JOURNAL OF PETROLOGY VOLUME 41 NUMBER 9 PAGES 1439–1453 2000

Geochemical Constraints from Zoned

Hydrothermal Tourmalines on Fluid
Evolution and Sn Mineralization:
an Example from Fault Breccias at Roche,
SW England

B. J. WILLIAMSON1∗, J. SPRATT2, J. T. ADAMS2, A. G. TINDLE3 AND

C. J. STANLEY2
1
DEPARTMENT OF EARTH SCIENCES, UNIVERSITY OF BRISTOL, WILLS MEMORIAL BUILDING, QUEEN’S ROAD,
BRISTOL BS8 1RJ, UK
2
DEPARTMENT OF MINERALOGY, THE NATURAL HISTORY MUSEUM, CROMWELL ROAD, LONDON SW7 5BD,
UK
3
DEPARTMENT OF EARTH SCIENCES, THE OPEN UNIVERSITY, MILTON KEYNES MK7 6AA, UK

RECEIVED JANUARY 26, 1999; REVISED TYPESCRIPT ACCEPTED FEBRUARY 7, 2000

Hydrothermal fluid evolution north of the St Austell granite, southwest
England, has been studied through geochemical analysis of tourmaline
from a fault breccia of <2 cm width within massive quartz–
tourmaline rocks at Roche. Brecciated tourmaline grains have
overgrowths of <400
m width [Fe/(Fe + Mg) = 0·31–0·99]
with four chemically distinct zones (1–4, towards the margins).
Variations in overgrowth composition were caused by episodic mixing
between Mg-, Al-rich magmatic hydrothermal fluids (dominant in
zone 1), with an increasing component of more oxidizing, Fe-rich
formation waters (zones 2 and 4). More oxidizing conditions are
supported by high Sn contents in zone 2 (<0·35 wt %), with Sn
probably present as Sn4+ rather than Sn2+, the usual form in
hydrothermal fluids. From X-ray maps, zones 1 and 3 occur
exclusively as overgrowths on pre-existing grains, indicating that
overgrowth formation was kinetically favoured over tourmaline
nucleation. In zones 2 and 4, nucleation and growth occurred,
possibly as a result of supersaturation with respect to tourmaline
during increased mixing with formation waters. Tourmaline is
associated with the main episode of mineralization in many important
mineral deposits, often unaffected by alteration. This method of
studying hydrothermal fluid evolution may therefore have uses in
exploration, particularly for tourmaline-breccia-hosted ores in Cu-
porphyry deposits.
KEY WORDS: breccia; Cornwall; hydrothermal; tin; tourmaline
INTRODUCTION
The nature and origin of hydrothermal fluids responsible
for tin–tungsten mineralization in SW England have
been the subject of numerous mineralogical (e.g. Farmer
et al., 1991; London & Manning, 1995; Smith &
Yardley, 1996; Williamson et al., 1997) and fluid
inclusion studies ( Jackson et al., 1982; Alderton &
Rankin, 1983; Shepherd et al., 1985; Bottrell & Yardley,
1988; Wilkinson et al., 1994, 1995; Smith et al., 1996;
Smith & Yardley, 1996). However, there are many
aspects of their evolution that remain poorly understood.
One of the most crucial of these is the source

∗Corresponding author. Telephone: +44-117-954-5235. Fax: +44-

117-925-3385. e-mail: Ben.Williamson@bristol.ac.uk  Oxford University Press 2000
 JOURNAL OF PETROLOGY VOLUME 41
and behaviour of elements involved in mineralization,
particularly whether they had a magmatic source or
were scavenged from crustal rocks by convective
circulation of magmatic hydrothermal fluids and/or
formation waters. What is also poorly understood is
the strong association between tourmalinization and
cassiterite mineralization. No direct role has been
identified for B and F in the formation of the Sn
deposits (Pollard et al., 1987; Taylor & Wall, 1993)
although it has been suggested that their presence in
granite magmas lowers the solidus temperature to
>650°C (Pichavant & Manning, 1984), allowing time
for more extensive fractionation and partitioning of
ore-forming elements into hydrothermal fluids (Pollard
et al., 1987).
Tourmaline is a particularly useful mineral in the study
of ore-forming fluids as it occurs in a wide variety of ore
deposits (e.g. Slack, 1982, 1996; Slack et al., 1993; London
& Manning, 1995; Griffin et al., 1996). It is also stable
during most types of weathering processes and potentially
up to upper amphibolite facies [reviewed by Henry &
Dutrow (1996)], limiting problems with later alteration,
which can be extremely extensive in mineral deposits. In
addition, it has a complex chemistry, with a number of
different site occupancies (e.g. Burt, 1989; London &
Manning, 1995; Hawthorne, 1996; Jiang et al., 1998),
which can accommodate a chemically varied suite of
ore-forming elements. Its trace element composition may
therefore provide a tool in exploration [see review by
Slack (1996)]. This was recently demonstrated by a world-
wide study of massive sulphide deposits, which identified
a good correlation between base metal proportions in
tourmaline and the bulk composition of ore being mined
(Griffin et al., 1996).
There have been several previous studies of tourmaline
in SW England. These have mainly focused on the
discrimination of different genetic environments using
‘major element’ variations (Manning, 1991; London &
Manning, 1995) and on the source of B using B isotopes
(Smith & Yardley, 1996). There have also been useful
trace element studies on tourmaline from different settings
in SW England. Li concentrations in tourmaline (and
other minerals) from granites and a wide variety of other
rock types [including Roche massive quartz–tourmaline
(MQT) rock] have been determined by ion microprobe
(Wilson & Long, 1983). Power (1968) has provided trace
element data (including Sn) for a similar variety of rocks.
There have, however, been no detailed studies on major
and trace element variations within zoned hydrothermal
tourmalines and the bearing of this on the evolution of
ore-forming fluids. Hydrothermal breccias at Roche are
ideal for such a study as they formed within MQT rocks,
which consist almost entirely of quartz and tourmaline,
the latter showing strong mechanical and chemical sta-
bility (Slack, 1982; Henry & Dutrow, 1996). This will
1440
NUMBER 9 SEPTEMBER 2000
have greatly limited the extent of wall-rock reaction,
which may otherwise have altered the chemical and
physical nature of the hydrothermal fluids. The com-
positions of hydrothermal vein minerals in the breccia
are therefore likely to yield valuable information on the
nature of the fluids from which they formed.
GEOLOGICAL SETTING
Roche Rock forms an isolated stock of MQT rock within
Lower Devonian, mainly calcareous slate, Meadfoot Bed
country rocks (Fig. 1). It lies >500 m north of the
northern margin of the St Austell granite, the smallest
of the five main apophyses of the Hercynian batholith
of SW England. The St Austell granite is a highly complex
body consisting of six main granite types, many of which
show extensive late magmatic and hydrothermal al-
teration (Manning et al., 1996). Granite magmatism in
the area occurred over an extended time period. The
main biotite granites have yielded an age, by Rb–Sr
whole-rock isochron, of 285 ± 4 Ma and the Brannel
elvan, a nearby late granite porphyry dyke, an age of
270 ± 9 Ma (Darbyshire & Shepherd, 1985). Present
outcrop is considered to have been close to the roof of
the intrusion as indicated by the presence of numerous
pegmatites occurring as sheets and containing abundant
miarolitic cavities carrying quartz, tourmaline, zinnwal-
dite, topaz and a wide range of other phases (Manning,
1983; Weiss, 1989). The St Austell granite contains
abundant cross-cutting tourmaline fissure veins and stock-
works, which are often mineralized, containing amongst
others cassiterite, stannite and wolframite. The veins are
commonly bordered by greisens of <10 cm width, which
have been dated in the Goonbarrow pit, >2 km SE of
Roche, at 273·6 ± 1·1 Ma (K–Ar, Bray & Spooner,
1983) and 276·8 ± 1·2 Ma ( 40Ar/39Ar, Chesley et al.,
1993).
MQT rocks mainly crop out at the margins of the
western part of the St Austell granite. In the past, probably
from observations made in long since disused un-
derground workings, these bodies have been described
as forming elongate, tin-bearing stripes proceeding from
the main mass of St Austell granite (De La Beche, 1839;
Collins, 1878). Today, their contacts with the granites
and surrounding country rocks cannot be seen. The
origin of MQT rocks, and particularly whether they
formed from highly evolved silicate liquids or magmatic
hydrothermal fluids, remains problematic despite numer-
ous studies (see London & Manning, 1995). Badham
(1980) pointed out the lack of replacement textures in the
MQT rocks and therefore suggested that they crystallized
from a quartz–tourmaline magma that rose above the
roof of the main mass of granite. Charoy (1982) con-
sidered that MQT rock is undoubtedly of magmatic
 WILLIAMSON et al. FLUID EVOLUTION FROM TOURMALINE COMPOSITIONS
Fig. 1. Map showing the location of Roche Rock and the main granite varieties of the St Austell granite [after Exley & Stone (1982) and
Manning et al. (1996)].
origin and formed by unmixing of a phase rich in B,
alkalis and silica. Smith & Yardley (1996) came to a similar
conclusion from B isotope studies, that the formation of
MQT rocks was due to crystallization of volatiles pro-
duced by the degassing of granite silicate liquids. There
is therefore little argument that MQT rocks formed
from magmatic ‘fluids’ of some kind, rather than from
formation waters. This is supported by the MQT rock
tourmaline being more similar in composition to mag-
matic tourmaline from SW England rather than breccia
and replacement deposit tourmaline in the country rocks
(London & Manning, 1995).
The hydrothermal breccia being studied forms a nar-
row (<2 cm width) vein in a large boulder of MQT rock
at the southwestern margin of Roche Rock. The MQT
rock immediately around the breccia is generally tour-
maline rich (70%) and contains a number of narrow
‘quartz’ veins with <5% acicular tourmaline. From the
elongate nature of the breccia, it is fault related rather
than representing an intrusive hydrothermal body, such
as those at Wheal Remfry, in the western part of the St
Austell granite (Allman-Ward et al., 1982, fig. 1).
PETROGRAPHY
The breccias contain 50–90% angular, fractured tour-
maline grains and abundant cavities, up to 0·5 cm in
1441
diameter, with tourmaline extending into the voids. The
tourmaline shows no embayments or other textures that
may indicate partial resorption, and often has con-
centrically zoned hydrothermal overgrowths. The over-
growths, which are developed parallel to the c-axis and
almost exclusively at one end of each tourmaline grain,
are dark blue–purple (
) to light blue or pale tan ()
(Fig. 2a and f ). The outermost part of the overgrowths
commonly consists of fine (<10 m width) feathery fibres
(Fig. 2). Tourmaline also occurs as fine-grained needles
and sprays of feathery crystallites throughout the quartz
matrix. The angular tourmaline fragments commonly
contain rounded inclusions of quartz and more rarely
inclusions of zircon, apatite and monazite.
Quartz in the breccia is granoblastic and considerably
more coarse grained (2–3 mm) than the tourmaline.
Large, optically continuous crystals commonly enclose a
number of broken tourmaline crystals. In tourmaline-rich
areas, quartz occasionally forms narrow veins (<100 m
width), which cross-cut tourmaline aggregates and com-
monly contain acicular tourmalines. The quartz contains
abundant fluid and mineral inclusions, the latter being
mainly haematite, and more rarely zircon, monazite and
apatite. Many of the fluid inclusions contain halite crystals
and fluids associated with the formation of the breccias
are, therefore, likely to have been relatively saline. Fluid
inclusion studies on Sn–W veins from SW England by
 JOURNAL OF PETROLOGY VOLUME 41 NUMBER 9 SEPTEMBER 2000

