistence of the pertinent phases rather
than the temperature of the vapor-saturated solidus.
1623
Many granites (sensu lato) probably do not have
hornblende + biotite coexisting stably at the water-sat-
urated solidus, except for melts with low ratios of norm-
ative or to an, such as diorites.
In the next section, the significance of the estimated
water concentrations and pressures of emplacement will
be related to the efficiency with which ore metals can be
removed into the magmatic volatile phase during volatile
exsolution.
TIMING OF VOLATILE SATURATION
VERSUS CRYSTALLIZATION
PROGRESS
The competition for ore metals (and other elements)
between the MVP and the crystallizing magmatic mineral
assemblage during crystallization is a first-order control
on the efficiency with which metals can be removed from
magma into an evolving MVP, or ‘ore fluid’ (Candela &
Holland, 1986). The values of the pertinent crystal–melt
and MVP–melt partition coefficients, together with the
timing of crystallization relative to saturation with respect
to a water-rich MVP, is a major factor in determining
whether the MVP or the crystallizing phases take up a
higher proportion of a given ore metal. My coworkers
and I have presented detailed accounts of ore metal
partitioning in felsic melt–crystal systems elsewhere (see
Candela, 1992; Jugo, 1997), and the interested reader is
referred to those and other papers for values of the
pertinent partition coefficients. Here, I will discuss briefly
the qualitative effects of oxygen fugacity and crys-
tallization on the behavior of ore metals in granitic
systems.
Based on field associations of mineral deposits as-
sociated with I- versus S-type granites, Burnham &
Ohmoto (1980) suggested that oxygen fugacity may affect
the type of mineralization associated with granites. They
suggested that copper and molybdenum deposits were
associated with oxidized granites, and tin and tungsten
deposits were associated with more reduced systems.
Similar conclusions had been reached by Ishihara (1977).
Candela (1992) summarized experimental data for the
crystal–melt partitioning of copper, molybdenum and
tungsten, and showed that the experimental results are
consistent with the effects of magmatic oxygen fugacity.
Those results suggest that f (O2)-dependent crystal–melt
equilibria play a significant role in producing the observed
relationships. Similarly, Lehmann (1990) showed, based
on field data, that tin behaves as a compatible element
at low oxygen fugacities. Of the elements copper (Lynton
et al., 1993), gold ( Jugo, 1997), molybdenum and tungsten
(Candela & Bouton, 1990), and tin (Lehmann, 1990),
only tin and tungsten do not experience enhanced par-
titioning in favor of the melt phase with increasing
 JOURNAL OF PETROLOGY VOLUME 38
oxygen fugacity, either as a direct effect on the partition
coefficient (e.g. as in the case of molybdenum), or because
of destabilization of the host phase (as in the case of
copper or gold).
Molybdenum and tungsten can be partitioned into Ti-
bearing accessory phases, such as ilmenite, magnetite,
sphene and biotite. Both gold and copper are partitioned
strongly into certain accessory magmatic sulfides [which
are stable at f (O2)<NNO + 1; Carroll & Rutherford,
1985] and protracted crystallization of magmatic sulfide
minerals, or exsolution of an immiscible sulfide melt at
higher temperatures, can result in the strong depletion
of melts with respect to copper and gold. If a significant
amount of copper- and gold-bearing sulfide is removed
from the magmatic system before saturation of the melt
with respect to a volatile phase, the probability of for-
mation of a copper- or gold-rich ore or protore will
be significantly reduced. In I-type systems with oxygen
fugacities between quartz–fayalite–magnetite (QFM) and
NNO, a small amount of pyrrhotite crystallization might
remove a significant amount of copper, but not gold. For
some metals, such as copper and gold, that possess high
MVP–melt partition coefficients (see Candela et al., 1996),
the role of f (O2) is obviated in cases where hot, primitive
melts with high initial water concentrations are emplaced
in the uppermost levels of the Earth’s crust. Under
these conditions, volatile exsolution may occur before
significant fractionation of sulfides, regardless of the oxy-
gen fugacity. However, copper mineralization is usually
restricted to relatively oxidized magmas at shallow levels,
as even small amounts of pyrrhotite can remove sig-
nificant amounts of copper from a crystallizing melt
( Jugo, 1997). Candela & Holland (1986) suggested that
magmatic sulfides might, in some cases, react out of the
hypersolidus mineral assemblage upon saturation of a
magma with a volatile phase, as sulfur is removed from
the melt along with other exsolving volatiles. Further,
Keith et al. (1997) argued convincingly that metals se-
questered in sulfides can be repartitioned into the MVP
after volatile phase saturation, if magmatic sulfides are
crystallized at, or brought to, the level of emplacement
where volatile exsolution occurs. This process depends
on the proportion of ore metals in a magmatic system
that is partitioned into these phases during in situ crys-
tallization at the level of emplacement versus the pro-
portion of ore metals that is partitioned into crystallizing
phases at depths below the level of emplacement (Candela
& Piccoli, 1995) and remains at sites distal to MVP
exsolution. In the latter case, ore metals are certainly lost
from the system.
In general, early saturation with a water-rich volatile
phase, as has been suggested for porphyry copper deposits
(Candela & Holland, 1986; Candela, 1989) allows the
evolving magmatic volatile phase maximum access to
crystal-, or sulfide liquid-compatible ore metals, such as
1624
NUMBER 12 DECEMBER 1997
copper or gold, while they are still dissolved in the melt.
In a relatively dry magma that is crystallizing at a
relatively deep level, exsolution of a water-rich magmatic
volatile phase will occur late in magmatic crystallization
progress. Given that copper generally behaves as a com-
patible element, and molybdenum and tungsten are, for
the most part, more incompatible than copper, early
magmatic crystallization of phases capable of sequestering
large proportions of magmatic copper can lead to low
efficiencies of removal of copper into ore fluids relative
to molybdenum and tungsten (Candela & Holland, 1986;
Candela, 1989). The general zonation of shallow Cu
deposits and deeper Mo and W deposits in some terranes
supports this idea (Candela, 1992).
