Journal of the Geological Society, London, Vol. 159, 2002, pp. 347–350. Printed in Great Britain.

SPECIAL

Formation of gigantic
gypsum crystals
J . G A R C I A - G U I N E A 1, S . M O R A L E S 2,
A . D E L G A D O 3, C . R E C I O 4
& J. M . C A L A F O R R A5
1
Dept. Geologı́a. Museo Nacional Ciencias Naturales
(CSIC) Madrid, 28006 Madrid, Spain
(e-mail: guinea@mncn.csic.es)
2
Dept. Mineralogia y Petrologia, Facultad Ciencias, Univ.
Granada, Granada 18002, Spain Fig. 1. Geode of the Jaravias mine (Pulpı́, Almerı́a, SE Spain).
3
Lab. Isotopos Estables, Estación Experimental Zaidin Transparent gypsum crystals
(CSIC), Granada 18008, Spain
4
Lab. Isotopos Estables, Facultad de Ciencias, Univ.
Salamanca, Salamanca 37008, Spain
5 Geological setting. The mineralized veins and masses of the
Dept. Hidrogeologia y Quimica Analı́tica, Univ. Almerı́a,
Almerı́a 04120, Spain Jaravias mine are related to the Miocene volcanic events, and are
analogous to those observed in nearby areas (sulphide–sulpho-
salts-bearing barite–siderite veins of the Sierra Almagrera
(Martinez Frias 1998; Morales 1994); volcano-sedimentary bar-
luid inclusions and stable isotopes of giant trans-

F
ite, Fe–Mn oxides and marine sediments at Herrerı́as (Martinez
parent gypsum crystals in a huge geode recently Frias 1998; Morales 1994); dacitic rocks with vertical veins of
discovered in Pulpı́ (SE Spain) are useful tools for Fe–Mn oxides and barite at La Serrecica). The geode is hosted
explaining its geological formation. Fluid inclusions by dolomites. The chronological sequence of mineralization is as
suggest a mixture of fluids, starting with freshwater and follows (from host rock to geode core): dolomite host rock,
including seawater in more advanced stages of the gypsum siderite, Fe–Mn oxy-hydroxides and gypsum crystals, which very
crystal growth. The isotopic composition of the hydration often included celestine needles.
water in the gypsum crystals agrees with a meteoric fluid.
The gypsum is enriched in 34 S, which denotes genetic links Experimental procedure. Optical and micro-analytical techniques (pet-
with marine sulphates via freshwater dissolution-recrystalli- rography, SEM, electron microprobe), micro-thermometry of fluid inclu-
zation of earlier marine evaporites. sions and stable isotopes (ä34 S CDT in sulphides and sulphates; ä18 O
SMOW in sulphates and gypsum hydration water; äD SMOW in gypsum
Keywords: gypsum, geodes, fluid inclusions, sulphur isotopes. hydration water) have been used to study the gypsum crystals and their
host rocks. Transparent gypsum crystals were also analysed by ICP-MS,
Spectacular colour pictures of huge gypsum crystals in a geode XRF and AAS. Melting temperatures of the last ice crystal of the fluid
inclusions can also be used (Attia et al. 1995) to infer the origin of the
from Pulpı́ (Almerı́a, Spain) were widely publicized in the
fluids. This micro-thermometric study was carried out on gypsum chips.
worldwide media. The impact of these images stems from an ä34 S and ä18 O in gypsum reflect the corresponding values of dissolved
exceptional combination of the huge size, transparency and SO4 2
in the mineralizing fluid. Data on natural samples and experi-
perfection of the crystals in a large rock cavity (Fig. 1). The mental results indicate that isotope ratios of the SO4 2
anion are not
Pulpı́ geode is 8 m long, 1.8 m wide and 1.7 m high and is significantly affected by dissolution-recrystallization processes at low
studded with 1 m long gypsum crystals. It is located in the temperature (Claypool et al. 1980; Chiba et al. 1981). Analytical results
Jaravias Fe–Pb mine (Pulpı́, Almerı́a) at about 50 m depth, 3 km can be obtained from the Society Library or the British Library
from the sea, and 0 m a.s.l. and is hosted by Triassic dolomite of Document Supply Centre, Boston Spa, Wetherby, West Yorkshire LS23
the Betic Cordillera. The gypsum crystallographic lattice has 7BQ, UK as Supplementary Publication No. SUP 18174 (3 pages).
layers of SO4 2
tetrahedra bound together by Ca2þ and layers of
water molecules, which gives a perfect cleavage parallel to (010). Results and discussion. The transparent gypsum crystals are
Simple precipitation experiments of Bosbach & Rammensee pure CaSO4. 2H2 O with small and variable Sr (0.02–0.13%) and
(1994) demonstrated that speed is constant for crystal growth in Hg (0.01–0.23%), and minor amounts of other cations (Zn, Fe,
different directions and the anisotropy of the crystal habit remain Mg, Mn, Ba). Fluid inclusions (Fig. 2) in the gypsum crystals
constant during crystal growth. Crystal and dissolution processes have polyhedral morphologies (six-sided, trapezoidal, rhomb-
on the (010) surface of gypsum is a layer-by-layer process, which shaped, negative crystal shape, etc.); are very large (up to
is dominated by the movement of [100] and [001]. The large 250 ìm), two-phase (liquid þ vapour) at room temperature, and
crystal sizes and perfect shape, high transparency and minor solid have a fairly constant degree of filling (liquid/(liquid þ
inclusions, resulting in such clean crystals, must have been vapour) ¼ 0.85 to 0.95). Petrographic observations have estab-
produced in a unique environment. A suitable explanation of the lished the presence of two types of fluid inclusions: (a) primary
geode’s origin must address both the cavity formation and fluid fluid inclusions (PFI), that are generally found isolated and
characteristics, which produced these exceptional crystals. randomly distributed throughout the gypsum crystals and (b)

