Composition of fluids during serpentinite breakdown in subduction

zones: Evidence for limited boron mobility

Eric Tenthorey
Jörg Hermann
Research School of Earth Sciences, Australian National University, Mills Road, Canberra ACT 0200, Australia

ABSTRACT
We experimentally investigated the trace element compositions of fluids released dur-
ing breakdown of subducted serpentinites. Serpentinites contain significant amounts of
fluid-mobile elements such as B, Cs, As, and Ba, which during seafloor alteration are
incorporated into mantle rocks. During the later high-pressure breakdown of the serpen-
tinites, these trace elements are redistributed among the residual olivine, orthopyroxene,
minor chlorite, and fluid. We find that B is far more compatible in these minerals than
previously assumed; it has a fluid/residue partition coefficient (F/RD) of 3–5. Most other
fluid-mobile elements (Cs, As, Ba, Pb) are strongly enriched in the fluid and exhibit ex-
pected F/RD values of 30–250. Serpentinites are possibly the most important sink for B in
subduction zones; the experimental results suggest that significant B is recycled into the
deep mantle. Furthermore, high B concentrations in mantle olivines might be a fingerprint
for previous metasomatism or serpentinization.

Keywords: boron, subduction, serpentinite, olivine.

INTRODUCTION tracer for oceanic crust involvement in mantle plume–related volcanic

Subduction of serpentinized oceanic lithosphere provides an im-
portant mechanism by which H2O is introduced to the deep mantle
(Ulmer and Trommsdorff, 1995). Although it has been suggested that
release of H2O during serpentinite dehydration may govern certain
physical processes in subduction zones (Dobson et al., 2002; Peacock,
2001), the geochemical signature of this fluid and the olivine-rich res-
idue have largely been ignored in terms of their importance to arc-
magma genesis and the recycling of oceanic crust, respectively. Only
recently has it been suggested that the fluids released during serpentin-
ite dehydration might impart a unique geochemical imprint on subduction-
zone magmas (Hattori and Guillot, 2003; Rüpke et al., 2002; Scam-
belluri et al., 2001b, 2004).
The serpentinization process is poorly understood, but is thought
to occur when seawater-derived fluids react with mantle rocks in the
oceanic crust. Large volumes of serpentinized mantle have been ob-
served at passive continental margins (Boillot et al., 1989) and at slow-
spreading ridges (Cannat, 1993). Additionally, serpentinization occurs
when fluids permeate down large oceanic transforms or normal faults
at the outer rise of subduction zones, transforming olivine and ortho-
pyroxene to the hydrated mineral antigorite (Ranero et al., 2003). Al-
though serpentinites are depleted in most trace elements, previous stud-
ies indicate that they are often enriched in certain fluid-mobile elements
such as B, Cl, Sr, and Ba (Bonatti et al., 1984; O’Hanley, 1996; Scam-
belluri et al., 2001a) and may also incorporate other trace elements
such as Cs, As, Be, and Li (Hattori and Guillot, 2003; Scambelluri et
al., 2004).
It is widely accepted that B behaves incompatibly during subduc-
tion and is strongly partitioned into the fluid phase during dewatering
of sediments and dehydration of altered basaltic crust (Ishikawa et al.,
2001; Morris et al., 1990). However, high B concentrations have been
documented in olivine that formed during high-pressure breakdown of
antigorite (Scambelluri et al., 2004), suggesting that partitioning of B
into fluids might be reduced if olivine is involved in the reaction.
Potential B retention by olivine is significant in that it may provide a
mechanism by which B is recycled deep into the mantle, providing a
rocks. Furthermore, if olivine is a potential B sink, then the B concen-
tration in the primitive mantle may be higher than previously thought,
because such estimates are based on current assumptions relating to
the mobility of B (McDonough and Sun, 1995). In this paper we pre-
sent results from a novel suite of experiments, in which we explore
the partitioning behavior of B and other trace elements during serpen-
tinite breakdown. The results from this study are new and unexpected
and suggest that B might behave differently than previously thought.
EXPERIMENTAL METHODOLOGY
To evaluate the importance of trace elements hosted in subduction-
related serpentinites, we have estimated the main reservoirs for a series
of trace elements in partly serpentinized subducting lithosphere (Fig.