This analysis suggests that magmas possessing early-
formed low Al-amphibole (high initial water content melts
emplaced at shallow levels) should have a high probability
of forming high (Cu + Au)/(Mo + W) ores. Conversely,
magmas with late, high Al-hornblende might be as-
sociated with a lower potential for high (Cu + Au)/
(Mo + W) ore formation.
In summary, there are two prime effects on the effi-
ciency of ore metal removal from magmas: namely, the
sequestering of ore metals into crystallizing magmatic
phases, which is a function of both the oxygen fugacity and
the mineralogy of the fractionating assemblage (Candela,
1992); and the timing of volatile saturation relative to
crystallization (Candela & Holland, 1986), with early
volatile exsolution allowing ore metals, especially those
that are crystal compatible, to be available for partitioning
into the MVP.
GRANITE TEXTURES
It was recognized rather early in the study of granite-
related ores that the granitic rocks associated with min-
eralization can be texturally complex. Variations in tex-
ture are observed commonly within a given pluton, or
among the plutons of a suite. The textural variants
present may include porphyritic, aplitic, and pegmatitic
domains, miarolitic cavities, graphic and micrographic
texture, and dendritic, acicular and/or skeletal crystals,
in addition to normal (i.e. hypidiomorphic granular)
granitic texture. Variations in texture and morphology
in igneous systems are functions of the mineral in ques-
tion, the composition of the phase and the associated
melt, the physical history of the melt, the presence of
pre-existing phase boundaries (e.g. nuclei), ripening, post-
crystallization deformation, and the magnitude of un-
dercooling. For a given phase and a given melt com-
position, the magnitude of undercooling exerts a major
influence on the crystallization kinetics and crystal mor-
phology in the crystallizing system (Swanson, 1977).
Undercooling (DT ) is defined as the difference between
 CANDELA A REVIEW OF SHALLOW, ORE-RELATED GRANITES
the temperature at which the melt saturates with respect
to a mineral, and the temperature at which the mineral
actually nucleates and grows. The undercooling is the
driving force for crystal nucleation and growth; i.e. the
free energy of crystallization is a function of DT. Further,
crystallization textures evolve with increasing under-
cooling (e.g. a given mafic melt may crystallize yielding
a gabbroic texture at low undercooling, a diabasic texture
at intermediate undercooling and a basaltic texture at
high undercooling). Much has been made in the pet-
rological literature about the necessity of pre-existing
nuclei (heterogeneous nucleation) in the crystallization of
igneous systems; however, the variations found in crystal
number density in shallow granites, and as a function of
the degree of undercooling in experimental studies, show
that this concept is not a prime factor in understanding
the textures of shallow granites. Granitic rocks that are
emplaced at shallow levels, where undercooling is ex-
pected to vary significantly, show the greatest variety of
textures. Magmas at deeper crustal levels, or magmas in
regions of high heat flow, will cool more closely and
acquire an equilibrium texture. Therefore, most un-
deformed granitic rocks fall on a continuum from near-
equilibrium textures, which tell us very little about mag-
matic conditions, to texturally rich rocks (far from equi-
librium), which possess a historical record.
Porphyritic texture is reported commonly in shallow,
ore-associated granites. When the same minerals are
found among both the phenocrysts and the groundmass,
it is safe to ascribe this texture to two-stage cooling,
with the fine-grained groundmass produced by a sudden
increase in undercooling. For a given saturation tem-
perature for the phases in question, the undercooling
may be increased by lowering the actual temperature at
which the crystals grow. The movement of an MVP-
undersaturated magma into colder country rock, pre-
sumably by magma ascent to shallower levels, increases
the temperature difference between the country rock and
the magma. The increasing rate of heat flow out of the
magma, in turn, results in a more rapid decrease in the
temperature of the melt before crystals grow (i.e. higher
DT ) and therefore a finer grain size. On the other
hand, ascent of an MVP-saturated magma results in the
lowering of the equilibrium water concentration in the
melt. As the water concentration in the melt decreases
(first boiling), the undercooling increases because of the
rise in temperature required for equilibrium between, for
example, quartz + feldspar and the melt at the new,
lower water concentration in the melt. In principle,
whether or not the melt rises volatile undersaturated may
be discerned from the texture of magmatic quartz. The
presence of rounded cores in quartz (Ratajeski, 1995)
could indicate that at least some MVP-undersaturated
ascent has occurred, because quartz solubility increases
with decreasing pressure in water-undersaturated melts.
1625
Volatile saturation following volatile phase under-
saturated rise should induce disequilibrium growth tex-
tures (e.g. skeletal quartz growth) over the corroded
quartz cores. If the magma rises further, increasing
undercooling may result, and skeletal quartz may be
overgrown by more dendritic forms (see Fig. 1), either in
the groundmass, or as decorations around the pre-existing
phenocrysts (Ratajeski, 1995). Further, undercooling can
produce, in the course of crystallization of felsic melts:
the generation of quartz eyes [produced because of the
ease of quartz nucleation relative to feldspar, yielding
quartz phenocrysts (Piccoli et al., 1996)]; pegmatite (Lon-
don, 1992); graphic or micrographic intergrowths (Fenn,
1986); granophyric (skeletal quartz) texture (Swanson &
Fenn, 1986); aplite; variants of, or intermediates between
these textures; miarolitic cavities, isolated, or in-
terconnected (Candela & Blevin, 1995b).
Piccoli et al. (1996) suggested tentatively that increasing
undercooling in dikes of the Tuolumne Intrusive Suite
produced a sequence of dike textures ranging from nor-
mal granitic texture, to finer, more allotriomorphic vari-
ants, to pegmatite–aplite. Piccoli et al. (1996) also found
that graphic texture in pegmatites and the abundance of
pegmatite also increased with undercooling. Graphic or
micrographic texture is defined as a cellular quartz–
feldspar intergrowth (see Fig. 2). Micrographic texture
occurs commonly as a component of the texture described
otherwise as granophyric, fine grained, and/or aplitic.