Fig. 2. (a) Photomicrograph in plane-polarized light of brecciated tourmaline (light brown cores with pale blue rims) with light blue to black
zoned overgrowths. T, tourmaline; Qz, quartz. The matrix is quartz containing fine tourmaline needles and abundant fluid inclusions. (b–e) X-
ray element maps for Mg, Fe, Sn and F for the same field of view as in (a). The map for Sn is overlaid (in white) with the outline of four discrete
zones in the overgrowths (1–4) and the brecciated MQT tourmaline (0) determined from the maps for Fe, Mg and F. (f ) Photomicrograph in
plane-polarized light for a second area of the breccia showing a brecciated tourmaline grain with secondary hydrothermal overgrowth. (g, h) X-
ray element maps for Fe and Sn for the field of view in (f ) and an overlay for the different zones on the map for Sn.

1442
 WILLIAMSON et al. FLUID EVOLUTION FROM TOURMALINE COMPOSITIONS
Alderton & Harmon (1991) indicated moderate salinities
of between 5 and 15 wt % NaCl equivalent, with ho-
mogenization temperatures of between 230 and 400°C.
More detailed fluid inclusion studies on the breccias
could not be carried out because when the quartz was
viewed in cathodoluminescence, it was found to be made
up of a complex ‘mosaic’ of angular (brecciated) fragments
cemented by several generations of quartz. It would
therefore be extremely difficult to identify fluid inclusions
in quartz cement, formed during the brecciation process,
from those in the breccia tourmaline inherited from the
MQT rocks.
The MQT rocks at Roche are for the most part
granoblastic and homogeneous. They are medium
grained (<1 to 3 mm) and consist almost entirely of
tourmaline (30–70%) and quartz. They show very little
in the way of internal structure apart from the presence of
rare, dark, commonly elongate, tourmaline-rich patches,
‘quartz veins’ (containing <10% tourmaline) and numer-
ous tourmaline- and quartz-lined vugs between <500 m
and >50 cm in diameter. Tourmaline from the MQT
rocks commonly has cores showing relatively strong pleo-
chroism from dark brown (
) to pale tan–yellow () with
rims that are mid-blue (
) to light blue–colourless ().
Some finer-grained tourmaline is entirely pale blue (
)
to colourless ().
ANALYTICAL TECHNIQUES
Quantitative microprobe analysis was carried out using
an automated WDS Cameca SX50 microprobe with an
accelerating voltage of 20 kV, a beam current of 25 nA
and a spot size of >3 m, identical conditions to those
used by London & Manning (1995). The following stand-
ards were used: jadeite for Na, wollastonite for Si and
Ca, olivine for Mg and Fe, pure manganese for Mn,
rutile for Ti, corundum for Al, synthetic KBr for K,
synthetic BaF for F, synthetic cassiterite for Sn. Detection
limits (3 SD) were <0·03 wt % (except for F, <0·1 wt %).
Two traverses, parallel to the c-axis, were carried out
across the contact of MQT fragments and secondary
overgrowths. The positions of the traverses are shown in
Fig. 2. Traverse 1 contains 25 points (length 190 m)
compared with 49 points (length 284 m) in traverse 2,
the latter being considerably more detailed. The data
were recalculated to atoms per formula unit (a.p.f.u.),
based on 24·5 O, which assumes fixed B at 3 a.p.f.u.
and fixed (OH, F) at 4 a.p.f.u. The data have been
plotted on tourmaline ‘discrimination diagrams’ in Fig.
3, and with distance along each traverse in Fig. 4. Data
for individual points from traverse 1, with an average
for the MQT tourmaline, are presented in Table 1.
The recalculation procedure used produced slightly high
values (<6·1) for Si, above the maximum T site occupancy
1443
of 6 a.p.f.u., mainly in zones 2 and 4 (especially for
traverse 2). Extensive checks were carried out to dem-
onstrate that this was not an analytical problem, as is
also evident from its effects being strongest in certain
zones. The high Si values are therefore considered to be
an artefact of the recalculation procedure used to express
the data in formula units, particularly the assumptions
of total Fe as Fe2+, (OH, F) fixed at 4 a.p.f.u. and B
fixed at 3 a.p.f.u., although the latter seems likely from
recent structural refinement studies (Bloodaxe et al.,
1999).
X-ray element maps were generated for small
(>0·5 mm2) areas of the sample surface using a moving
stage rastering program on an automated Cameca
SX50 microprobe. The maps were obtained using an
accelerating voltage of 15 kV, a beam current of 100 nA,
a spot size of >1 m, a spot interval of 2 m and a
count time of 200 ms per spot. Counts per second gain
for each element were measured and plotted on separate
maps using a colour scale, shown in the right-hand
margin, from black to red representing low to high
counts, respectively. X-ray element maps for Mg, Fe, Sn
and F are shown in Fig. 2.
TOURMALINE CHEMISTRY
Tourmaline chemistry is complex with the possibility for
a wide range of substitutions (e.g. Burt, 1989; London &
Manning, 1995; Hawthorne, 1996). A great deal of
research has been undertaken on tourmaline structure
and chemistry, which has recently culminated in a re-
defined chemical classification scheme (Hawthorne &
Henry, 1999). The basic formula for tourmaline is:
XY3Z6(T6O18)(BO3)3V3W with common site occupancies
being:
X = Ca, Na, K or vacant;
Y = Li, Fe2+, Mg, Mn, Al, Cr3+, Fe3+, V3+, Ti4+;
Z = Mg, Al, Fe3+, V3+, Cr3+;
T = Si, Al, B;
B = B, (vacant);
V = OH, O;
W = OH, F, O.
The chemistry of tourmaline from the breccia is, how-
ever, ‘relatively simple’ because of its restricted contents
(in a.p.f.u.) of Ca <0·06, K <0·01, Mn <0·04 and Ti
<0·06 (Table 1). The main end-member compositions
between which the tourmalines in this study may fall
have the general formula XY3Z6(Si6O18)(BO3)3V3W with
[after Hawthorne & Henry (1999)]:
 JOURNAL OF PETROLOGY VOLUME 41 NUMBER 9 SEPTEMBER 2000