1). Although the proportion of serpentinized oceanic lithospere is poor-
ly constrained, recent seismic evidence suggests as much as 17% ser-
pentinization over 20 km (Ranero and Salarès, 2004). In Figure 1 we
assume a value nearly half this large, assuming 2 km of pure serpen-
tinite. Even using this value, it appears that serpentinites are the main
host of B and can contain significant amounts of As and Cs, but are
not important for other fluid-mobile elements such as Li, Pb, Ba, and
Rb. Because the B budget in subduction zones is likely to be controlled
mainly by serpentinites, this work focuses mainly on the partitioning
behavior of B during serpentinite dehydration.
To determine fluid/residue partitioning behavior during dehydra-
tion of a natural serpentinite, we have performed piston-cylinder ex-
periments in which the fluid was captured in diamond traps and the
mobility of B and a suite of other elements was determined by laser-
ablation inductively coupled plasma–mass spectrometry. Our experi-
ments are similar to those of other studies that have used this method
on other rock types (Johnson and Plank, 1999; Stalder et al., 1998),
but we have made certain modifications to overcome important limi-
tations. Unlike previous serpentinite dehydration experiments that were
conducted through the use of perforated capsules (Tatsumi et al., 1986),
the fluids generated by the dehydration reactions remain in the system
and are allowed to equilibrate with the residual assemblage. Further-

q 2004 Geological Society of America. For permission to copy, contact Copyright Permissions, GSA, or editing@geosociety.org.
Geology; October 2004; v. 32; no. 10; p. 865–868; doi: 10.1130/G20610.1; 5 figures; 1 table; Data Repository item 2004143. 865

Figure 1. Mass balance of trace elements between different litholo-
gies in subducting oceanic lithosphere. Of B in subducting slabs,
~60% is contained within serpentinites, in contrast to most other
fluid-mobile elements that are transported mainly by sedimentary
rocks or altered oceanic crust (OC). Owing to predominance of B in
serpentinite, partitioning of B during dehydration will exert major
control on how much B is incorporated into island-arc magma
sources and how much is recycled to deep mantle. Serpentinites
may also be moderate sinks for other fluid-mobile elements such as
As and Cs. Lithologic thicknesses used: sedimentary rocks (400 m),
altered oceanic crust (1000 m), fresh oceanic crust (1000 m), gabbro
(3000 m), depleted mantle (6000 m), serpentinite (2000 m). Elemental
concentrations used are presented in Appendix DR1 and Table DR1
(see footnote 1 in text).
more, for the first time, trace element analysis of the residual phases
is complemented by accurate analysis of the dehydration fluid as mea-
sured from the quench precipitates in the diamond trap (see Appendix
DR11).
RESULTS
Three experiments were conducted at 30 kbar and 750 8C, a
pressure-temperature regime beyond the stability limit of serpentine
and typical for a subduction-zone setting. Following reaction for 7
days, specimens were completely transformed to 90 wt% of a residual
assemblage dominated by olivine, orthopyroxene, and minor chlorite
(Fig. 2) and 10 wt% fluid, from which quench material precipitated in
the diamond trap. Chlorite occurs mainly as domains within larger
olivine zones, whereas orthopyroxene is present mainly as fine laths.
1GSA Data Repository item 2004143, Appendix DR1, methods, and Table
DR1, oceanic lithosphere mass balance, is available online at www.geosociety.org/
pubs/ft2004.htm, or on request from editing@geosociety.org or Documents Sec-
retary, GSA, P.O. Box 9140, Boulder, CO 80301-9140, USA.
Figure 2. Scanning electron microscope backscatter micrograph
showing microtextures in dehydrated serpentinite specimen run at
750 8C and 30 kbar for 1 week. Ol is olivine, Opx is orthopyroxene,
and Chl is chlorite.
866
Figure 3. Profile of raw Cs concentration and porosity through
olivine-rich residue following dehydration of serpentinite. This
experiment was initially doped with 100 ppm Cs (assuming 10
wt% H2O from serpentinite dehydration). That Cs and porosity
closely track each other indicates that Cs is partitioned
strongly into fluid phase and is negligible in residual phase.
This allows Cs to be used as internal standard in determina-
tion of fluid’s trace element composition (Appendix DR1; see
footnote 1 in text).