Apparently, during growth of feldspar, a planar feldspar
interface degrades into a cellular micrographic quartz–
feldspar intergrowth, which, in turn, can pass into dend-
ritic quartz in a feldspar matrix (see Fig. 3). The term
granophyric is usually used to described the less regular
quartz–feldspar intergrowths found in aplites and fine-
grained groundmass; these textures usually comprise, in
part, skeletal quartz, quartz dendrites, irregular micro-
graphic patches or feathery feldspar. The formation of
skeletal quartz–quartz dendrites–granophyre, and
graphic or micrographic textures also require degrees of
undercooling beyond the range that obtains during the
production of normal granite texture.
Miarolitic cavities and pegmatite
Miarolitic cavities (see Fig. 4) comprise crystals of super-
solidus minerals that terminate in, and project into, a
quasi-spherical void (or mass of hydrothermally pre-
cipitated minerals). Further, miarolitic cavities and peg-
matites share a common textural characteristic: external
nucleation. External nucleation (London, 1992) is defined
as nucleation on a bounding surface of a system (as in a
typical, crustified, hydrothermal vein), rather than at a
point that is internal to the system (as in the crystallization
of an aplite or a normal-textured granitic rock). One can
 JOURNAL OF PETROLOGY VOLUME 38 NUMBER 12 DECEMBER 1997

Fig. 1. A skeletal quartz phenocryst in the Gulgui granite from the new England Orogen, Eastern Australia. The phenocryst acted as a nucleus
for the quartz dendrite that grew as part of the groundmass. (Feldspar is at extinction.) The photo was taken with crossed Nicols. The field of
view is 5 mm across.

Fig. 2. Large alkali feldspar crystals in the porphyritic Gulgui granite. The feldspar crystals are set in a finer-grained groundmass characterized
by micrographic quartz–feldspar intergrowths, and dendritic and skeletal quartz. The photo was taken with crossed Nicols. The field of view is
5 mm across.

readily identify crystals that have nucleated internally never apparent, and therefore is not a very useful concept
versus externally; however, whether internal nucleation when examining real rocks. External nucleation is the
has occurred heterogeneously or homogeneously is almost main characteristic of pegmatites. Because of the close

1626
 CANDELA A REVIEW OF SHALLOW, ORE-RELATED GRANITES

Fig. 3. A feldspar crystal (near extinction) from the Gulgui granite, with bright areas of quartz. Within the feldspar crystal, a regular micrographic
intergrowth degrades toward the upper right into a region containing skeletal quartz. The photo was taken with crossed Nicols. The field of
view is 5 mm across.

textural similarity between pegmatitic and miarolitic tex-
ture, a short digression on the pegmatite texture is in
order. Pegmatites are recognized by many geologists by
the fact that the crystals form encrustations along the
bounding surfaces of the body (again, as in a well-
crustified hydrothermal vein) rather than by grain size;
hence the term ‘micropegmatite’, which I think is an apt
term for the externally nucleated texture found in some
small dikes, or other minor bodies of externally nucleated
igneous minerals, where the crystals are <1 cm in length.
London et al. (1989) and London (1992) demonstrated
experimentally that pegmatitic textures do not require
water saturation for their formation, and also pointed
out that other experimental studies (e.g. Swanson, 1977)
in addition to their own have shown that feldspar growth
rates are lower at water saturation than under water-
undersaturated conditions. The average size of crystals
in a rock is inversely proportional to the number of
crystals in a given volume, and is proportional to the
ratio of the growth rate to the nucleation rate. The
variation in the ratio of growth to nucleation rates can
result from a multiplicity of causes, including the mag-
nitude of undercooling, and the build-up of the con-
centration of non-quartzofeldspathic components in the
melt, especially water, fluorine and phosphorus (London
et al., 1989; London, 1992). In my opinion, based on the
aforementioned experimental studies, my observations in
the field, and observations and interpretations reported
by others (e.g. Swanson & Fenn, 1986; London, 1992)
pegmatite and aplite both form, at least in part, as a result
of undercooling. Subtle variations in melt chemistry,
temperature, and undercooling affect the relationship
between growth and nucleation rates of the constituent
minerals, thereby determining whether pegmatite (dom-
inated by external nucleation of at least feldspar) or aplite
(dominated by internal nucleation) forms; however, aplite
seems to dominate with increasing undercooling. Gen-
erally, there can be multiple kinetic pathways to achieving
external nucleation, and pegmatitic texture may be gen-
erated in different environments for different reasons.
Miarolitic cavities share with pegmatite the characteristic
of external nucleation; however, this does not suggest
that miarolitic cavities, along with the many modes of
occurrence of pegmatite, share a common genesis. In-
deed, unless a pegmatite is miarolitic, there is no com-
pelling reason to suppose the melt was water saturated
during its crystallization (although many pegmatite
magmas may well have crystallized at or near water
saturation). Some pegmatites probably crystallize from
near-solidus melts that passively flow into sub-horizontal
voids created by the foundering of layers or mattes of
crystals. The space created in this way may be infilled
with nuclei-poor, relatively cool (i.e. near-solidus) melt,
wherein external nucleation of crystals may occur more
rapidly than internal nucleation. Similarly, pluton borders
may contain pegmatite (stockscheider when it is composed

1627
 JOURNAL OF PETROLOGY
Fig. 4. A miarolitic cavity in the Watergums granite of the Lachlan Fold Belt, Southeastern Australia. Larger crystals of mostly feldspar nucleated
externally on the surrounding finer-grained, skeletal quartz-bearing matrix. The photo was taken with crossed Nicols. The field of view is 5 mm
across.
primarily of quartz, feldspar and mica), not because of
intrusion of ‘late, water-saturated, magmas along the
contact’, but because of intermediate undercooling, or
because of dilational pumping of residual melt away from
preexisting crystals that normally would serve as nuclei
at lower, near-solidus temperatures.