Fig. 3. Variation diagrams for points in traverses 1 and 2 (except MQT rocks, which are shown as averages for each traverse) of (a) Mg/(Fe
+ Mg) vs Xvac/(Na + Xvac), (b) Fe vs Mg, (c) Al vs Fe and (d) Fe vs Xvac (in a.p.f.u.). Xvac (X vacancy) = 1 − (Na + K + Ca). Exchange
vector arrows and reference lines important in this study are indicated (Φ, vacancy). Also shown are data from SW England for tourmalines
from granites and pegmatites, MQT rocks, hydrothermal veins, breccias within granites, and replacements and breccias in country rocks [from
London & Manning (1995)].

broad geochemical trends, identified from the X-ray

(X) (Y3) (Z6) V3 W element maps (Fig. 2), and their outlines have been
overlaid on the X-ray element maps for Sn (Fig. 2d and
Elbaite Na+ Li+ 3+
1·5 Al1·3 Al63+ (OH)3 (OH) h). They have also been given different symbols in the
Schorl Na +
Fe 2+
3 Al63+ (OH)3 (OH) variation diagrams in Fig. 3 and are marked on the
Dravite Na+ Mg32+ Al63+ (OH)3 (OH) traverses in Fig. 4.
Olenite Na+ Al33+ Al63+ (O)3 (OH)
In the Xvac vs Mg/(Fe + Mg) discrimination diagram
Buergerite Na+ Fe33+ Al63+ (O)3 (F)
in Fig. 3a, the MQT tourmalines lie just within the
Povondraite Na+ Fe33+ Fe43+Mg22+ (OH)3 (O)
Schorl field, close to the boundary with Foitite. Because
of their high F contents (>0·5 a.p.f.u.), they may be
Rossmanite [] Li+ Al23+ Al63+ (OH)3 (OH)
described as F-Schorl (D. J. Henry, personal com-
Foitite [] Fe22+ Al63+ Al63+ (OH)3 (OH)
munication, 1999). The MQT tourmalines show a limited
Magnesio-foitite [] Mg22+ Al3+ Al63+ (OH)3 (OH)
range in compositions [Fe/(Fe + Mg) = 0·92–0·99],
similar to MQT tourmaline from Roche Rock, which
has a mean value of 0·937 ± 0·043 (3 SD; London &
For ease of description of chemical variations in the Manning, 1995). Their high Fe/(Fe + Mg), F and
tourmalines studied, the overgrowths have been divided Al are more similar to magmatic than hydrothermal
into four zones and the MQT tourmaline has been tourmaline in the Cornubian batholith and are very
described separately. The overgrowths can be clearly different from tourmaline formed in replacement deposits
distinguished from the MQT tourmaline on the X-ray and breccias in the country rock (Fig. 3).
element maps from their much lower F contents and Zone 1 contains considerably higher Mg and lower F
from their strong chemical zonation [Fe/(Fe + Mg) = compared with the MQT tourmaline. From a comparison
0·31–0·99]. The zones were delimited on the basis of of Figs 2, 3 and 4, the inner part of zone 1 (two points

1444
 WILLIAMSON et al. FLUID EVOLUTION FROM TOURMALINE COMPOSITIONS
Fig. 4. (a, b) Microprobe traverses from within the MQT tourmaline
and across the overgrowths. The locations of traverses 1 and 2 are
shown in Fig. 2. The units on the left-hand axis are in a.p.f.u.
in traverse 1 and one point in traverse 2) is particularly
Mg rich, and lies within the Dravite field of Fig. 3a. The
remainder of zone 1 straddles the Mg-Foitite, Foitite and
Schorl fields, showing a poorly defined trend of decreasing
Mg/(Fe + Mg) and Xvac in the direction of the exchange
vectors (Fe2+Fe3+)(MgAl)−1, (NaFe2+)(ΦAl)−1 or
(Fe)(Mg)−1, or more probably a combination of these
vectors (Fig. 3c and d). Zones 2, 3 and 4 fall within the
Schorl field with generally lower Xvac and Mg/(Fe +
Mg). From zone 2 to 3 there is a marked increase in
Xvac and Al and a slight increase in Mg/(Fe + Mg),
indicating the substitution (ΦAl)(NaFe2+)−1. The trend
within zone 3 is parallel to the exchange vector
(ΦAl)(NaMg)−1 (Fig. 3c and d). Zone 4 shows higher Fe
and lower Al compared with zone 3 and similar Al, Fe
and Mg values to zone 2. The main difference between
zone 2 and 4 is the considerably higher Xvac in zone 4.
Because of the similarities in Al, Mg and Fe contents,
this cannot be explained by the substitution
(ΦAl)(NaMg)−1 or (ΦAl)(NaFe2+)−1 and is more likely
to be due to (ΦFe3+)(NaFe2+)−1.
Li contents in the tourmaline have been estimated
using the technique of Burns et al. (1994) (see Table
1). Calculated Li values are relatively low, with Li2O
<0·53 wt % (<1231 ppm) in the hydrothermal over-
growths. This is consistent with the hundred to several
1445
thousands of ppm Li found during ion probe studies of
tourmaline from Cornish hydrothermal breccias (Wilson
& Long, 1983).
Zones 2 and 4 contain high Fe (>3 a.p.f.u.) and low
Mg (<0·5 a.p.f.u.), and lie above the Schorl–Dravite
vector in Fig. 3b. The Y site has a maximum occupancy of
3 a.p.f.u. and the ‘excess’ Fe must therefore be contained
within a different site. It is highly unlikely to substitute
into the X site, in nine-fold co-ordination (Henry &
Dutrow, 1996), as its ionic radius is too small and because
Fe increases with Na towards the maximum Na site
occupancy in X of 1 a.p.f.u. (Table 1). The most likely
site for the ‘excess’ Fe is as Fe3+ in Z, because of the
substitution FeAl−1 (London & Manning, 1995; Dyar
et al., 1998). This is supported by the most Fe-rich
compositions showing proportionately large negative val-
ues for Al in the Y site (Fig. 4), which indicates a ‘deficit’
of Al in Z allowing the substitution of Fe3+ (London &
Manning, 1995). It is also possible that a component of
Fe in Y is contained as Fe3+ towards the composition of
Buergerite.
Tourmaline from the breccia and MQT tourmaline
were also analysed for a variety of ore-forming elements,
Sn, W and Cu. The only element to be found above
detection limits (dl) was Sn (<0·35 wt %; dl = 0·03 wt %),
and this was only in the overgrowths. Both the X-ray
element maps and traverses show significant con-
centrations of Sn in zone 2 and to a lesser extent in
zones 1 and 4. The homogeneous distribution of Sn
throughout zone 2 is likely to indicate that the Sn is
‘structurally bound’ rather than being contained as dis-
crete mineral or amorphous inclusions.
A comparison of the photomicrograph in plane-po-
larized light (Fig. 2a and f ) with the X-ray element maps
shows that there is some correspondence between the
chemical composition of the tourmaline and its colour,
although this is not always the case in tourmaline from
SW England (London & Manning, 1995). The more Fe-
rich zones are dark blue and the more Mg-rich zones
are pale blue. The brecciated MQT tourmalines have
high F and are generally colourless–pale blue with light
brown cores.
DISCUSSION
Constraints on fluid evolution from
tourmaline chemistry
The relative contributions from magmatic waters, defined
as having equilibrated with a silicate melt (Bottrell &
Yardley, 1988), and formation waters in hydrothermal
fluids are notoriously difficult to establish (Manning,
1985). A dominantly magmatic source for early min-
eralizing fluids in SW England is indicated by the close
spatial and temporal association of mineralized veins to
 JOURNAL OF PETROLOGY VOLUME 41 NUMBER 9 SEPTEMBER 2000

