Volume 80, Issue 3-4

The intracrystalline partitioning of Fe and Mg in the octahedral sites of olivine is known to be dependent on temperature, pressure, and composition. Interpretations of the temperature effect on the partitioning have been mostly based on heating and quenching experiments, which seem to indicate that Fe 2+ preferentially orders into the Ml site with increasing temperature. The present single-crystal neutron diffraction experiments yield the first in-situ high-temperature structure refinements above 900 ºC and clearly indicate an ordering reversal above this temperature. Three data sets collected at 960, 1030, and 1060 ºC show a remarkable progressive decrease in the K D parameter with temperature, whereas the data at 880 ºC are consistent with a slight preference of Fe 2+ for M1, as reported in the literature. The effect is tentatively interpreted on the basis of competing contributions of configurational and vibrational entropy at high temperature, and it is expected to have profound implications for the thermodynamic modeling of olivine in the Earth’s mantle and in planetary processes.

We have investigated the structure of hydrated (Mg,Fe)SiO 3 glasses both at high pressure and after quenching, using infrared spectroscopy. Glasses were synthesized by fusion at pressures of up to 41 GPa and probed in situ to 39 GPa. Strong H bonding, as manifested by the presence of distinct OH stretching bands between 2250 and 2600 cm -1 , plays an important role in the speciation of OH groups at pressures between 15 and 25 GPa. At higher pressures, strong H bonding is also observed in the glasses but is completely unquenchable to ambient pressure; major structural changes are observed during decompression from pressures in excess of 20 GPa. Both the absence of a significant concentration of strongly H-bound OH groups and the similarity in frequency of the primary O-H stretching peak in the spectra of samples quenched from pressures above 30 GPa and those formed below 15 GPa indicate that similar structural features exist in glasses synthesized and quenched from these two separate pressure regimes. We infer that changes in Si coordination occur primarily between these two pressures and are largely unquenchable. Intratetrahedral H bonding is suggested as a possible mechanism for densification at pressures lower than ~25 GPa. Estimated bond lengths for high-pressure hydroxylated species imply that the hydrous component in silicate melts undergoes significant densification with pressure, and that the volume of dissolved H 2 O in high-pressure silicate melts is controlled almost solely by the volume of O anions.

Magnetic hysteresis properties of the Fe 3 O 4 -MgAl 2 O 4 solid solution have been measured at 4.4 K and in fields up to 12 T. The trend in saturation magnetization, M s (4.4 K), as a function of composition is consistent with the cation distribution model of Nell and Wood (1989) and suggests that M s (4.4 K) changes sign at a composition of 30 mol% Fe 3 O 4 . The change in sign of M s (4.4 K) at this composition is correlated with an anomalous peak in coercivity of over 700 mT, This effect has been explained in terms of fluctuations in both composition and the degree of nonconvergent cation order, which lead to fine-scale magnetic domains with antiferromagnetic exchange coupling.

Brucite, Mg(OH) 2 , was investigated by Raman spectroscopy to pressures of 36.6 GPa under nonhydrostatic conditions and to 19.7 GPa under quasi-hydrostatic conditions. Several new Raman lines are first observed at 4 GPa, demonstrating the existence of a high-pressure structural change. One of the new lines grows over a broad pressure interval, and this growth can be explained by a resonant interaction with the Egtranslational mode. Raman data are compared with recent infrared spectroscopy, X-ray diffraction, and neutron diffraction studies of brucite. The structural change is likely to involve displacement or disordering of the H atoms, consistent with neutron diffraction results. The Ramanactive O-H stretching vibration in brucite decreases with pressure at the rate of ~7 cm -1 / GPa, larger than the pressure dependence of the infrared-active stretching vibration by more than a factor of ten. The primary differences in the Raman spectra of Mg(OH) 2 and Ca(OH) 2 are that the O-H vibrational frequencies in Mg(OH) 2 vary linearly with pressure, and the O-H stretching vibration band width increases with pressure at a rate that is an order of magnitude lower for Mg(OH) 2 than for Ca(OH) 2 .