Miarolitic cavities are generally taken as the best evi-
dence, found in plutonic rocks, for volatile phase sat-
uration in magmas. The crystals project into what is now
a void, and what was occupied by the magmatic volatile
phase. The paucity of internally nucleated crystals in
miarolitic cavities reflects, in part, the fact that silicate
melt did not exist in that volume, and only crystals
nucleating near the melt–MVP interface could grow into
the cavity. Miarolitic cavities are generally restricted,
though not entirely so, to epizonal granites, particularly
those that are finer grained. Miarolitic cavities are not
found in most granites, even though the overwhelming
majority of most granites have seen volatile saturation at
some stage in their history; that is, the absence of mi-
arolitic cavities from most granites tells us nothing about
whether those granites were saturated with a volatile
phase or not. These factors suggest the hypothesis that
the formation of miarolitic cavities requires bubble growth
during magma ascent and decompression, and (at least
partial) quenching of melt during that process. Vapor
bubbles that form during isobaric crystallization of a
water-saturated melt are probably small compared with
VOLUME 38 NUMBER 12 DECEMBER 1997
the size of most of the mineral grains in a normal, granite-
textured rock. Therefore, the overwhelming majority of
bubbles that form during the crystallization of a granite
are not preserved as miarolitic cavities. The probability
of forming miarolitic cavities increases with decreasing
crystal size because bubble size and crystal size approach
each other. Although bubbles can grow by coalescence
and by addition of volatile constituents from the de-
volatilizing melt, those processes can occur in all crys-
tallizing, devolatilizing granitic melts. However, in an
ascending, volatile-saturated melt, bubbles increase in
size by decompression. Together with the concomitant
increase in solidus temperature, the simultaneity of bubble
growth and crystallization driven by increasing under-
cooling results in the simultaneous development of mi-
arolitic cavities, micropegmatite, and micrographic and/
or dendrite-bearing (‘granophyric’) aplite (Candela &
Blevin, 1995b). The combination of larger bubbles and
smaller crystals allows for some small portion of the
volume of the magmatic volatile phase to be preserved
as miarolitic cavities. However, this textural preservation
is selective, and the volume of miarolitic cavities is
unrelated to the actual proportion of the MVP vs melt.
The absence of miarolitic cavities from most granites
indicates that decompression to pressures below the vapor
pressure of the melt did not occur to such an extent that
significant crystallization accompanied decompression
and bubble growth. When a relatively low crystallinity
1628
 CANDELA A REVIEW OF SHALLOW, ORE-RELATED GRANITES
magma decompresses and saturates with respect to a
volatile phase, and crystallizes significantly, then mi-
arolitic cavity formation can occur. For example, if a
melt with ~4 wt % water ascends to within about 4 km
of the Earth’s surface, then vapor bubbles begin to form
as the 100 MPa isobar is crossed, and miarolitic cavity
generation may begin thereafter during crystallization
and progressive rise of the melt to higher levels (as
long as the rate of ascent is sufficiently slow to allow
crystallization). As higher initial water contents obtain,
miarolitic cavity formation may occur at deeper levels in
the crust. If we accept the data from melt inclusions in
volcanic rocks, felsic melts can have vapor pressures as
high as a few hundreds of MPa. Therefore, formation of
miarolitic cavities by bubble growth and decompressive
crystallization will be limited, generally, to the uppermost
crust. As a generalization, the deeper the interval of
crystallization is in the crust, the lower the probability
of forming miarolitic cavities; miarolitic cavities should
become relatively uncommon at pressures of crys-
tallization of 300 MPa or greater. There is, however, no
maximum pressure for the formation of miarolitic cavities.
Many of the textures that have been ascribed to water
saturation or volatile exsolution, such as pegmatite–aplite,
granophyre, or graphic or micrographic textures, have
been produced experimentally at volatile-undersaturated
conditions (Fenn, 1986; Swanson & Fenn, 1986; London,
1992). Rather, as indicated herein and elsewhere, these
textures indicate that the magma experienced varying
degrees of undercooling, which may be related to the
temperature of the country rock and/or the exsolution
of a volatile phase upon ascent to higher levels in the
crust. For magmas of a given composition, water content
and crystallinity, emplaced at a given level in the epizone,
higher undercoolings will be produced in regions of lower
heat flow. This may be why many porphyry-Mo related
plutons possess highly variable textures, complete with
unidirectional solidification textures (some components
of which are magmatic, and possess dendritic, skeletal,
or pegmatitic components).
In summary, most of the textural variability found
in high-level and/or ore-related granites is probably
generated by various degrees of undercooling. This un-
dercooling may result in part from first boiling, so that
many of the observed textures are indirectly related to
volatile exsolution. However, many of these textures have
been produced experimentally in the absence of volatile
saturation, so saturation is not a necessary prerequisite
for the formation of pegmatite, aplite, ‘granophyre’ or
graphic textures. Miarolitic cavities, however, appear to
be an exception to the rule. Further, the generation of
undercooling by volatile phase exsolution is restricted to
first boiling. Granite textures are moot with regard to
volatile phase exsolution produced by second boiling, as
this process occurs without generating any substantive
1629
undercooling. Fine-grained rocks that crystallized during
second boiling owe their texture to undercooling pro-
duced by conductive heat loss at shallow levels or co-
incidental decompression that may have accompanied
second boiling. Further research, building upon the sem-
inal works of Shannon et al. (1982), Kirkham & Sinclair
(1988), and Lowenstern & Sinclair (1996), is required to
determine how these many granite textures relate to the
genesis of specific mineral deposits, and, more im-
portantly, how they differ among deposit types.
MVP egress and interconnected miarolitic
texture
Whereas magmatic vapor±brine may remove ore metals
from the apical regions of an intrusion into the overlying
hydrothermal system, magmatic transport of ore metals
to the apical regions of the magma chamber is essential,
as the volume of the apical magma, by itself, is rarely
sufficient to provide the full complement of ore metals
to an ore zone. How then does the magmatic volatile
phase gain egress from a magma? To a large extent, this
depends upon: whether the magma is saturated with
respect to a volatile phase upon ascent, and whether a
vapor, or vapor plus brine are present; the pressure
during volatile exsolution, and the volume proportion of
crystals:silicate liquid:brine:vapor; the rate of magma rise
after volatile saturation; the rate of cooling of the magma
at a given pressure; and many other factors. If the
magma is already saturated with a vapor phase upon
emplacement, and the crystallinity of the magma is low,
then some bubble rise and/or bubble-laden plume rise
is possible (Candela, 1991). Indeed, ascent of the magma
may be considered as the rise of a bubble-laden plume,
with ascent driven in part by the buoyancy imparted to
the magma by bubbles of vapor. Buoyancy imparted by
brine is lower in magnitude, but magmatic brines formed
at near-solidus temperatures and at ~100 MPa have
densities on the order of 1000 kg/m3, significantly lower
than the bulk density of magmas.