Table 1: Tourmaline compositions for each point in traverse 1 (1–18 towards the margin) and an
average for the MQT tourmaline (point 0)

Point no.: Av. MQT 1 2 3 4 5 6

Zone: 0 1 1 1 1 1 1

Na2O 1·64 1·95 1·99 1·42 1·49 1·47 1·52

CaO 0·09 0·05 0·16 0·07 0·05 0·06 0·03
K2O 0·03 0·01 0·00 0·00 0·01 0·00 0·00
MgO 0·28 6·64 6·10 4·71 4·74 4·20 3·55
FeO 14·41 5·34 5·79 6·78 6·96 7·67 9·09
MnO 0·10 0·01 0·00 0·00 0·00 0·02 0·01
TiO2 0·08 0·09 0·17 0·21 0·20 0·19 0·19
Al2O3 35·43 34·73 35·28 36·00 35·87 35·95 35·50
SiO2 35·69 38·66 38·58 37·70 37·89 37·75 37·78
F 1·28 0·34 0·38 0·19 0·18 0·20 0·23
Li2O∗ 0·07 0·52 0·53 0·39 0·42 0·41 0·45
Total 89·09 88·34 88·98 87·47 87·81 87·92 88·35

Na 0·525 0·593 0·603 0·437 0·457 0·452 0·469

Ca 0·015 0·008 0·027 0·012 0·008 0·010 0·005
K 0·007 0·002 0·000 0·000 0·002 0·000 0·000
Mg 0·068 1·553 1·420 1·115 1·119 0·994 0·842
Fe 1·988 0·700 0·756 0·900 0·922 1·018 1·210
Mn 0·014 0·001 0·000 0·000 0·000 0·003 0·001
Ti 0·010 0·011 0·020 0·025 0·024 0·023 0·023
Al 6·887 6·418 6·493 6·737 6·693 6·725 6·659
Si 5·888 6·063 6·025 5·987 6·000 5·993 6·014
F 0·665 0·169 0·188 0·095 0·090 0·100 0·116
Li∗ 0·044 0·327 0·331 0·248 0·266 0·260 0·287
X 0·540 0·601 0·629 0·449 0·466 0·463 0·474
Y 2·069 2·254 2·176 2·015 2·041 2·015 2·054
Z 6·900 6·433 6·519 6·770 6·725 6·755 6·689
X+Y 2·610 2·855 2·806 2·464 2·506 2·477 2·528
Fe/(Fe + Mg) 0·967 0·311 0·347 0·447 0·452 0·506 0·590
Fe + Mg 2·056 2·253 2·176 2·015 2·041 2·012 2·052
XAl 0·788 0·496 0·544 0·757 0·725 0·747 0·703
Xvac 0·452 0·397 0·371 0·551 0·532 0·537 0·526

Sn wt % 0·01 0·04 0·06 0·14 0·19 0·17 0·19

granite plutons ( Jackson, 1979; Bray & Spooner, 1983;
Chen et al., 1993) and from the high concentrations
of transition metal chlorides in fluid inclusions from
tourmaline–topaz–quartz rocks at St Mewan Beacon
(Fig. 1; Bottrell & Yardley, 1988). Fluid inclusion and
isotopic studies suggest that Sn–W deposits at Cligga
Head, Cornwall, were largely derived from magmatic
fluids (Smith et al., 1996). Boron in mineralizing fluids is
also likely to have had a dominantly magmatic source,
as indicated from B isotope studies of tourmalinized veins
at Cligga Head, >30 km west of Roche (Smith & Yardley,
1996). A contribution from formation waters is, however,
difficult to disprove, as they can commonly have B
contents up to 300 ppm (Leeman & Sisson, 1996), or
even higher if associated with evaporites, although these
have not been found in the local country rocks (Smith
& Yardley, 1996).
Evidence for a magmatic component during the initial
formation of zone 1 is the relatively high levels of Al
(Table 1, Fig. 4). Al is particularly soluble in low-pH

1446
 WILLIAMSON et al. FLUID EVOLUTION FROM TOURMALINE COMPOSITIONS

Point no.: 7 8 9 10 11 12 13
Zone: 1 1 1 1 1 1 2

Na2O 1·69 1·85 1·95 1·89 1·82 2·12 2·54

CaO 0·03 0·04 0·03 0·02 0·07 0·12 0·22
K2O 0·01 0·00 0·00 0·01 0·02 0·02 0·03
MgO 3·26 3·55 3·82 2·37 1·93 0·78 0·16
FeO 10·21 10·03 10·05 12·37 13·70 17·04 21·00
MnO 0·02 0·02 0·00 0·03 0·03 0·02 0·03
TiO2 0·22 0·20 0·22 0·23 0·22 0·10 0·02
Al2O3 34·84 34·68 34·27 33·94 33·45 31·19 29·07
SiO2 37·34 37·27 37·57 37·11 36·92 35·95 35·07
F 0·24 0·24 0·19 0·25 0·28 0·38 0·44
Li2O∗ 0·38 0·35 0·42 0·40 0·35 0·31 0·04
Total 88·24 88·23 88·52 88·62 88·79 88·03 88·62

Na 0·527 0·577 0·606 0·594 0·575 0·692 0·849

Ca 0·005 0·007 0·005 0·003 0·012 0·022 0·041
K 0·002 0·000 0·000 0·002 0·004 0·004 0·007
Mg 0·781 0·851 0·912 0·573 0·469 0·196 0·041
Fe 1·372 1·348 1·346 1·677 1·868 2·400 3·026
Mn 0·003 0·003 0·000 0·004 0·004 0·003 0·004
Ti 0·027 0·024 0·027 0·028 0·027 0·013 0·003
Al 6·597 6·570 6·470 6·483 6·429 6·192 5·903
Si 6·001 5·992 6·019 6·016 6·022 6·056 6·043
F 0·122 0·122 0·096 0·128 0·144 0·202 0·240
Li∗ 0·247 0·228 0·271 0·263 0·230 0·209 0·026
X 0·532 0·584 0·611 0·597 0·588 0·714 0·889
Y 2·156 2·202 2·259 2·254 2·342 2·599 3·071
Z 6·633 6·602 6·505 6·521 6·465 6·208 5·907
X+Y 2·687 2·786 2·870 2·851 2·930 3·313 3·961
Fe/(Fe + Mg) 0·637 0·613 0·596 0·745 0·799 0·925 0·987
Fe + Mg 2·153 2·199 2·258 2·250 2·337 2·596 3·067
XAl 0·633 0·595 0·525 0·536 0·486 0·265 −0·050
Xvac 0·466 0·416 0·389 0·400 0·408 0·282 0·104

Sn wt % 0·16 0·12 0·08 0·08 0·07 0·08 0·11

(May et al., 1979), B- (Li + F)-rich aqueous fluids (Morgan
& London, 1989). This type of fluid is likely to be of
magmatic origin in the Roche area as formation waters
will have originated in, or at least passed through, local
country rock calcareous slates. Al is considered to be
largely immobile in ‘typical’ aqueous crustal fluids
[Al(aq) = 5 × 10−5 to 10−4 m, at 400–700°C and
1 kbar; Ragnarsdottir & Walther, 1985]. There is no
source of Al in the MQT rocks as MQT tourmaline, the
only major local Al-bearing phase, shows no embayments
or other resorption textures. Under normal cir-
cumstances, for tourmaline formed within the granites
or country rocks, tourmaline will have derived its Al
from the breakdown of local minerals such as feldspar
and micas.
Further evidence for a magmatic fluid component is
from the relatively reduced nature of the hydrothermal
fluids during the initial formation of zone 1. The change
in composition from the inner to outer part of zone 1 is
along the exchange vector (ΦAl)(NaMg)−1, combined