We have determined the quartz-coesite transition by reversed experiments in a pistoncylinder apparatus in the range 500-1200 °C. The difference between the sample pressure and apparent pressure was calibrated by (1) studying the friction decay in the hysteresis loop defined by the relationship between apparent ("nominal") pressure and piston position in the compression and decompression cycles, and (2) determining the melting temperature of LiCl by DTA in pressure cells similar to those used in the reversal experiments and comparing the results with those determined in the gas apparatus. The equilibrium transition boundary can be expressed as P (kbar) = 21.945 (±1855) + 0.006901 (±0003)T (K). It is subparallel to, but ~1.5 kbar higher than, the transition boundary determined by Bohlen and Boettcher (1982). We have also retrieved the entropy [39.56 ±0.2 J/(mol. K)] and enthalpy off formation (-907.25 ±0.007 kJ/mol) from elements of coesite at 1 bar, 298 K., from our phase-equilibrium data and selected thermochemical data from the literature. From the characteristics of the hysteresis loop we conclude that the often-used practice of maintaining a constant nominal pressure by repeated pressure adjustment during an experiment leads to variation of pressure on the sample.

The p1̅ to I1̅ phase transition in end-member anorthite, CaAl 2 Si 2 O 8 , has been reversed in-situ at high pressures and temperatures in an internally heated diamond-anvil pressure cell using single-crystal X-ray diffraction. The unit-cell parameters of anorthite were determined at 48 temperature-pressure points up to 255 °C and 2.6 GPa. At high pressures and temperatures, from room temperature up to ~225°C, the transition is marked by a first-order step in the unit-cell volume and the complete disappearance of c and d reflections from the diffraction pattern. From room temperature to ~195 °C and 2.1 GPa, the transition boundary is linear and nearly isobaric, with a slope, dP/ dT, of -0.003 GPa/°C. From 195 to 240°C and 1.5 GPa the boundary is curved, with increasingly negative dP/dT, and the magnitudes of ΔV and the scalar strain associated with the transition decrease to about 75% of their high-pressure values of ~0.2 and ~0.011 %. From 1.5 GPa to room pressure, the boundary is isothermal at 240°C and is marked by the disappearance of the c and d reflections, although there are no detectable discontinuities in the unit-cell parameters. The phase transition remains unquenchable over the entire boundary. The distinct changes in character of the phase transition and the trajectory of the equilibrium boundary in P-T space are associated with a crossover in the I1̅ phase field that is marked by a sudden change in cell parameters of this phase over the pressure interval of ~1.5-2.0 GPa.

Two rectorite samples, a Na- and Ca-rich rectorite (sample 1) from Beatrix Mine, South Africa, and a Na-rich rectorite (sample 2) from Garland County, Arkansas, have been investigated quantitatively by solid-state 23 Na, 27 Al, and 29 Si magic-angle spinning (MAS) NMR spectroscopy, total chemical dissolution, cation exchange, and X-ray diffraction (XRD). Comparison of modeled and experimental diffractograms shows that both rectorite samples have a mica to smectite ratio of 1:1 and an ideal ordering in units of one mica plus one smectite layer. The average thicknesses of the coherent scattering domains are ten and eight 2:1 layers for samples 1 and 2, respectively. Quantification of the 23 Na, 27 Al, and 29 Si MAS NMR spectra has allowed determination of the compositions for the octahedral sheets and for the mica (paragonite and margarite) and smectite tetrahedral sheets and determination of the distribution of interlayer Na + ions between paragonite and margarite interlayers. Each 2:1 layer has an octahedral sheet of weak positive charge sandwiched between two tetrahedral sheets of weak and strong negative charge. The top and bottom tetrahedral sheets of the coherent scattering domains have strong negative charges and very high cation-exchange capacities for sample 1 and low negative charges for sample 2. Sample 1 is a three-component mixed layer (margarite, paragonite, and smectite), and sample 2 a two-component mixed layer (paragonite and smectite).