Rise of individual bubbles certainly occurs in magmas
during some stages of evolution after volatile saturation.
However, as the proportion of crystals increases, the
pathway for bubble rise becomes tortuous. Further,
bubble rise hypotheses suffer from the fact that volatile
exsolution, resulting from rapid crystallization or rapid
pressure reduction, probably produces many small
bubbles, rather than a few large bubbles. Smaller bubbles
have higher ratios of drag forces to buoyancy forces, and
therefore rise more slowly. Both of these factors are likely
to limit models that depend upon bubble rise in shallow
ore environments, except for large bubbles that form
during ascent in crystal-poor magma. In addition to the
rise of bubble-laden magma plumes, Candela (1991)
 JOURNAL OF PETROLOGY VOLUME 38
suggested that the MVP may advect through in-
terconnected regions of the volatile phase, allowing for
transport of hydrothermal ‘ore fluids’ through the
magma. Shinohara & Kazahaya (1995) have criticized
this hypothesis, suggesting that the model relies on ‘chan-
nels’ between bubbles that are unstable with respect to
coalescence, and the large channel itself would rise be-
cause of its buoyancy, relegating this mechanism to a
transient phenomenon. However, the model was pro-
posed as a phenomenological ‘fractal percolation cluster’
model, not as a detailed physical model. Indeed, co-
alescence of volumes of the MVP is not an unreasonable
mechanism for connectivity. Whether the percolation
cluster of volatile phase volumes rises as discrete packets,
or simply remains open, with the volatile phase flowing
through it, depends upon the rate of rise of the MVP
relative to the rate at which it is produced by exsolution.
Whether MVP expulsion is steady state, or occurs by
pulse-like slug flow, is beyond what is currently testable.
Although researchers may perform simplified calculations
meant to model the complex process of MVP exsolution,
the suggestion of specific textural geometries, which can
be tested in the field by ‘postdiction’, provides a definitive
test of process-related hypotheses. Whereas the per-
colation model proposes ‘interconnected regions of
MVP’, and because miarolitic cavities represent the best
evidence for the existence of a separate MVP in plutonic
environs, interconnected miarolitic cavities are wholly
consistent with the supersolidus MVP percolation hypo-
thesis. In Candela & Blevin (1995b), we have described,
illustrated, and documented, both macroscopically and
microscopically, the occurrence of interconnected mi-
arolitic cavities in shallow ore-related granites (see
Fig. 5a–c). These textures have been described in the
Ruby Creek phase of the Stanthorpe Granite (New
England Orogen, Eastern Australia), where they provide
strong field evidence for the supersolidus MVP per-
colation hypothesis. The MVP percolation model also
suggests that supersolidus percolation will occur only
under certain circumstances (relatively low pressures and
high water concentrations in the ‘initial’ magma); that
is, the process is critical, and does not ‘work’ in every
case, consistent with the observation that mineralization
is less probable at greater depths (a few kilobars). Igneous
structures supporting this model have also been found
in the Capitan Pluton, New Mexico (Dunbar et al., 1996).
Candela (1991) and Candela & Blevin (1995b) considered
this process to be a general one for shallow (pressures
~200 MPa and less) wet magmas (initial water con-
centrations close to saturation). We suggested that this
process operates during second boiling under the ap-
propriate conditions, and is sometimes ‘frozen in’ and
preserved when plumes of magma, containing in-
terconnected channels, become upwardly mobile because
of buoyancy. High-level granites exhibiting this texture
1630
NUMBER 12 DECEMBER 1997
should be considered ‘prospective’ from an exploration
viewpoint.
GENERAL NARRATIVE ON THE
DEVELOPMENT OF GRANITE-
RELATED ORE SYSTEMS AT THE
HYDROTHERMAL STAGE
During crystallization of an MVP-saturated melt, pressure
may increase above the total load pressure because of
the exsolution of the low-density vapor phase. Burnham
(1979) suggested that this excess pressure may cause
failure of the country rock leading to flow of fluids (melt,
magma or MVP), or even to volcanic eruption. However,
the preferred orientations found in many mineralized
systems suggested that these near-field stresses are coupled
strongly to the far-field stress regime, producing a com-
plex geometry of fractures that may vary from regionally
controlled to those that are related to the geometry of
the pluton. Venting at more than one locus within a
large chamber creating apophyses, veins, or vein dikes
(or even subvolcanic necks) will lower the prevailing total
pressure of the chamber to lithostatic or lower (if the
boundary region between the lithostatically and hydro-
statically pressured systems is breached), possibly trig-
gering volatile exsolution by first boiling and groundmass
crystallization. If the decrease in pressure is large (tens
of MPa) chamber failure can provide a large percolation
network (by increasing the molar volume of a vapor
phase). Under these conditions melt flow may also occur,
with crystals left behind. This hydrodynamic dispersion
of melt and crystals may, for example, produce non-
miarolitic, crowded and overcrowded porphyries of mon-
zodioritic composition, on one hand, and more highly
fractionated quartz-rich, and possibly miarolitic rocks
elsewhere down pressure in the system.