347
348 J. G A R C I A - G U I N E A E T A L .

to 0.5 and 3.7 wt% eq. NaCl). The decrease in Tmice values
(mainly in PSFI), which reflects an increase in solute content
with time, suggests a fluid mixing, starting with freshwater and
including a different type of water in more advanced stages of
the system. The dispersion of the Tmice values to
2.2 8C
suggests that this later fluid may have been seawater (B in Figure
2) or recycled seawater (C in Fig. 2). All values of Tmice are very
different from D in Figure 2, which excludes a genesis from
evaporated seawater.
The evidence is compelling, therefore, that the area has been
exposed to both marine and meteoric environments, thus water of
different isotopic composition may have been involved in the
formation of the geode gypsum crystals. The crystals are
relatively enriched in 34 S (ä34 S ¼ þ18.6‰ to þ19.8‰, see Fig.
3b), which strongly suggests a genetic link with marine
sulphates, probably by means of dissolution-recrystallization of
Fig. 2. Fluid inclusion in negative crystal displaying the monoclinic earlier marine evaporites. äD values for the hydration water
symmetry of gypsum from a (010) section. Frequency histogram of Tmice extracted from the gypsum crystals (ranging from
64.2‰ to
values (melting temperatures of the last ice crystal of the fluid

65.4‰ SMOW), however, demand involvement of fresh meteo-
inclusions). Dots (A), (B), (C) and (D) show published Tmice values by
ric waters.
Attia et al. (1995), as follows: (A) Freshwater satured with gypsum
(Tmice ¼ 0.08C); (B) Normal seawater (Tmice ¼
1.9 8C); (C) ‘Recycled’
Experiments at temperatures from 17 to 57 8C (Fontes &
seawater (seawater saturated with gypsum derived from dissolution)
Gonfiantini 1967) have shown that hydrogen isotope fractionation
(Tmice ¼
2.0 8C) and (D) Seawater evaporated to gypsum saturation between hydration water of gypsum crystals and the water from
point (Tmice ¼
7.0 to
8.0 8C). which the crystals grow in equilibrium is not temperature
dependent; hydration water being 15‰ depleted in D with
respect to environmental water. Therefore, an estimated äD value