The porosity in the residue decreases from 12 to 8 vol% in the first 4
mm away from the diamond trap and then quickly drops below 5%
(Fig. 3). The presence of porosity indicates that the fluid released was
able to equilibrate with the residue. For the determination of the residue
composition, we used data obtained from this well-equilibrated higher-
porosity zone and corrected for trapped fluid. Owing to laser spot-size
restrictions, it was impossible to acquire trace element analyses of in-
dividual minerals; therefore, the partition coefficients that we report are
for the bulk olivine-rich residue.
Most elements in the starting serpentinite were present in concen-
trations lower than that of primitive mantle, except for B, As, Cs, Pb,
and Sr (Fig. 4), which are variably enriched. During dehydration of the
serpentinite, most of the fluid-mobile species are partitioned strongly
into the fluid phase. The fluid exhibits elemental enrichments of 1–2.5
orders of magnitude relative to the olivine residue, which corresponds
to fluid/residue partition coefficients (F/RD) of 10–500. The fluid pro-
duced is water rich and contains dissolved SiO2 (;6 wt%), MgO (;2
wt%), CaO (;1 wt%), and Al2O3 (;1 wt%). Most of the incompatible
elements are enriched in the fluid by a factor of 5–10 with respect to
Figure 4. Trace element concentrations (normalized to primi-
tive mantle values) of starting serpentinite (squares), dehy-
drated olivine-rich residual phase (dark gray shading), and de-
hydration fluid (light gray shading) as measured from quench
precipitate in diamond traps. In starting material only B, As,
and Cs are significantly enriched with respect to primitive-
mantle values. Nearly all fluid-mobile elements are greatly en-
riched in fluid phase relative to their concentration in residue,
except for B, which is only enriched by factor of ~5. For B,
lines that deviate from shaded areas represent data points
from reversal experiment in which starting B concentration in
fluid was 1000 ppm.
GEOLOGY, October 2004

Figure 5. Starting material (S, solid symbols) and postex-
periment concentrations (fluid—F, open symbols; residue—
R, shaded symbols) for B and Cs from three experiments.
Reversal experiment was performed in which B spike (1000
ppm) was introduced to fluid. During this experiment, sig-
nificant B was extracted by olivine-rich residue, and B par-
tition coefficients (F/RDB) were nearly identical to those in
undoped experiments. This points to efficacy of olivine in
hosting B within crystal structure. All three experiments
yield F/RDB 5 ~3–5, in contrast to those for Cs that are at
least two orders of magnitude greater, graphically empha-
sizing difference in behavior between two supposedly fluid
mobile elements.
the starting material, which is consistent with 10 wt% fluid being pro-
duced during antigorite breakdown. The exception is B, which exhibits
a significantly lower partitioning into the fluid: F/RDB 5 ;3–5. This
result is in marked contrast to previous experimental studies, which
mimic either B mobilization in fluids during low-temperature (200–
350 8C) interaction with sediments (You et al., 1995) or high-
temperature mantle conditions (900 8C, Brenan et al., 1998). These
studies document strong partitioning of B into the fluid phase (F/RDB
. 10). We suggest that B release is limited during dehydration of
serpentinite because of the high amount of olivine in the residue. Boron
partitioning between olivine and orthopyroxene (Dol/opxB ) and olivine
and chlorite (Dol/chl
B ) from natural subduction-zone rocks range from
1.3–3 and 3–4, respectively, showing that olivine is the most important
host for B (J. Hermann, 2004, personal commun.).
To test the obtained partitioning we ran a compositional reversal,
in which the diamond trap was doped with B so that the liberated fluid
would contain 1000 ppm B. The measured fluid composition after the
run was much lower at ;300 ppm, whereas the residue (75 ppm B)
showed a strong enrichment with respect to the undoped experiment
(Fig. 5). This result demonstrates that the fluid is able to equilibrate
with the residue in the experiments and confirms that B is much less
incompatible than other fluid-mobile elements in an olivine-orthopyroxene
residue. Therefore, the combined evidence from analyses of residue
and quench material from three experiments using different spikes in-
dicates a F/RDB during serpentine breakdown of 3–5. Best-estimate par-
tition coefficients for most trace elements analyzed are given in Table
1 for serpentinite.