During first or second boiling at low (<150 MPa)
pressures, a magma may saturate with respect to a vapor,
a brine (Webster, 1997), or both, depending upon the
ratio of chlorine to water in the magma, the initial water
concentration, and the pressure of crystallization (Metrich
& Rutherford, 1992; Webster, 1992; Shinohara, 1994;
Candela & Piccoli, 1995). When two volatile phases
exsolve they may be transported through the magma
toward the top of the chamber by a variety of mechanisms
(Candela, 1991; Candela & Blevin, 1995a, b). However,
the dynamics of vapor movement through a magma is
likely to be wholly different from that of a brine. MVP
percolation will probably be dominated by vapor. The
vapor, brine and melt will tend to stratify gravitationally,
producing regions within the upper reaches of the magma
chamber and the superjacent subsolidus hydrothermal
system enriched in the lower-density aqueous vapor
 CANDELA A REVIEW OF SHALLOW, ORE-RELATED GRANITES
Fig. 5. (a) Interconnected miarolitic texture. Ruby Creek phase of the
Stanthorpe granite, New England Orogen, Eastern Australia. The two-
domain nature of the texture should be noted. The coarser material is
externally nucleated quartz and feldspar that connect miarolitic cavities
(now mostly filled with subsolidus quartz). The finer-grained material
is aplitic and contains many skeletal crystals. Also noteworthy is the
high degree of connectivity of the coarser domain, referred to as
interconnected miarolitic texture (Candela & Blevin, 1995b). Field of
view is 14·5 cm across. (b) Close-up view of (a), showing miarolitic
cavities and their interconnections. Field of view is 9 cm across. (c)
Miarolitic cavity in the Ruby Creek phase of the Stanthorpe granite.
Quartz, plagioclase and alkali feldspar are visible. The photo was taken
with crossed Nicols. The field of view is 2·5 mm across.
1631
phase. The higher-density brine may be trapped within
the crystallizing melt below. Because of its high density
relative to both the vapor and the surrounding hydro-
logical regime, the brine phase would tend to resist
migration away from the immediate vicinity of the magma
chamber (Henley & McNabb, 1978). Brine may probably
remain dispersed. In fact, early, deep generation of brine
within the pluton may not be optimum for ore formation.
If both brine and vapor form simultaneously, the vapor
from different portions of the crystallization interval may
mix, and the only meaningful vapor compositions would
be averaged compositions. The vapor volumes would
form percolation clusters (under the appropriate con-
ditions of pressure and initial water concentration of the
melt) with vapor streaming toward the apical regions of
the pluton. Brine bubbles may not form percolation
clusters (their mass fraction is too low, and their density
is too high); most probably, they sit where they form.
They would be buoyant in the magma, rising small
distances tortuously; a large proportion of brine probably
becomes trapped in the pluton, only to be removed by
circulating exogeneous fluids. If, on the other hand, the
brine forms late (e.g. near the solidus), a cupola near the
maximum pressure at which brine can form at near-
solidus temperatures (~120 MPa) would be the locus of
brine formation and storage; otherwise a brine dispersed
through the crystallized pluton could become just another
phase of plutonic dispersal of chloride-complexed ore
metals. It should be noted that in the magma, brine is at
a pressure of ‘lithostatic’ or greater, and upon catastrophic
failure of the roof, the brine may be propelled into the
lower reaches of the overlying hydrologic system (see
Rye, 1993). The brine may indeed re-equilibrate with
the subsolidus pluton at temperatures of 400–700°C,
profoundly altering its character (Frank, 1996). The re-
gion between the carapace and the brittle geothermal
system above is a quasiplastic region where meteoric
water is scarce, and fractures heal rapidly. As Fournier
(1987) has pointed out, and as Rye and later Hedenquist
& Lowenstern (1994) have emphasized, the continual
accession of vapor plumes from the deep magmatic zone
into the overlying geothermal system may be punctuated
by catastrophic periodic pulses of more saline brine
that are driven upward from the upper reaches of the
magmatic system by the large pressure gradients that are
produced when the brittle–ductile transition is breached.
Hence, vapor or supercritical gas on one hand, and brine
on the other hand, may be transported through the
magmatic chamber and into the overlying hydrothermal
system in very different ways. Further, it is worth em-
phasizing that chlorinity, and hence the metal-carrying
capacity of vapor and brine, can vary strongly over rather
small regions of P–T–X space, especially for bulk MVP
compositions that are near-critical at the pressures and
temperatures close to the water-saturated granite solidus,
 JOURNAL OF PETROLOGY VOLUME 38
and the proportion of ore substance transported in gas-
like or liquid-like form can vary appreciably from one
system to another.
Miarolitic textures in high-level granites may grud-
gingly yield hints as to means of egress of MVPs from
crystallizing plutons, and other granite textures may,
indirectly, bear on the complex relations among variables
such as pressure, volatile concentrations in the melt,
ascent rate, and the local thermal structure of the crust.
The actual concentrations of ore metals, sulfur, HCl,
etc., in the MVP that enter the deep parts of ore, protore,
or geothermal systems, depend upon how magmatic
evolution has affected critical parameters such as: the
concentration of H2O, CO2, HCl, and total sulfur in the
magma; the f (O2) of the magma; when, relative to volatile
saturation, the magma saturates with sulfides or other
phases capable of sequestering ore metals; and where
those crystalline materials reside relative to the regions
of ‘ore-fluid’ exsolution. Fortunately, experimental work
on granite textures (London, 1992), ore metals (Candela,
1992; Williams et al., 1995), phase equilibria in igneous
systems (Rutherford, 1993), apatite and estimates of mag-
matic chloride concentrations (Piccoli & Candela, 1994),
granite geobarometry (see Anderson, 1996), ore metal
sequestering in sphene (Piccoli et al., 1994), sulfide and
oxide chemistry (see Whitney, 1985; Stimac & Hickmott,
1995; Jugo, 1997), and many other studies are beginning
to yield, albeit stubbornly, information on many of the
intensive parameters important in determining the chem-
ical and physical properties of devolatilizing magmatic
systems and the associated volatile phases that give rise
to granite-related ore systems.
ACKNOWLEDGEMENTS
This research benefited from the support of the NSF
(Grant EAR9506631) and the US DOE (Grant DE-
FG0790ID13025). I would like to thank the many past
and present members of the Laboratory for Mineral
Deposits Research, University of Maryland (their papers
are cited in this work) who have contributed to the overall
model presented here, and especially Dr Philip Piccoli.
Thanks are due to Professor Jay Ague, who kindly made
a preprint of his paper available to me. I would also like
to thank my colleagues in the Geology Department at
the Australian National University, especially Dr Phillip
Blevin and Professor Bruce Chappell, who have not only
supported my many forays into Eastern Australia, but
have also shared their knowledge and wisdom related to
granites and mineralization, much of which inspired the
material presented in the latter half of this paper. I am
grateful to Professors Bernd Lehmann and Jeffrey Keith,
and an anonymous reviewer, for insightful and con-
structive reviews of the manuscript.