1447
 JOURNAL OF PETROLOGY VOLUME 41 NUMBER 9 SEPTEMBER 2000

Table 1: continued

Point no.: 14 15 16 17 18
Zone: 2 3 3 3 4

Na2O 2·59 2·29 1·79 1·52 2·01

CaO 0·16 0·18 0·16 0·09 0·16
K2O 0·04 0·03 0·01 0·02 0·02
MgO 0·16 0·96 0·72 0·22 0·35
FeO 21·84 18·76 17·76 17·76 20·39
MnO 0·04 0·03 0·10 0·06 0·13
TiO2 0·02 0·02 0·02 0·01 0·01
Al2O3 28·31 30·71 31·62 32·10 29·31
SiO2 35·37 36·03 36·26 36·25 35·70
F 0·46 0·58 0·41 0·44 0·42
Li2O∗ 0·09 0·09 0·15 0·18 0·10
Total 89·08 89·68 89·00 88·65 88·60

Na 0·865 0·744 0·580 0·493 0·667

Ca 0·030 0·032 0·029 0·016 0·029
K 0·009 0·006 0·002 0·004 0·004
Mg 0·041 0·240 0·179 0·055 0·089
Fe 3·146 2·630 2·481 2·485 2·916
Mn 0·006 0·004 0·014 0·009 0·019
Ti 0·003 0·003 0·003 0·001 0·001
Al 5·746 6·067 6·224 6·328 5·907
Si 6·093 6·041 6·057 6·065 6·106
F 0·251 0·307 0·217 0·233 0·227
Li∗ 0·061 0·058 0·102 0·124 0·068
X 0·894 0·777 0·608 0·509 0·696
Y 3·193 2·874 2·674 2·548 3·024
Z 5·750 6·071 6·227 6·330 5·909
X+Y 4·087 3·651 3·282 3·057 3·720
Fe/(Fe + Mg) 0·987 0·916 0·933 0·978 0·970
Fe + Mg 3·187 2·870 2·660 2·540 3·005
XAl –0·158 0·112 0·284 0·395 0·015
Xvac 0·097 0·217 0·390 0·487 0·300

Sn wt % 0·35 0·10 0·07 0·10 0·11

Values are expressed as wt % oxides and a.p.f.u. (assuming 24·5 O), except Sn, which is in wt %. X = Na + Ca, Y = Fe +
Mg + Mn, Z = Al + 1·33Ti, Al in Y = Al + 1·33Ti + Si − 12 [following the method of London & Manning (1995)].
∗Calculated assuming 24·5 O (total Fe as Fe2+), assigning a value of three minus the Y-site total to Li, adding back the
equivalent wt % Li2O to the oxides, renormalizing again, then iterating this procedure until convergence is reached.

with a more minor component of (Fe)(Mg)−1, with no
evidence for substitutions involving Fe3+. Magmatic wa-
ters will have been relatively reducing as the granites
from which they were derived show low magnetite abund-
ance, low magnetic susceptibility (>1 × 10−5 e.m.u./g;
Willis-Richards & Jackson, 1989) and have been classified
as belonging to the ilmenite series of Ishihara (1977).
This argument is supported by the presence of detectable
concentrations of Sn in the inner part of zone 1
(0·06 wt %, Fig. 4). Sn transport in aqueous fluids requires
low pH and reducing conditions (Taylor & Wall, 1993),
far more likely for magmatic than formation fluids, the
latter having originated in or passed through calcareous
slate country rocks.
Substantial chemical changes in the hydrothermal fluid
are reflected by variations in tourmaline composition