The experiments show that the incompatible elements incorporat-
ed into serpentinites during interaction of oceanic lithosphere with sea-
water are later released during breakdown of serpentinite. Our parti-
tioning data (Fig. 4) combined with the distribution of key trace
elements in subducting lithosphere (Fig. 1) suggest that such fluids may
contribute greatly to the enrichment of B, As, Cs, and to a lesser extent
Pb, in arc magmas. Although our experiments show that elements such
as Li, Be, Ba, and Rb are quite mobile, their generally low initial
concentrations in serpentinite indicate that enrichment of these ele-
ments in arc lavas must be derived mainly from subducted sediments
GEOLOGY, October 2004
TABLE 1. BEST ESTIMATE OF PARTITION COEFFICIENTS
Element Serpentinite F/RD*
B 4
As 60
Cs 250
Li 100
Be 5
Ba 30
Rb .10
Pb 250
Th 1.5
U .20
Sr .10
*F/RD is fluid/residue partition coefficient.
(Fig. 1). The Th and U contents of serpentinites are too low to be a
significant reservoir, but dehydration of serpentinite fractionates Th and
U, and U is enriched in the fluid. The experimentally determined fluid
trace element pattern is remarkably similar to trace element character-
istics determined in olivine-hosted fluid inclusions that formed during
serpentine breakdown (Scambelluri et al., 2001a). The estimated B con-
centration of these fluid inclusions combined with the measured B con-
centration of olivine yields a F/OlDB ; 5 (Scambelluri et al., 2004), in
excellent agreement with our experiments. However, the experiments
do not provide evidence for elevated concentrations of high field
strength elements such as Ti and Nb in the fluid, as suggested by
Scambelluri et al. (2001a).
DISCUSSION
Boron is the main focus of this work because serpentinites are by
far the most important reservoir for B in the subducting lithosphere
(Fig. 1) and because the partitioning behavior of B is different from
the other fluid-mobile elements (Fig. 4). Our results, together with oth-
er field-based studies (Kent and Rossman, 2002; Scambelluri et al.,
2004), suggest that the olivine crystal structure can accommodate sig-
nificant B under certain conditions. The relatively high partition co-
efficients reported here for B have several large-scale implications. By
using our partition coefficients for B, a typical serpentinized oceanic
crust containing ;60 ppm B (Leeman and Sisson, 1996) would retain
12–20 ppm B upon dehydration, well above the primitive-mantle value
of 0.3 ppm (McDonough and Sun, 1995). In fact, it has been observed
that natural olivine originating from high-pressure breakdown of anti-
gorite retains high B contents of ;10–20 ppm (Scambelluri et al.,
2004). If significant proportions of subducting slabs are serpentinized,
then the B-bearing olivine represents a potentially large flux of B that
is recycled back into the deep mantle. The results from this work sug-
gest that high-B olivines from xenoliths (Kent and Rossman, 2002) or
ophiolitic sections are a fingerprint for metasomatized or previously
serpentinized mantle. Furthermore, high ratios of B to other incom-
patible elements in intraplate lavas might indicate the involvement of
recycled oceanic lithosphere with a B concentration greatly exceeding
that of the primitive mantle. High B concentrations and positive B
isotope in certain backarc basin systems have also been used to infer
that significant B is transported to the deep mantle (Palmer, 1992; You
et al., 1994). Serpentinized peridotites have isotopically heavy B (d11B
ø 110‰; Spivack and Edmond, 1987) and thus provide a possible
source of heavy B, either via dehydrating fluids or by incorporation of
B-rich ultramafic rocks into the source region of backarc-basin basalts.
The results presented here also raise questions as to whether the
mantle wedge filters B from migrating slab-derived fluids. It is clear
from many studies that arc magmas are highly enriched in B relative
to mid-oceanic-ridge basalt and oceanic-island basalt, reflecting the
high B flux from oceanic sediments, altered oceanic crust, and possibly
serpentinized lithospheric mantle (Rüpke et al., 2002). However, the
relatively low partition coefficients reported here suggest that flow
867
through the mantle wedge must be highly channelized in order for
fluids to maintain high B concentrations until partial melting eventually
occurs in hotter parts of the mantle wedge. Channelized flow would
result in a high fluid/rock ratio during transit and ensure that filtration
of B by olivine was limited. Variable uptake of B by olivine might
also be partly responsible for the reduced B concentrations of lavas
toward the backarc of many subduction zones (Ishikawa et al., 2001;
Morris et al., 1990; Noll et al., 1996; Ryan et al., 1996). Such fluids
originating from deeper parts of the subducting slab would have a
greater transit distance across the mantle wedge and could therefore be
expected to lose more B before partial melting.