1632
NUMBER 12 DECEMBER 1997
REFERENCES
Abbott, R. N., Jr, 1981. AFM liquidus projections for granitic magmas,
with special reference to hornblende, biotite, and garnet. Canadian
Mineralogist 19, 103–110.
Ague, J. J., 1997. Thermodynamic calculation of emplacement pressures
for batholithic rocks, California: implications for the Al-in-horn-
blende barometer. Geology 25, 563–566.
Anderson, J. L., 1996. Status of thermobarometry in granitic batholiths.
Transactions of the Royal Society of Edinburgh: Earth Sciences 87, 125–138.
Burnham, C. W., 1979. Magmas and hydrothermal fluids. In: Barnes,
H. L. (ed.) Geochemistry of Hydrothermal Ore Deposits, 2nd edn. New
York: John Wiley, pp. 71–136.
Burnham, C.W. & Ohmoto, H., 1980. Late-state processes of felsic
magmatism. Society of Mining Geologists of Japan, Special Issue 8, 1–11.
Candela, P. A., 1989. Magmas, volatiles and metallogenesis. In: Nal-
drett, A. & Whitney, J. (eds) Ore Deposition Associated with Magmas.
Reviews in Economic Geology, Society of Economic Geologists 4, 223–233.
Candela, P. A., 1991. Physics of aqueous phase exsolution in plutonic
environments. American Mineralogist 76, 1081–1091.
Candela, P. A., 1992. Controls on ore metal ratios in granite-related ore
systems: an experimental and computational approach. Transactions of
the Royal Society of Edinburgh: Earth Sciences 83, 317–326.
Candela, P. A. & Blevin, P. L., 1995a. Physical and chemical magmatic
controls on the size of magmatic–hydrothermal ore deposits. In:
Clark, A. H. (ed.) Giant Ore Deposits, II. Kingston: Queens University,
pp. 2–37.
Candela, P. A. & Blevin, P. L., 1995b. Do some miarolitic granites
preserve evidence of magmatic volatile phase permeability? Economic
Geology 90, 2310–2316.
Candela, P. A. & Bouton, S. L., 1990. The influence of oxygen fugacity
on tungsten and molybdenum partitioning between silicate melts
and ilmenite. Economic Geology 85, 633–640.
Candela, P. A. & Holland, H. D., 1986. A mass transfer model for
copper and molybdenum in magmatic hydrothermal systems: the
origin of porphyry-type ore deposits. Economic Geology 81, 1–19.
Candela, P. A. & Piccoli, P. M., 1995. Model ore–metal partitioning
from melts into vapor and vapor/brine mixtures. In: Thompson, J.
F. H. (ed.) Magmas Fluids and Ore Deposits. Mineralogical Association of
Canada Short Course 23, 101–128.
Candela, P. A., Piccoli, P. M. & Williams, T. J., 1996. Preliminary
study of gold partitioning in a low-sulfur, high oxygen fugacity melt/
volatile phase system. Geological Society of America, Abstracts with Programs
28, A-402.
Carrington, D. P. & Harley, S. L., 1996. Cordierite as a monitor of
fluid and melt H2O contents in the lower crust: an experimental
calibration. Geology 24, 577–672.
Carroll, M. R. & Rutherford, M. J., 1985. Sulfide and sulfate saturation
in hydrous silicate melts. Journal of Geophysical Research 90, C601–C612.
Dilles, J. H. & Einaudi, M. T., 1992. Wall-rock alteration and hydro-
thermal flow paths about the Ann-Mason porphyry copper deposit,
Nevada—a 6-km vertical reconstruction. Economic Geology 87, 963–
2001.
Dingwell, D. B., 1997. The brittle–ductile transition in high-level
granitic magmas: material constraints. Journal of Petrology 38, 000–000.
Dunbar, N. W., Campbell, A. R. & Candela, P. A., 1996. Physical,
chemical, and mineralogical evidence for magmatic fluid migration
within the Capitan pluton, southeastern New Mexico. Geological
Society of America Bulletin 108, 318–333.
Fenn, P. M., 1986. On the origin of graphic granite. American Mineralogist
71, 325–330.
Fournier, R. O., 1987. Conceptual models of brine evolution. US
Geological Survey Professional Paper 1350(2), 45–61.
 CANDELA A REVIEW OF SHALLOW, ORE-RELATED GRANITES
Frank, M. R., Candela, P. A. & Piccolo, P. M., 1996. Thermodynamics
of alteration equilibria in a high salinity quartz-saturated portion of
the system Al2O3–SiO2–H2O–KCl–KH–1. Geological Society of America,
Abstracts with Programs 28, A-153.
Hanson, R. B. & Glazner, A. F., 1995. Thermal requirements for
extensional emplacement of granitoids. Geology 23, 213–216.
Hedenquist, J. W. & Lowenstern, J. B., 1994. The role of magmas in
the formation of hydrothermal ore deposits. Nature 370, 519–527.
Henley, R. W. & McNabb, A., 1978. Magmatic vapor plumes and
ground-water interaction in porphyry copper emplacement. Economic
Geology 73, 1–20.
Ishihara, S., 1977. The magnetite-series and ilmenite-series granitic
rocks. Mining Geology 27, 293–305.
Jugo, P., 1997. Experimental study of the behaviour of copper and
gold in the system haplogranite–CuFeS2–FeS–Au–H2O–HCl–O2–S2
at 850°C and 100 MPa. M.Sc. Thesis, University of Maryland,
College Park.
Keith, J. D., Whitney, J. A., Hattori, K., Ballantyne, G. H., Christiansen,
E. H., Barr, D. L., Cannan, T. M. & Hook, C. J., 1997. The role
of magmatic sulphides and mafic alkaline magmas in the Bingham
and Tintic mining districts, Utah. Journal of Petrology 38, 1679–1690.
Kirkham, R. V. & Sinclair, W. D., 1988. Comb quartz layers in felsic
intrusions and their relationship to porphyry deposits. In: Taylor,
R. P. & Strong, D. F. (eds) Recent Advances in the Geology of Granite-
related Mineral Deposits. Canadian Institute of Mining 31, 50–71.