1448
 WILLIAMSON et al. FLUID EVOLUTION FROM TOURMALINE COMPOSITIONS
through zones 1–4. The main trend in zone 1 is parallel
to one or a combination of the exchange vectors
(Fe2+Fe3+)(MgAl)−1, (NaFe)(ΦAl)−1 or (Fe)(Mg)−1 (Fig.
3c and d). The field for zone 2 lies roughly along the
trend of zone 1, but is offset to higher Fe and lower
Xvac. Within zone 2, there is a poorly defined trend,
roughly in the direction of the exchange vector
(NaFe2+)(ΦAl)−1. Fe contents in zone 2 exceed the
maximum that can be accommodated in the tourmaline
Y site (3 a.p.f.u.). From the proportionate decrease in Al
in Z (Fig. 4), it is probable that the ‘excess’ Fe is
accommodated in the Z site as Fe3+, indicating relatively
oxidizing conditions during the formation of zone 2.
Slightly higher Mg contents in zone 3 are probably due
to the substitution (MgAl)(Fe2+Fe3+)−1 whereas the trend
within zone 3 is parallel with the exchange vector
(ΦAl)(NaMg)−1. Zone 4 shows almost identical com-
positions to zone 2, in terms of Fe, Mg and Al, but with
a considerably higher Xvac. This is likely to be due
to the substitution (ΦFe3+)(NaFe2+)−1, in addition to
(Fe2+Fe3+)(MgAl)−1, which caused the main com-
positional shift from zone 3 to 4. This may be indicating
even more oxidizing conditions in zone 4 than in zone
2. Relatively oxidizing conditions are indicated from the
presence of haematite in matrix quartz, which is likely
to have formed contemporaneously with, or soon after,
zone 4.
The more oxidizing conditions could have been caused
by either mixing with more oxidizing fluids or boiling.
The latter was suggested for the formation of tourmalines
under relatively oxidizing conditions in the Larderello
Geothermal Field, Italy (Cavarretta & Puxeddu, 1990).
Boiling causes the loss of H into the vapour phase
and the resulting oxidation of Fe in the aqueous phase
(Drummond & Ohmoto, 1985). However, boiling is
considered unlikely to have caused the apparent increase
in f O2 because F concentrations in the overgrowths
increase in zones 2 and 4 (Table 1). This is the reverse
of what would be expected during boiling, as, above
250°C, HF is preferentially partitioned into the vapour
phase (Reed & Spycher, 1985; Smith, 1995).
The increase in Fe and apparent increase in Fe3+/
Fe2+ through zone 1, which peak in zones 2 and 4,
suggest open-system behaviour during breccia formation,
with an episodic but generally increasing contribution
from relatively oxidizing formation waters. A significant
component of formation waters in the hydrothermal
fluids of SW England has been suggested from numerous
stable isotope determinations and fluid inclusion studies
(e.g. Jackson et al., 1982; Wilkinson, 1990; Alderton &
Harmon, 1991; Wilkinson et al., 1995).
One of the limiting factors for the formation of tour-
maline in the breccia may have been a decrease in Al
in the hydrothermal fluid, as indicated from its general
decline through zones 1–4. From previous discussion
1449
(above) it is likely that formation waters were relatively
depleted in Al and that magmatic fluids were the main
source for this element. Another limiting factor may have
been an increase in pH as a result of progressively higher
degrees of mixing with formation waters from local
calcareous slates. The amount of B required to stabilize
tourmaline increases with increasing pH, with tourmaline
becoming unstable at pH >6·5–7 (Morgan & London,
1989). Other ‘ingredients’ for tourmaline formation,
mainly Fe, Na, OH and Si, were abundant in the
hydrothermal fluids at this time, as indicated from the
presence of inclusions of haematite in matrix quartz and
high-salinity fluid inclusions. High hydrothermal fluid Fe
contents are evident in the St Austell area from the
occurrence of bleached, biotite-depleted aureoles around
hydrothermal quartz–tourmaline veins in biotite granites
(London & Manning, 1995). Similar bleached zones are
also commonly seen in country rock metasediments.
Open-system formation of hydrothermal tourmaline,
involving mixing between magmatic (providing B, Al,
Mg and Fe) and formation waters (probably providing
further Fe), is in agreement with the findings of London
& Manning (1995) from geochemical studies of many
different types of tourmaline from across SW England.
Ore-carrying potential of the hydrothermal
fluids
In a comprehensive microprobe study of tourmaline
compositions from many different occurrences in Corn-
wall, none were found to contain Sn (London & Manning,
1995). Why Sn was found in the overgrowths is not clear.
It may simply be due to the refractory nature of the
MQT rocks in which the breccia formed, with there
having been little cassiterite crystallization because of
limited wall rock reaction. Alternatively, Sn con-
centrations in the hydrothermal fluid may have been
below cassiterite saturation.
Sn concentrations show marked variations through the
overgrowths. In zone 1, Sn shows oscillatory zoning from
>0·09 to 0·18 wt %, at a scale of >20 m, independent
of any other element (Fig. 4, traverse 2). Zone 2 shows
a peak for Sn of 0·35 wt %, which correlates with the
highest peak for Fe. Zones 3 and 4 show a decrease in
Sn to similar levels as in zone 1. The origin of the
oscillatory zoning in zone 1 is uncertain, as this process
can be caused by a number of mechanisms, which
are often difficult to identify (e.g. Yardley et al., 1991;
Lanzirotti, 1995), particularly in minerals formed in open
systems [reviewed by Shore & Fowler (1996)]. These
mechanisms are either extrinsic, caused by external
changes such as episodic fluid flow, variations in pressure,
temperature, pH or f O2, or intrinsic, being caused by
‘feed-back between crystal growth and solute diffusion
 JOURNAL OF PETROLOGY VOLUME 41
or surface effects as a source of non-linearity in the
crystal-growth kinetics’ (Shore & Fowler, 1996). The
effects of extrinsic mechanisms can be recognized between
zones in the overgrowths as it is likely that each was
formed from different mixtures of fluids. The oscillatory
zoning for Sn in zone 1 is better explained by intrinsic
mechanisms, as this is the only element to show such
regular oscillatory trends (Fig. 4).
How Sn was incorporated within the tourmaline struc-
ture in zone 1 is problematic. From previous discussion,
zone 1 (particularly the inner part) is likely to have been
formed under relatively reducing conditions. The Sn was
therefore probably contained in the hydrothermal fluid
as Sn2+. Sn is dominantly thought to be transported as
Sn2+ in low redox hydrothermal fluids responsible for
the formation of Sn deposits (commonly in the form of
chloride and hydroxy chloride complexes; Jackson &
Helgeson, 1985; Wilson & Eugster, 1990; Taylor & Wall,
1993). Sn2+ has an extremely large ionic radius of 0·122
nm in eight co-ordination (Shannon & Prewitt, 1969),
0·093 nm in six co-ordination (see Taylor, 1979). It is
therefore highly unlikely to substitute into an octahedral
tourmaline site for Fe2+ (0·078 nm), Mg2+ (0·072 nm) or
Al3+ (0·0535 nm) (six co-ordination; Shannon & Prewitt,
1969). The most likely substitution is within the X site,
which is a nine co-ordinated trigonal antiprism located
along the three-fold axis of symmetry (Henry & Dutrow,
1996). This site is usually occupied by Na (0·102 nm),
Ca (0·10 nm), with minor K (0·138 nm), or can be
vacant.
The increasing concentrations of Sn in zone 2 can be
explained either by an increase in Sn2+ concentrations
in the hydrothermal fluid or by more oxidizing conditions,
with Sn4+ being more easily accommodated within the
tourmaline structure than Sn2+. The first possibility is
unlikely, as Sn in zone 2 is much higher than normally
found in tourmaline from SW England or mineral de-
posits world-wide. Southwest England tourmaline was
found to have maximum Sn contents of <500 ppm from
emission spectrography studies on 48 samples by Power
(1968). Proton microprobe studies of 32 samples of tour-
malines from massive sulphide deposits and tourmalinites
world-wide indicated Sn concentrations of <745 ppm
(Griffin et al., 1996). Nemec (1973) found a maximum of
500 ppm in 80 tourmaline samples from different settings,
Sn-mineralized and non-mineralized, world-wide. This
suggests maximum Sn contents in tourmalines much less
than 1000 ppm, which is likely to reflect the relatively
reducing conditions that usually prevail during tour-
maline crystallization and the dominance, and in-
compatible nature, of Sn2+ in hydrothermal fluids,
especially where associated with Sn deposits.
The increase in Sn concentration in zone 2 to an-
omalously high levels is far more likely to be due to Sn4+,
with an ionic radius of 0·069 nm [six co-ordination; see
1450
NUMBER 9 SEPTEMBER 2000
Taylor (1979)], being more easily accommodated within
the tourmaline structure than Sn2+. This provides further
evidence in support of more oxidizing conditions during
the formation of zone 2. Which site it occupies is unclear,
as again there are a variety of possible coupled sub-
stitutions that could occur. Assuming that the high Sn4+
content of the tourmaline reflects high concentrations in
the hydrothermal fluid, its presence in the fluid is likely
to be a transient phenomenon, as it would usually be