If significant olivine-hosted B is recycled into the deep mantle
during subduction, then current estimates for B concentrations in prim-
itive mantle may be significant underestimations. The most recent es-
timate of 0.3 ppm (McDonough and Sun, 1995), in the middle of pre-
vious estimates, is based on K contents from various oceanic basalts
and assumes that K/B remains approximately constant during mag-
matic processes. Our partitioning data suggest that B is much more
compatible in olivine than K and that higher B concentrations are pos-
sible. However, high-B mantle domains are likely to be heterogeneous
and restricted to regions that have undergone B metasomatism or in-
corporated large amounts of subducted oceanic lithosphere. Consider-
ing the importance of B to a number of global petrologic problems,
more work is needed to fully understand the behavior of B during high-
temperature hydrothermal processes; much of this work will be made
possible by using new and innovative experimental techniques.
ACKNOWLEDGMENTS
We thank B. Leeman and an anonymous reviewer for constructive reviews
of the manuscript and M. Norman and M. Scambelluri for their comments on
the paper. We also thank to C. Allen, H. Kokkonen, and M. Shelley for technical
assistance related to the experiments. Support for this work was provided by
the Australian Research Council and the Swiss National Science Foundation.
REFERENCES CITED
Boillot, G., Feraud, G., Recq, M., and Girardeau, J., 1989, Undercrusting by
serpentinite beneath rifted margins: Nature, v. 341, p. 523–525.
Bonatti, E., Lawrence, J.R., and Morandi, N., 1984, Serpentinization of oceanic
peridotites: Temperature dependence of mineralogy and boron content:
Earth and Planetary Science Letters, v. 70, p. 88–94.
Brenan, J.M., Ryerson, F.J., and Shaw, H.F., 1998, The role of aqueous fluids
in the slab-to-mantle transfer of boron, beryllium, and lithium during sub-
duction: Experiments and models: Geochimica et Cosmochimica Acta,
v. 62, p. 3337–3347.
Cannat, M., 1993, Emplacement of mantle rocks in the sea-floor at midocean
ridges: Journal of Geophysical Research, v. 98, p. 4163–4172.
Dobson, D.P., Meredith, P.G., and Boon, S.A., 2002, Simulation of subduction
zone seismicity by dehydration of serpentine: Science, v. 298,
p. 1407–1410.
Hattori, K.H., and Guillot, S., 2003, Volcanic fronts form as a consequence of
serpentinite dehydration in the forearc mantle wedge: Geology, v. 31,
p. 525–528.
Ishikawa, T., Tera, F., and Nakazawa, T., 2001, Boron isotope and trace element
systematics of the three volcanic zones in the Kamchatka arc: Geochimica
et Cosmochimica Acta, v. 65, p. 4523–4537.
Johnson, M.C., and Plank, T., 1999, Dehydration and melting experiments con-
strain the fate of subducted sediments: Geochemistry, Geophysics, Geo-
systems, v. 1, doi: 10.1029/1999GC000014.
Kent, A.J.R., and Rossman, G.R., 2002, Hydrogen, lithium, and boron in
mantle-derived olivine: The role of coupled substitutions: American Min-
eralogist, v. 87, p. 1432–1436.
Leeman, W.P., and Sisson, V.B., 1996, Geochemistry of boron and its impli-
cations for crustal and mantle processes, in Grew, E.S., and Anovitz, L.M.,
868
eds., Boron: Mineralogy, petrology and geochemistry: Mineralogical So-
ciety of America Reviews in Mineralogy, v. 33, p. 645–707.
McDonough, W.F., and Sun, S.-s., 1995, The composition of the Earth: Chem-
ical Geology, v. 120, p. 223–253.
Morris, J., Leeman, W.P., and Tera, F., 1990, The subducted component in island
arc lavas: Constraints from Be isotopes and B-Be systematics: Nature,
v. 344, p. 31–36.
Noll, P.D., Jr., Newsom, H.E., Leeman, W.P., and Ryan, J.G., 1996, The role of
hydrothermal fluids in the production of subduction zone magmas: Evi-
dence from siderophile and chalcophile trace elements and boron: Geo-
chimica et Cosmochimica Acta, v. 60, p. 587–611.