Lehmann, B., 1990. Metallogeny of Tin. Springer-Verlag: Berlin.
London, D., 1992. The application of experimental petrology to the
genesis and crystallization of granitic pegmatites. Canadian Mineralogist
30, 499–540.
London, D., Morgan VI, G. B. & Hervig, R. L., 1989. Vapor-
undersaturated experiments with Macusani glass + H2O at 200
MPa, and the internal differentiation of granitic pegmatites. Con-
tributions to Mineralogy and Petrology 102, 1–17.
Lowenstern, J. B., 1994. Dissolved volatile concentrations in an ore-
forming magma. Geology 22, 893–896.
Lowenstern, J. B. & Sinclair, W. D., 1996. Exsolved magmatic fluid
and its role in the formation of combined layered quartz at the
Cretaceous Logtung W–Ma deposit, Yukon Territory, Canada.
Transactions of the Royal Society of Edinburgh: Earth Sciences 87, 291–304.
Lynton, S. A., Candela, P. A. & Piccoli, P. M., 1993. Experimental
study of the partitioning of copper between pyrrhotite and a high-
silica rhyolitic melt. Economic Geology 88, 901–915.
Metrich, N. & Rutherford, M. J., 1992. Experimental study of chlorine
behavior in hydrous silicic melts. Geochimica et Cosmochimica Acta 56,
607–616.
Piccoli, P. M. & Candela, P. A., 1994. Apatite in felsic rocks: a model
for the estimation on initial halogen concentrations in the Bishop
Tuff (Long Valley) and Tuolumne Intrusive Suite (Sierra Nevada
Batholith) magmas. American Journal of Science 294, 92–135.
Piccoli, P. M., Candela, P. A., Jugo, P. J. & Frank, M. R., 1996.
Contrasting syn–late magmatic intrusive behavior of aplite dikes in
the Tuolumne Intrusive Suite, California: implications for magma
rheology. Cordilleran Section of the Geological Society of America (a symposium
in honor of Paul Bateman) 28, 101.
Piccoli, P., Candela, P. & Rivers, M., 1994. Titanite microchemistry
and the interpretation of magmatic and hydrothermal processes in
granitic systems. Geological Society of America, Abstracts with Programs 26,
499.
1633
Ratajeski, K., 1995. Estimation of initial and saturation water con-
centrations for three granitic plutons in the North–Central Great
Basin, Nevada. M.Sc. Thesis, University of Maryland, College Park.
Rutherford, M. J., 1993. Experimental petrology applied to volcanic
processes. EOS Transactions, American Geophysical Union 74, 49–55.
Rutherford, M. J. & Devine, J. D., 1988. The May 18, 1990 eruption
of Mt. St. Helens: 3, Stability and chemistry of amphibole in the
magma chamber. Journal of Geophysical Research 93, 949–959.
Rye, R. O., 1993. The evolution of magmatic fluids in the epithermal
environment: the stable isotope perspective. Economic Geology 88,
733–753.
Shannon, J. R., Walker, B. M., Carten, R. B. & Geraghty, E. P.,
1982. Unidirectional solidification textures and their significance in
determining relative ages of intrusions at the Henderson Mine,
Colorado. Geology 10, 293–297.
Shinohara, H., 1994. Exsolution of immiscible vapor and liquid phases
from a crystallizing silicate melt: implications for chlorine and metal
transport. Geochimica et Cosmochimica Acta 58, 5215–5221.
Shinohara, H. & Kazahaya, K., 1995. Degassing processes related to
magma chamber crystallization. In: Thompson, J. F. H. (ed.) Magmas
Fluids and Ore Deposits. Mineralogical Association of Canada Short Course
Volume 23, 47–70.
Stimic, J. & Hickmott, D., 1995. Ore metal partitioning in intermediate
to silicic magmas: PIXE results on natural mineral/melt assemblages.
In: Clark, A. H. (ed.) Giant Ore Deposits, II. Kingston: Queens
University, pp. 182–220.
Swanson, S. E., 1977. Relation of nucleation and crystal growth rate
to the development of granitic textures. American Mineralogist 62,
966–978.
Swanson, S. E. & Fenn, P. M., 1986. Quartz crystallization in igneous
rocks. American Mineralogist 71, 331–342.
Thomas, R., 1994. Estimation of the viscosity and water content of
silicate melts from melt inclusion data. European Journal of Mineralogy
6, 513–536.
Vigneresse, J. L. & Bouchez, J. L., 1997. Successive granitic magma
batches during pluton emplacement: the case of Cabeza de Araya
(Spain). Journal of Petrology 38, 1767–1776.
Vythnal, C. R., McSween, H. Y., Jr & Speer, J. A., 1991. Hornblende
chemistry in southern Appalachian granitoids: implications for alu-
minum hornblende thermobarometry and magmatic apatite. American
Mineralogist 76, 176–188.
Webster, J. D., 1992. Fluid–melt interactions involving Cl-rich granites:
experimental study from 2 to 8 bar. Geochimica et Cosmochimica Acta
56, 679–687.
Webster, J. D., 1997. Chloride solubility in felsic melts and the role of
chloride in magmatic degassing. Journal of Petrology 38, 1793–1807.
White, A. J. R. & Chappell, B. W., 1977. Ultrametamorphism and
granitoid genesis. Tectonophysics 43, 7–22.
Whitney, J. A., 1985. Fugacities of sulfurous gases in pyrrhotite-bearing
silicic magmas. American Mineralogist 69, 69–78.
Williams, T. J., Candela, P. A. & Piccoli, P. M., 1995. The partitioning
of copper between silicate melts and two-phase aqueous fluids: an
experimental investigation at 1 kilobar, 800°C and 0·5 kilobar,
850°C. Contributions to Mineralogy and Petrology 121, 388–399.
Williams, T. J., Candela, P. A. & Piccoli, P. M., 1997. Hydrogen–alkali
exchange between silicate melts and two-phase aqueous fluids: an
experimental investigation. Contributions to Mineralogy and Petrology 128,
114–126.