expected to precipitate out, forming cassiterite. It is
possible that Sn partitioning within tourmaline was kin-
etically favoured over cassiterite nucleation. Alternatively,
Sn concentrations may have been below cassiterite sat-
uration, although this seems unlikely because of the
rare occurrence of such high tourmaline Sn contents.
Unfortunately, as the amount of Sn in the tourmaline
probably relates to its valence state during tourmaline
crystallization rather than to its concentration, no firm
conclusions can be reached on its magmatic or formation
water source. An interesting observation, however, is that
the Sn content of zone 4 is very much lower than that
of zone 2, whereas (with the exception of Xvac) these
zones are almost identical in all other aspects of their
geochemical compositions. Both zones were apparently
formed under relatively oxidizing conditions with the
hydrothermal fluid being dominated by formation waters.
The most likely explanation for this is that the fluids
were depleted in Sn by the time zone 4 was formed.
Whether this was because of a decrease in cassiterite
solubility in the magmatic fluid over time or was due to
the diluting effect of mixing with formation waters is
unclear.
Breccia formation
The X-ray element maps in Fig. 2 give an indication of
the timing of tourmaline nucleation and growth. Zones
1 and 3, which show low Fe/Mg, are present only as
overgrowths on, or fracture fills within, brecciated MQT
tourmaline. Nowhere can they be seen to have formed
separate nucleation sites. Tourmaline nucleation appears
to have occurred only during the formation of zones 2
and 4, with evidence for the latter being the common
occurrence of feathery filaments of high-Fe tourmaline
in the quartz matrix. The lack of nucleation during
the formation of zones 1 and 3 was probably due to
crystallization on pre-existing tourmaline being kinetically
favoured. Nucleation during the formation of zones 2
and 4 coincides with the proposed peak in mixing with
formation waters and may therefore have been due to
this process having led to supersaturation of the fluid
with respect to tourmaline.
 WILLIAMSON et al. FLUID EVOLUTION FROM TOURMALINE COMPOSITIONS
CONCLUSIONS AND WIDER
IMPLICATIONS OF THE STUDY
Compositional variations across secondary tourmaline
overgrowths in breccias at Roche have provided detailed
information on the evolution of mineralizing fluids. Four
distinct zones have been recognized in the overgrowths,
zones 1–4 towards the margins. The overgrowths are
interpreted as having initially crystallized from relatively
Mg-, Al-rich hydrothermal fluids of magmatic origin,
which progressively mixed with more oxidized, Fe-rich
formation waters, dominant during the formation of
zones 2 and 4. All four zones contain detectable con-
centrations of Sn, which reach a peak in zone 2. This
peak is considered to be due to more oxidizing conditions
during the formation of zone 2, with Sn being more
easily accommodated within the tourmaline structure as
Sn4+ than as Sn2+. The presence of Sn4+ in hydrothermal
fluids is likely to be a transient phenomenon, as it would
normally be expected to be precipitated out as cassiterite.
The identification of chemically distinct zones in the
overgrowths has also yielded valuable information on the
timing of tourmaline nucleation and growth. During the
formation of zones 1 and 3, tourmaline crystallization
was restricted to overgrowths on pre-existing brecciated
MQT tourmaline. During the formation of zones 2 and
4 there was also widespread tourmaline nucleation away
from the MQT tourmaline. These nucleation events are
likely to relate to periods of supersaturation with respect
to tourmaline, probably as a result of a sudden increase
in mixing with formation waters.
The study of just a few grains of tourmaline has
revealed details of the evolution of mineralizing fluids in
the area. It is hoped that this will encourage more
extensive studies of chemical variations within single
tourmalines as a tool in mineral exploration and in
petrogenetic studies. Tourmalinization is commonly as-
sociated with mineralization in many of the world’s most
economically important mineral deposit types, par-
ticularly in breccias related to Cu porphyry deposits. In
the Los Bronces Cu porphyry, for example, which is one
of the largest known of its type in the world, mineralization
is concentrated within a tourmaline breccia, with tour-
malinization having occurred before, during and after
the main episode of mineralization (Warnaars et al., 1985).
Many of these types of deposit are highly altered from
the ingress of later fluids, which commonly produce
multiple generations of fluid inclusions and complex
mineral assemblages, which hamper the study of fluid
chemistry related to primary mineralization. Tourmaline
is commonly unaffected by later alteration and so can
preserve original information on the conditions of min-
eralization. Its chemistry also often directly reflects that
of related ores (Griffin et al., 1996). The study of individual
tourmaline grains will complement the recently more
1451
popular approach of comparing the compositions of
single or a few tourmalines from many deposit types to
identify genetic criteria to be used in exploration and
genetic studies (e.g. Clarke et al., 1989; London & Man-
ning, 1995; Griffin et al., 1996; reviewed by Slack, 1996).
The type of study presented here is likely to become
increasingly worth while with the development and rou-
tine use of more sophisticated methods of ‘point analysis’,
with progressively lower detection limits, such as laser
ablation ICP-MS and ion probe, particularly for B and
ore-forming elements. Methods for studying the valency
of Sn and Fe in situ (and non-destructively) in polished
sections of mineral samples are also eagerly awaited.
ACKNOWLEDGEMENTS
All analyses were carried out at the Natural History
Museum, London, through internal funding. The authors
wish to thank Terry Williams and Tony Wighton for
technical help, and Martin Smith, Darrell Henry, David
Manning and an anonymous referee for comments on
the manuscript.
REFERENCES
Alderton, D. H. M. & Harmon, R. S. (1991). Fluid inclusion and stable
isotope evidence for the origin of mineralizing fluids in south-west
England. Mineralogical Magazine 55, 605–611.
Alderton, D. H. M. & Rankin, A. H. (1983). The character and
evolution of hydrothermal fluids associated with the kaolinised St
Austell granite, SW England. Journal of the Geological Society, London
140, 297–310.
Allman-Ward, P., Halls, C., Rankin, A. & Bristow, C. M. (1982). An
intrusive hydrothermal breccia body at Wheal Remfry in the western
part of the St Austell granite pluton, Cornwall, England. In: Evans,
A. M. (ed.) Metallization Associated with Acid Magmatism. Chichester:
John Wiley, pp. 1–28.
Badham, J. P. N. (1980). Late magmatic phenomena in the Cornish
batholith—useful field guides for tin mineralisation. Proceedings of the
Ussher Society 5, 44–53.
Bloodaxe, E. S., Hughes, J. M., Dyar, M. D., Grew, E. S. & Guidotti,
C.V. (1999). Linking structure and chemistry in the Schorl–Dravite
series. American Mineralogist 84, 922–928.
Bottrell, S. H. & Yardley, B. W. D. (1988). The composition of a
primary granite derived ore fluid from S.W. England, determined
by fluid inclusion analysis. Geochimica et Cosmochimica Acta 52, 585–588.
Bray, C. J. & Spooner, E. T. C. (1983). Sheeted vein Sn–W min-
eralization and greisenization associated with economic ka-
olinization, Goonbarrow China Clay Pit, St Austell, Cornwall,
England: geologic relationships and geochronology. Economic Geology
78, 1064–1089.
Burns, P. C., MacDonald, D. J. & Hawthorne, F. C. (1994). The crystal
chemistry of manganese-bearing Elbaite. Canadian Mineralogist 32,
31–41.
Burt, D. M. (1989). Vector representation of tourmaline compositions.
American Mineralogist 74, 826–839.
Cavarretta, G. & Puxeddu, M. (1990). Schorl–dravite–ferridravite
tourmalines deposited by hydrothermal magmatic fluids during early
 JOURNAL OF PETROLOGY
evolution of the Larderello Geothermal Field, Italy. Economic Geology
85, 1236–1251.
Charoy, B. (1982). Tourmalinisation in Cornwall, England. In: Evans,
A. M. (ed.) Metallization Associated with Acid Magmatism. Chichester:
John Wiley, pp. 63–70.
Chen, Y., Clark, A. H., Farrar, E., Wasteneys, H. A. H. P., Hodgson,
M. J. & Bromley, A. V. (1993). Diachronous and independent
histories of plutonism and mineralization in the Cornubian batholith,
southwest England. Journal of the Geological Society, London 150, 1183–
1191.
Chesley, J. T., Halliday, A. N., Snee, L. W., Mezger, K., Shepherd,
T. J. & Scrivener, R. C. (1993). Thermochronology of the Cornubian
batholith in south-west England: implications for pluton em-
placement and protracted hydrothermal mineralization. Geochimica
et Cosmochimica Acta 57, 1817–1835.
Clarke, D. B., Reardon, N. C., Chatterjee, A. K. & Gregoire, D. C.
(1989). Tourmaline composition as a guide to mineral exploration:
a reconnaissance study from Nova Scotia using discriminant function
analysis. Economic Geology 84, 1921–1935.
Collins, J. H. (1878). The Hensbarrow Granite District: a Geological Description
and a Trade History. Truro: Lake and Lake.