The study of a crystal-chemical model for Pbca orthopyroxene was undertaken using the results of X-ray single-crystal structure refinements and chemical analyses of about 200 samples with different compositions and degrees of order. Multiple linear correlations between compositional and structural variables were searched for using the statistic package SPSS. The coefficients and constant terms oflinear equations that allow the prediction of cell parameters and interatomic distances for any orthopyroxene, starting from its crystal- chemical formula, were calculated. From the predicted distances, the geometric refinement program DLS-76 yields the relevant atomic positions, which agree satisfactorily with those measured experimentally. The same results were obtained more quickly using coefficients that directly express the correlation between the atomic fractions at the structural sites and the atomic positions. The calculation of atomic positions makes it possible to study the effects induced on the structure by any variation of chemical composition and cation distribution; in particular it has allowed the prediction of structural properties for fictive end-members. The coefficients for the calculation of cation-O mean bond distances of the regular polyhedra M1 and SiB appear to be nearly correlated to the ionic radii of the cations at these sites, thus justifying the introduction of (M1-O) and (SiB-O) bond distances in the linear equations used for determining or checking M1 and SiB site populations. The prediction of atomic positions can be used for lattice-energy calculations, in order to develop a structure-energy model for orthopyroxene. The coefficients and constant terms of linear equations for the atomic fractions as a function of cell parameters, atomic positions, and mean atomic numbers at M1 and M2 sites were also calculated in order to test the possibility of predicting, for any orthopyroxene, its cation distribution in the absence of chemical analysis.

The dehydration behavior of an analcime formed by ion exchange from leucite (X-type analcime, or secondary analcime) is contrasted with the behavior of an analcime formed by hydrothermal recrystallization from structurally dissimilar materials (hydrothermal analcime, or H-type analcime). The kinetics of dehydration of X-type analcime is consistent with a high specific surface area, showing surface equilibration with H 2 O vapor and fast initial H 2 O loss. The kinetic data approximately fit the Austin-Rickett equation, yielding an empirical activation energy for the dehydration of X-type analcime of 33 kl/mol, compared with 92 kl/mol for H-type analcime. Washed and ultrasonically cleaned samples of the same grain size have very different BET (N adsorption) surface areas, being 20 m 2 /g for X-type analcime and 2 m 2 /g for H-type analcime, consistent with the high porosity of the surface of X-type analcime, as observed by SEM. Further analysis of the N adsorption data shows the pore size distribution for the X-type analcime to have a maximum at 100 Å. The rate of Na drift under electron microprobe is more than 15 times higher for X-type analcime than for H-type analcime and could be a convenient means of distinguishing the two parageneses.

Enthalpies of solution have been measured for an II-member analbite-sanidine series at 704 °C in molten lead borate. Enthalpies of K-Na mixing determined from the data are statistically equivalent to those determined by HF solution calorimetry at 50 °C (Hovis, 1988). This indicates that regular solution thermodynamic models, which assume enthalpies of mixing to be independent of temperature, are valid for the disordered alkali feldspars. The experiments also demonstrate the consistency of the HF and lead borate solution calorimetric techniques. Volumes of K-Na mixing at room temperature also are reported.

The technique of palladium oxide equilibration was used to measure activities of MgO and Al 2 O 3 in stoichiometric MgAl 2 O 4 spinel at 1150-1400 ºC. Activities of MgO range from 0.63 ± 0.03 to 1.00 ± 0.07, and activities of Al 2 O 3 range from 0.09 ± 0.01 to 0.05 ± 0.01 (10). The activities yield free energies of formation of spinel from the oxides ΔG o f , ranging from -31 kJ/mol at 1150 °C to -39 kJ/mol at 1400 °C, with a precision of 2-19% (10). The derived values of ΔG o f are consistent with the equilibrium amount of disorder present in the spinel under experimental conditions because the measured activities reflect equilibration at high temperature. Calorimetric heat contents were corrected for the state of order using a Landau formulation for the equilibrium amount of disorder in spinel as a function of temperature. The corrected heat contents, together with the measured ΔG o f , were used to produce a new, self-consistent C p function that can account for vailable experimental data. The modeling results are consistent with previous suggestions that short-range ordering may be important in spinel.