pseudo-secondary fluid inclusions (PSFI), arranged in rows that

never cut the crystal boundaries. The presence of both types of
fluid inclusion allows characterization of two different fluids
during gypsum crystal development, and hence we can describe
the evolution of the mineralizing solutions from initial (PFI) to
more advanced stages (PSFI). The main results of microthermo-
metric characterization of the fluid inclusions are the following:
(a) eutectic temperatures are very similar (from –51 to –38 8C)
in both PFI and PSFI, suggesting an H2 O–NaCl–CaCl2 system
and (b) melting temperatures of the last ice crystal (Tmice ) (Fig.
2), in PFI fall in a narrow interval of temperature (between –0.3
and 0.0 8C) whereas this interval is broader for values obtained
in PSFI (between –2.2 and –0.1 8C). Homogenization tempera-
tures could not be determined due to loss of water, and
consequent loss of transparency, in gypsum during heating. From
microthermometric study, salinity and density of fluid inclusion
can be estimated (Potter et al. 1978; Shepherd et al. 1985).
Values of salinity range from 0.0 to 0.5 wt% eq. NaCl for PFI
and from 0.2 to 3.7 wt% eq. NaCl for PSFI. Densities are similar
for both PFI and PSFI (0.8–0.9 gr cm
3 ), suggesting that fluids
were always less dense than seawater (dseawater ¼1.02 gr cm
3 ).
In view of the fluid inclusion study, a possible direct sedimen-
tary (evaporitic or vadose zone) origin (e.g. Mediterranean
Messinian salinity crisis) must be disregarded since the random
distribution of the fluid inclusions in the crystals differs signifi-
cantly from those described in gypsum of evaporitic origin (Attia
et al. 1995), where fluid inclusions are distributed in bands of
uniform salinity, with marked contrast in salinity between bands,
and the degree of filling in PFI and PSFI in this study is quite Fig. 3. (a) ä18 O of gypsum crystals, barite and celestite from Pulpı́,
constant (from 0.85 to 0.95), which indicates that gypsum compared to marine sulphate and Messinian evaporites. A1 and A2
precipitation could not have occurred in a fluid regime typical of values from Longinelli & Craig (1967) and Longinelli (1979). A3 and
the vadose zone (Goldstein & Reynolds 1994). The Tmice values A4 from this study. (b) ä34 S of gypsum crystals, barite, celestite and
of fluid inclusions in Pulpı́ gypsum (Fig. 2) show a mode near sulphides from Pulpı́ compared to marine sulphate, Messinian evaporites
0 8C, which suggests precipitation from non-marine water (i.e., and barite from neighbouring hydrothermal deposits. B1 and B2 values
surface water or ground-water) (A in Fig. 2). However, the Tmice from Claypool et al. (1980) and Longinelli (1979) respectively. B3 values
values have a dispersion range (
0.3 to 0.0 8C for PFI and
2.2 from Martinez Frias (1998) and Morales Ruano (1994). B4 to B6 from
to
0.1 8C for PSFI, corresponding to respective salinities of up this study.
 G I G A N T I C G Y P S U M C RY S TA L S 349

near
50‰ (SMOW) can be derived for environmental water

from the measured Pulpı́ values. Moreover, sulphate ä18 O values
(þ17.6‰ to þ20.3‰ SMOW) are rather high with respect to
present marine sulphate values (þ9.4‰). These data are incon-
sistent with direct precipitation of gypsum from dissolved marine
sulphate (see Fig. 3a). However, different, repeated low tempera-
ture episodes of dissolution-precipitation could explain these
values, as kinetic effects during sulphate crystallization can
produce a small increase of several units per mil in the ä18 O
values (Longinelli & Craig 1967; Lloyd 1968; Longinelli 1979).
This hypothesis of ‘inherited-recycled SO4 2
’ agrees with 87 Sr/
86
Sr ratios and ä18 O data in sulphates and crystallization water
of Messinian gypsum, supporting the involvement of meteoric
water in the genesis of the evaporitic deposits in the area
(Longinelli 1979; Playa et al. 2000). These values are also
coincident with those of sedimentary gypsum from neighbouring
basins where originally marine sulphates, recycled by meteoric
waters, present relatively high ä18 O values (Playa et al. 2000),
and are consistent with ä13 C and ä18 O values in the host
carbonate (authors’ own data), that require interaction with fluids
of meteoric derivation.
The chronological sequence of mineralization at Jaravias mine
provides a framework for the formation of the cave hosting the
geode and its subsequent filling by gypsum mega-crystals, as
follows (Fig. 4a, b and c). Alternating hydrothermal–volcanic
fluid pulses deposited barite–strontianite (anhydrous Ba–Sr
sulphates), sulphides and sulphosalts and Fe–Mn oxy-hydroxides,
controlled by discontinuities (fissures, vertical faults and metape-
lite–dolomite lithologic contacts). The incoming fluid reacted
with the dolomite (CaMg(CO3 )2 ) host rock producing siderite
(FeCO3 ); excess oxy-hydroxides remained in vein cores, e.g. as
powdered limonite (Fe(OH)3 .nH2 O) that can be observed in the
nuclei of other stuffed minor geodes. Subsequent karstification
processes (favoured by increased acidity caused by H2 SO4 in
solution) dissolved carbonates (dolomite and siderite) from
fissures and removed other soluble materials, producing cavities,
e.g. the large Pulpı́ geode cave where some siderite–goethite
remains on the internal walls. The final stage was dominated by
infiltration of supergene fluids, epsomite- and gypsum-bearing
(hydrous Mg–Ca sulphates), from which the gigantic gypsum
crystals grew. These crystals enclosed, protected, and thus
preserved, the very delicate celestine (SrSO4 ) needles that had
precipitated at earlier stages.
In summary, the formation of the giant gypsum crystals in the
geode may be explained in two steps: (1) formation of the cavity
by removal of soluble materials, previously deposited as result of
reaction between hydrothermal fluids and the host rock and (2)
infiltration of fluid causing the growth of large gypsum crystals.
Combined stable isotope and fluid inclusion data indicate that
this later fluid varied with time from freshwater in the earlier
stages to seawater-like in later stages, most likely as a result of
dissolution-recrystallization of earlier marine evaporites.