Darbyshire, D. P. F. & Shepherd, T. J. (1985). Chronology of granite
magmatism and associated mineralisation, SW England. Journal of
the Geological Society, London 142, 1159–1177.
De La Beche, H. T. (1839). Report on the Geology of Cornwall, Devon
and West Somerset. London: Longman, Orme, Brown, Green and
Longmans.
Drummond, S. E. & Ohmoto, H. (1985). Chemical evolution and
mineral deposition in boiling hydrothermal systems. Economic Geology
80, 126–147.
Dyar, M. D., Taylor, M. E., Lutz, T. M., Francis, C. A., Guidotti,
C. V. & Wise, M. (1998). Inclusive chemical characterization of
tourmaline: Mössbauer study of Fe valence and site occupancy.
American Mineralogist 83, 848–864.
Exley, C. S. & Stone, M. (1982). Hercynian intrusive rocks. In:
Sutherland, D. S. (ed.) The Variscan Fold Belt in the British Isles.
Chichester: John Wiley, pp. 287–311.
Farmer, C. B., Searl, A. & Halls, C. (1991). Cathodoluminescence and
growth of cassiterite in the composite lodes at South Crofty Mine,
Cornwall, England. Mineralogical Magazine 55, 447–458.
Griffin, W. L., Slack, J. F., Ramsden, A. R., Tin Win, T. & Ryan, C.
G. (1996). Trace elements in tourmalines from massive sulfide
deposits and tourmalinites: geochemical controls and exploration
applications. Economic Geology 91, 657–675.
Hawthorne, F. C. (1996). Structural mechanisms for light-element
variations in tourmaline. Canadian Mineralogist 34, 123–132.
Hawthorne, F. C. & Henry, D. J. (1999). Classification of the minerals
of the tourmaline group. European Journal of Mineralogy 11, 201–215.
Henry, D. J. & Dutrow, B. L. (1996). Metamorphic tourmaline and
its petrologic applications. Mineralogical Society of America, Reviews in
Mineralogy 33, 503–557.
Ishihara, S. (1977). The magnetite-series and ilmenite-series granitic
rocks. Mining Geology 27, 293–305.
Jackson, K. J. & Helgeson, H. C. (1985). Chemical and thermodynamic
constraints on the hydrothermal transport and deposition of tin: I.
Calculation of the solubility of cassiterite at high pressures and
temperatures. Geochimica et Cosmochimica Acta 49, 1–22.
Jackson, N. J. (1979). Geology of the Cornubian tin field: ‘a review’.
Bulletin of the Geological Society of Malaysia 11, 209–237.
Jackson, N. J., Halliday, A. N., Sheppard, S. M. F. & Mitchell, J. G.
(1982). Hydrothermal activity in the St Just mining district, Cornwall.
In: Evans, A. M. (ed.) Metallisation Associated with Acid Magmatism.
Chichester: John Wiley, pp. 137–179.
VOLUME 41 NUMBER 9 SEPTEMBER 2000
Jiang, S.-Y., Palmer, M. R., Slack, J. F. & Shaw, D. R. (1998).
Paragenesis and chemistry of multistage tourmaline formation in
the Sullivan Pb–Zn–Ag Deposit, British Columbia. Economic Geology
93, 47–67.
Lanzirotti, A. (1995). Yttrium zoning in metamorphic garnets. Geochimica
et Cosmochimica Acta 59, 4105–4110.
Leeman, W. P. & Sisson, V. B. (1996). Geochemistry of boron and its
implications for crustal and mantle processes. Mineralogical Society of
America, Reviews in Mineralogy 33, 645–695.
London, D. & Manning, D. A. C. (1995). Chemical variation and
significance of tourmaline from Southwest England. Economic Geology
90, 495–519.
Manning, D. A. C. (1983). Disseminated tin sulphides in the St Austell
granite. Proceedings of the Ussher Society 5, 411–416.
Manning, D. A. C. (1985). Comparison of influence of magmatic
water on form of granite-hosted Sn–W deposits and associated
tourmalinisation from Thailand and southwest England. In: Halls,
C. (ed.) High Heat Production Granites, Hydrothermal Circulation and Ore
Genesis. London: Institute of Mining and Metallurgy, pp. 203–212.
Manning, D. A. C. (1991). Chemical variation in tourmalines from
south-west England. Proceedings of the Ussher Society 7, 327–332.
Manning, D. A. C., Hill, P. I. & Howe, J. H. (1996). Primary lithological
variation in the kaolinized St Austell Granite, Cornwall, England.
Journal of the Geological Society, London 153, 827–838.
May, H. M., Helmke, P. A. & Jackson, M. L. (1979). Gibbsite solubility
and thermodynamic properties of hydroxyaluminum ions in aqueous
solution at 25°C. Geochimica et Cosmochimica Acta 43, 861–868.
Morgan, G. B. & London, D. (1989). Experimental reactions of am-
phibolite with boron-bearing aqueous fluids at 200 MPa: implication
for tourmaline stability and partial melting in mafic rocks. Contributions
to Mineralogy and Petrology 102, 218–297.
Nemec, D. (1973). Tin in tourmalines. Neues Jahrbuch für Mineralogie,
Monatshefte 2, 58–63.
Pichavant, M. & Manning, D. A. C. (1984). Petrogenesis of tourmaline
and topaz granites; the contribution of experimental data. Physics of
the Earth and Planetary Interiors 35, 31–50.
Pollard, P. J., Pichavant, M. & Charoy, B. (1987). Contrasting evolution
of fluorine- and boron-rich tin systems. Mineralium Deposita 22,
315–321.
Power, G. M. (1968). Chemical variation in tourmalines from south-
west England. Mineralogical Magazine 36, 1078–1089.
Ragnarsdottir, K. V. & Walther, J. V. (1985). Experimental de-
termination of corundum solubilities in pure water between
400–700°C and 1–3 kbar. Geochimica et Cosmochimica Acta 49, 2109–
2115.
Reed, M. H. & Spycher, N. (1985). Boiling, cooling, and oxidation in
epithermal systems: a numerical modelling approach. In: Berger, B.
R. & Bethke, P. M. (eds) Geology and Chemistry of Epithermal Systems.
Littleton, CO: Society of Economic Geologists, pp. 249–272.
Shannon, R. D. & Prewitt, C. T. (1969). Effective ionic radii in oxides
and fluorides. Acta Crystallographica B25, 925–946.
Shepherd, T. J., Miller, M. F., Scrivener, R. C. & Darbyshire, D. P.
F. (1985). Hydrothermal fluid evolution in the relation to min-
eralization in southwest England with special reference to the Dart-
moor–Bodmin area. In: Halls, C. (ed.) High Heat Production Granites,
Hydrothermal Circulation and Ore Genesis. London: Institute of Mining
and Metallurgy, pp. 345–364.
Shore, M. & Fowler, A. D. (1996). Oscillatory zoning in minerals: a
common phenomenon. Canadian Mineralogist 34, 1111–1126.
Slack, J. F. (1982). Tourmaline in Appalachian–Caledonian massive
sulphide deposits and its exploration significance. Transactions of the
Institute of Mining and Metallurgy 91, B81–B89.
1452
 WILLIAMSON et al. FLUID EVOLUTION FROM TOURMALINE COMPOSITIONS
Slack, J. F. (1996). Tourmaline associations with hydrothermal ore
deposits. Mineralogical Society of America, Reviews in Mineralogy 33,
559–643.
Slack, J. F., Palmer, M. R., Stevens, B. P. J. & Barnes, R. G. (1993).
Origin and significance of tourmaline-rich rocks in the Broken Hill
district, Australia. Economic Geology 88, 505–541.
Smith, M., Banks, D. A., Yardley, B. W. D. & Boyce, A. (1996). Fluid
inclusion and stable isotope constraints on the genesis of the Cligga
Head Sn–W deposit, S. W. England. European Journal of Mineralogy
8, 961–974.
Smith, M. P. (1995). The geochemistry of water–rock interaction in
progressively focused hydrothermal flow. Ph.D. Thesis, University
of Leeds.
Smith, M. P. & Yardley, B. W. D. (1996). The boron isotopic com-
position of tourmaline as a guide to fluid processes in the southwestern
England orefield: an ion microprobe study. Geochimica et Cosmochimica
Acta 60, 1415–1427.
Taylor, J. R. & Wall, V. J. (1993). Cassiterite solubility, tin speciation
and transport in a magmatic aqueous phase. Economic Geology 88,
437–460.
Taylor, R. G. (1979). Geology of Tin Deposits. New York: Elsevier.
Warnaars, F. W., Holmgren, D. C. & Barassi, F. S. (1985). Porphyry
copper and tourmaline breccias at Los Bronces–Rio Blanco, Chile.
Economic Geology 80, 1544–1565.
Weiss, S. (1989). The minerals of the china clay pits in Cornwall. Lapis
14, 11–24.
1453
Wilkinson, J. J. (1990). The role of metamorphic fluids in the evolution
of the Cornubian orefield: fluid inclusion evidence from south
Cornwall. Mineralogical Magazine 54, 219–230.
Wilkinson, J. J., Rankin, A. H., Mulshaw, S. C., Nolan, J. & Ramsey,
M. H. (1994). Laser ablation–ICP–AES for the determination of
metals in fluid inclusions: an application to the study of magmatic
ore fluids. Geochimica et Cosmochimica Acta 58, 1133–1146.
Wilkinson, J. J., Jenkin, G. R. T., Fallick, A. E. & Foster, R. P. (1995).
Oxygen and hydrogen isotopic evolution of Variscan crustal fluids,
south Cornwall, U.K. Chemical Geology 123, 239–254.
Williamson, B. J., Stanley, C. J. & Wilkinson, J. J. (1997). Implications
from inclusions in topaz for greisenisation and mineralisation in
the Hensbarrow topaz granite, Cornwall, England. Contributions to
Mineralogy and Petrology 127, 119–128.
Willis-Richards, J. & Jackson, N. J. (1989). Evolution of the Cornubian
ore field, southwest England: Part I. Batholith modelling and ore
distribution. Economic Geology 84, 1078–1100.
Wilson, G. A. & Eugster, H. P. (1990). Cassiterite solubility and tin
speciation in supercritical chloride solutions. Geochemical Society Special
Publication 2, 179–195.
Wilson, G. C. & Long, J. V. P. (1983). The distribution of lithium in
some Cornish minerals: ion microprobe measurements. Mineralogical
Magazine 47, 191–199.
Yardley, B. W. D., Rochelle, C. A., Barnicoat, A. C. & Lloyd, G. E.
(1991). Oscillatory zoning in metamorphic minerals: an indicator of
infiltration metasomatism. Mineralogical Magazine 55, 357–365.