The effect of the addition of 5, 10, and 20 wt% of the alkali oxides on the viscosity of a haplogranitic melt composition has been investigated at 1 atm and in the temperature range of 400-1650 °C. The high-temperature viscosity data were obtained with concentric cylinder viscometry and the low-temperature viscosity data using micropenetration viscometry. The combined data sets for low- and high-temperature viscosities have been fitted for each composition using the Tamann-Vogel-Fulcher (TVF) equation. The effect of alkali oxide on the viscosity of a haplogranite melt is extreme. The viscosity decreases with added alkali oxide content in a nonlinear fashion. The first few mole percent of alkali oxide added decreases viscosity several orders of magnitude, whereas subsequent addition of alkali oxide has a much smaller effect. The effects of each of the alkalis are broadly similar, implying that the structural role of the alkalis is common to all. In detail however, the viscosity of the strongly peralkaline melts investigated here increases with the size of the added alkali cation in the order Li < Na < K,Rb,Cs. This trend probably reflects a minor influence of the alkali-O bond strengths on the melt viscosity. This distinction of a dominant depolymerizing influence and a minor alkali specific bond-strength influence has important implications for the comparison of these data with those for the addition of other depolymerizing agents on the viscosity ofhaplogranitic melt (e.g., H 2 O, F 2 O -1 ).

Extrapolation of laboratory measurements of the viscosity (η)of silicate melts is frequently needed in order to analyze petrological and volcanological processes. Therefore a general understanding of silicate melt viscosities is required. In this paper we survey the present state of our knowledge and distinguish three flow regimes for homogeneous silicate liquids: (1) a low-viscosity regime (η < 1 Pa·s), where the viscosity obeys a temperaturedependence power law in accordance with mode coupling theory (these low viscosities are typical for depolymerized melts); (2) an intermediate regime (1 < η< 10 12 Pa·s), where silicate melt viscosity is determined by the availability of configurational states (the dependence of the viscosity on the temperature is described aptly by the configurational entropy theory of Adam-Gibbs); and (3) a high-viscosity regime, where the liquid has been transformed into a glass (η> 10 12 Pa·s) (this regime is not well known, but available measurements indicate an Adam-Gibbs or an Arrhenian temperature dependence of the glass viscosity). Examples are given of igneous rocks whose geneses were affected by these flow regimes.

Three methods of estimating H20 contents of geologic glasses are compared: (1) ion microprobe analysis (secondary ion mass spectrometry), (2) Fourier-transform infrared spectroscopy (FTIR), and (3) electron microprobe analysis using the Na decay-curve method. Each analytical method has its own advantages under certain conditions, depending on the relative importance of analytical accuracy, precision, sensitivity, spatial resolution, and convenience, and each is capable of providing reasonably accurate estimates of the H 2 O, or total volatile, content of geologic glasses. The accuracy of ion microprobe analyses depends critically on the availability of well-characterized hydrous standard glasses. Precision is often better than 0,2 wt% (1σ). The method provides good spatial resolution (~15 μm) and the capability to determine simultaneously the abundance of other volatile species of interest (e.g., F, B). FTIR spectroscopy provides excellent analytical sensitivity (~10 ppm), accuracy and precision (<0.1 wt%), and the capability to determine the abundance of H 2 O and CO 2 species (H 2 O, OH - , CO 2 CO 2- 3 ) in analyzed glasses, although the spatial resolution (> 25-35 μm) is not as good as that of the ion microprobe. The main advantages of the estimation of H20 contents of hydrous glasses using the electron microprobe are excellent spatial resolution (~10 μm) and analytical convenience. The disadvantages are that accuracy and precision (>0.5 wt%) are not as good as those associated with the other methods, but, for certain applications, these uncertainties may be acceptable for the estimation of H 2 O contents of H 2 O-rich (> 1 wt%) samples.