The work has been supported by Spanish projects DGICYT PB-98-0501,

PB-97-1211 and BTE-2001-3308 and RNM-0131 of the Junta de
Andalucia Research Group. Environment Council of Junta Andalucia is
also thanked for help. The authors wish to thank Ross Both (University
of Adelaide, Australia) and Adrian Boyce (SUERC, United Kingdom) for
helpful discussions concerning to different aspects of the manuscript.

References
Fig. 4. Simplified sketch of the geode formation: (a) hydrothermal Attia, O.E., Lowenstein, T.K. & Wali, A.M.A. 1995. Middle Miocene gypsum,
mineralization; (b) karstification and hydrothermal growth of celestite Gulf of Suez: marine or nonmarine? Journal of Sedimentary Research, A65-
needles; (c) growth of large gypsum crystals. 4, 614–626.
350 J. G A R C I A - G U I N E A E T A L .

Bosbach, D. & Rammensee, W. 1994. In situ investigation of growth and Longinelli, A. 1979. Isotope geochemistry of some Messinian evaporites;
dissolution on the (010) surface of gypsum by Scanning Force Microscopy. paleoenvironmental implications. Palaeogeography, Palaeoclimatology, Pa-
Geochimica et Cosmochimica Acta, 58, 843–849. laeoecology., 29, 95–123.
Claypool, G.E., Holser, W.T., Kaplan, I.R., Sakai, H. & Zak, I. 1980. The age Longinelli, A. & Craig, H. 1967. Oxygen-18 variations in sulfate ions in sea-
curves of sulphur and oxygen isotopes in marine sulphate and their mutual water and saline lakes. Science, 146, 56–59.
interpretation. Chemical Geology, 28, 199–260. Martinez Frias, J. 1998. An ancient Ba–Sb–Ag–Fe–Hg bearing hydrothermal
Chiba, H., Kusakabe, M., Hirano, S., Matsuo, S. & Somiya, S. 1981. Oxygen system in SE Spain. Episodes, 21, 248–251.
isotope fractionation factors between anhydrite and water from 1008C and Morales Ruano, S. 1994. Mineralogı́a, geoquı́mica y metalogenia de los
5008C. Earth & Planetary Science Letters., 53, 55–62. yacimientos hidrotermales del SE de España (Aguilas–Sierra Almagrera).
Fontes, C.H. & Gonfiantini, R. 1967. Fractionement isotopique de l’hidrogene PhD Thesis, Universidad de Granada.
dans l’eau de cristallisation du gypse. Comptes Rendus de l’Academie des Playa, E., Orti, F. & Rosell, L. 2000. Marine to non-marine sedimentation in the
Sciences, Paris, 65, 4–6. upper Miocene evaporites of the Eastern Betics, SE Spain; sedimentological
Goldstein, R.H. & Reynolds, T.J. 1994. Systematic of fluid inclusions in and geochemical evidence. Sedimentary Geology., 133, 135–166.
diagenetic minerals. SEMP (Society for Sedimentary Geology), Tulsa, Potter, R.W., Clayne, M.A. & Brown, D.L. 1978. Freezing point depression of
Oklahoma. aqueous sodium chloride solutions. Economic Geology., 73, 284–285.
Lloyd, R.M. 1968. Oxygen isotope behaviour in the sulphate-water system. Shepherd, T., Rankin, A.H. & Alderton, D.H.M. 1985. A practical guide to fluid
Journal of Geophysical Research, 73, 6099–6110. inclusions studies. Chapman and Hall, New York.

Received 10 January 2002; revised typescript accepted 4 April 2002.

Scientific editing by Richard England.