O’Hanley, D.S., 1996, Serpentinites: Records of tectonic and petrological his-
tory: New York, Oxford University Press, 296 p.
Palmer, M.R., 1992, The boron isotope systematics of hydrothermal fluids and
tourmaline: A synthesis: Chemical Geology, v. 94, p. 111–121.
Peacock, S.M., 2001, Are the lower planes of double seismic zones caused by
serpentine dehydration in subducting oceanic mantle?: Geology, v. 29,
p. 299–302.
Ranero, C.R., and Sallarès, V., 2004, Geophysical evidence for hydration of the
crust and mantle of the Nazca plate during bending at the north Chile
trench: Geology, v. 32, p. 549–552.
Ranero, C.R., Morgan, J.P., McIntosh, K., and Reichert, C., 2003, Bending-
related faulting and mantle serpentinization at the Middle America Trench:
Nature, v. 425, p. 367–373.
Rüpke, L.H., Morgan, J.P., Hort, M., and Connolly, J.A.D., 2002, Are the re-
gional variations in Central American arc lavas due to differing basaltic
versus peridotitic slab sources of fluids?: Geology, v. 30, p. 1035–1038.
Ryan, J., Morris, J., Bebout, G., and Leeman, B., 1996, Describing chemical
fluxes in subduction zones: Insights from ‘‘depth-profiling’’ studies of arc
and forearc rocks, in Bebout, G.E., et al., eds., Subduction: Top to bottom:
American Geophysical Union Geophysical Monograph 96, p. 263–269.
Scambelluri, M., Bottazzi, P., Trommsdorff, V., Vannucci, R., Hermann, J.,
Gomez-Pugnaire, M.T., and Vizcaino, V.L.S., 2001a, Incompatible ele-
ment–rich fluids released by antigorite breakdown in deeply subducted
mantle: Earth and Planetary Science Letters, v. 192, p. 457–470.
Scambelluri, M., Rampone, E., and Piccardo, G.B., 2001b, Fluid and element
cycling in subducted serpentinite: A trace-element study of the Erro-
Tobbio high-pressure ultramafites (Western Alps, NW Italy): Journal of
Petrology, v. 42, p. 55–67.
Scambelluri, M., Müntener, O., Ottolini, L., Pettke, T.T., and Vannucci, R., 2004,
The fate of B, Cl and Li in the subducted oceanic mantle and in the
antigorite-breakdown fluids: Earth and Planetary Science Letters, v. 222,
p. 217–234.
Spivack, A.J., and Edmond, J.M., 1987, Boron isotopic exchange between sea-
water and the ocean crust: Geochimica et Cosmochimica Acta, v. 51,
p. 1033–1043.
Stalder, R., Foley, S.F., Brey, G.P., and Horn, I., 1998, Mineral–aqueous fluid
partitioning of trace elements at 900–1200 8C and 3.0–5.7 GPa: New
experimental data for garnet, clinopyroxene and rutile and implications
for mantle metasomatism: Geochimica et Cosmochimica Acta, v. 62,
p. 1781–1801.
Tatsumi, Y., Hamilton, D.L., and Nesbitt, R.W., 1986, Chemical characteristics
of fluid phase from a subducted lithosphere and origin of arc magmas:
Evidence from high-pressure experiments and natural rocks: Journal of
Volcanology and Geothermal Research, v. 29, p. 293–309.
Ulmer, P., and Trommsdorff, V., 1995, Serpentine stability to mantle depths and
subduction-related magmatism: Science, v. 268, p. 858–861.
You, C.F., Butterfield, D.A., Spivack, A.J., Gieskes, J.M., Gamo, T., and Camp-
bell, A.J., 1994, Boron and halide systematics in submarine hydrothermal
systems: Effects of phase separation and sedimentary contributions: Earth
and Planetary Science Letters, v. 123, p. 227–238.
You, C.F., Spivack, A.J., Gieskes, J.M., Rosenbauer, R., and Bischoff, J.L.,
1995, Experimental study of boron geochemistry—Implications for fluid
processes in subduction zones: Geochimica et Cosmochimica Acta, v. 59,
p. 2435–2442.
Manuscript received 2 March 2004
Revised manuscript received 3 June 2004
Manuscript accepted 9 June 2004
Printed in USA
GEOLOGY, October 2004