Crystals of La-, Gd-, and Dy-bearing fluorapatite [La-FAp, Gd-FAp, Dy-FAp; Ca 10-x-2y- Na y REE x+ y (P 1-x Si x O 4 ) 6 A 2 , with x = 0.24-0.29, y = 0,32-0.36; P6 3 /m] have been synthesized hydrothermally, and their structures refined at room temperature with single-crystal X-ray intensities to R = 0.015-0.018. Na is essentially restricted to the Cal position in La-FAp, Gd-FAp, and Dy-FAp, in contrast to Nd-FAp, which was synthesized under slightly different conditions and has appreciable Na in Ca2 as well. Site occupancies for REE in Cal and Ca2, respectively, are 0.023(1) and 0.093(1) in LaFAp, 0.038(1) and 0.111(1) in Nd-FAp, 0.038(0) and 0.077(0) in Gd-FAp, and 0.039(1) and 0.060(1) in Dy- FAp. The REE site occupancy ratio (REE-Ca2 to REE-Ca1) appears to decrease monotonically for REE 3+ cations through the 4f transition-metal series. With this assumption, site occupancy ratios (REE-Ca2 to REE-Ca1) for some other REE in natural apatite are estimated to be: La 4.04, Ce 3.67, Pr 3.30, Nd 2.92, Sm 2.47, Eu 2.25, Gd 2.03, Dy 1.54, Y 1.29, Er 1.05. These single-REE site occupancy ratios may not be transferrable to natural apatite. The Ca1 and Ca2 site occupancies are generally consistent with site preferences deduced from bond-valence calculations, which show that the substitutions for Ca lead to equalization of Ca1 and Ca2 bond valences. Also, the REE site occupancy ratio correlates inversely with F bond valence.

We present a method to measure carbonate content of apatite that uses infrared spectroscopy, The method was calibrated against the amount of CO 2 released by reaction of apatite with phosphoric acid and may be used to study apatite with CO 2 content below I wt%. We have applied the method to apatite from carbonatite (≈0.3-0.4 wt% CO 2 ) and metasomatic (≈0.6-1.0 wt% CO 2 ) and alkaline silicate (<0.3 wt% CO 2 ) rocks from the Jacupiranga alkaline complex. The results show that higher concentration of carbonate was found in apatite related to metasomatic processes and, in particular, to metasomatism related to the intrusion of carbonatite from J acupiranga. Major and trace elements of the analyzed apatites do not show a simple substitution mechanism that would explain the presence of carbonate in the apatite structure. In a series of exploratory laboratory experiments in which apatite was synthesized under high temperature and pressure, no clear relationship was observed between the carbonate content of the synthetic apatite and the total C content in the coexisting fluid (molecular CO 2 or carbonate ions in solution). However, even though these experiments were not conclusive, they suggest that apatite coexisting with carbonate ions in solution has a CO 2 content above 0.3 wt%, whereas apatite coexisting with molecular CO 2 has a CO 2 content <0.3 wt%. These results indicate that the CO 2 content of apatite depends not only on total fluid C content, but also on the C speciation in the fluid.

The oxidation of synthetic ferruginous biotite in hydrothermal systems was studied for two ideal compositions, annite, KFe 3 AlSi 3 O 10 (OH) 2 and siderophyllite, KFe 2.5 Al 2 Si 2.5 O 10 (OH) 2 . These biotite samples were annealed in Shaw-type vessels under controlled f H2 at a total pressure of 1 kbar. The ratio of Fe 2+ to total Fe, determined by wet-chemical analysis, is directly proportional to the f H2 regardless of temperature variations. For f H2 values of < I bar to 102 bars, this ratio varies from 0.72 to 0.90 for annite and from 0.76 to 0.96 for siderophyllite. The H contents of the biotite do not vary with FeH content for either composition and result in 94 and 100% of the OH sites being occupied in annite and siderophyllite, respectively. A consistent decrease in total octahedral cation content with increasing FeH content for both annite and siderophyllite suggests that the Fe-vacancy substitution 3Fe 2+ ⇌ 2FeH 3+ + [6] Vac plays a dominant role in the oxidation of these synthetic biotite samples.

New data are presented for the ΔG 0 f for huntite, CaMg 3 (CO 3 ) 4 . Measurements were made using new third-kind electrochemical cells that do not require a complicated model for solute speciation or corrections for liquid-liquid junction potentials. The average value of ΔG 0 f [CaMg 3 (CO 3 ).(S)] (-4195.6 ± 6.4 kJ/mol) measured in these electrochemical cells agrees, within experimental error, with values determined calorimetrically (e.g., -4203.4 ± 1.6 kJ/mol: Hemingway and Robie, 1973) and electrochemically (-4188.6 ± 1.0 kJI mol: Gamsjäger, 1989). Our results demonstrate the utility of this new electrochemical method in the study of metal carbonate equilibria of geochemical importance.

Sillimanite from the regional sillimanite zone in west-central New Hampshire is fibrolitic and overprints F 2 folds (nappe stage) of earlier mica foliation. Regional sillimanite zone samples show no evidence for earlier staurolite parageneses, despite the fact that staurolite is abundant at lower grade, because sillimanite was produced directly from garnet + chlorite by the prograde (heating) reaction garnet + chlorite + muscovite + quartz = sillimanite + biotite + H 2 O. The pressure at which this reaction occurs is sensitive to the MnO and CaO contents of garnet, and phase-equilibrium arguments reveal that at the regional pressures of west-central New Hampshire (2-4 kbar), staurolite parageneses are only possible in rocks with low MnO + CaO. The inferred P- T path is counterclockwise with nearly isobaric initial heating at 2 kbar, followed by loading (± heating) to a peak metamorphic temperature of 600 ± 25 °C at 4 kbar, followed by nearly isobaric cooling. Garnets were compositionally homogenized near the metamorphic peak, but subsequent cooling was rapid (> 100 °C/m.y.). These data argue against the folded isotherm model of Chamberlain (1986). Instead, the present distribution of metamorphic grades is interpreted to be the result of regional stacking of high-grade thrust sheets on lower grade rocks, followed by depression of high-grade rocks to lower structural levels.

Nchwaningite, Mn 2+ 2 SiO 3 (OH) 2 · H 2 O, is an orthorhombic chain silicate [space group Pca2 1 Z = 4, a = 12.672(9), b = 7.217(3), e = 5.341(2) Å], occurring as aggregates shaped like pin cushions, together with calcite, bultfonteinite, and chlorite at N'chwaning mine, located in the Kalahari manganese field, northern Cape Province, South Africa. The aggregates consist of light brown, transparent needles, which average 1.0 × 0.1 × 0.05 mm in size. Nchwaningite is named after the mine N'chwaning II, where it was found first. Nchwaningite is biaxially negative with the refractive indices α= 1.681(2), β = 1.688(2), γ= 1.690(2), 2V x = 54.4(4)°. The optical orientation is X = b, Y = a, Z = c. Nchwaningite has two perfect cleavages parallel to (010) and (100). The calculated density is 3.202 glcm3. The chemical composition, as determined by electron microprobe analyses, indicates minor substitutions of Mg for Mn. The crystal structure, including H positions, was solved and refined from X-ray singlecrystal data to R = 2.14%, R w = 2.91%. The nchwaningite structure consists of double layers of laterally linked so-called truncated pyroxene-building units formed by a double chain of octahedra, topped with a Zweier single chain of Si tetrahedra. Symmetry-equivalent units are linked laterally but turned upside down. This yields a double-layer structure with H bridges linking the layers. A striking feature of the structure is that one Mn06 comer is formed by a H 2 O molecule. The tetrahedral chain and octahedral distortion of the new mineral is compared with pyroxenes having MnH in Ml [synthetic MnSiO 3 (P2 1 /c clinopyroxene) and johannsenite CaMn(SiO 3 ) 2 ]' To test a complete Ca and Mg substitution for Mn in the new nchwaningite structure type, distance least-square refinements were performed. It was found that Ca and Mg analogues would also yield reasonable interatomic distances and polyhedra.

The crystal structure of manandonite-2H 2 , an Al-, Li-, B-rich analogue of amesite, has been refined in space group C1 to R = 5.7%. Tetrahedral Si, AI, and B are partly ordered to give two mean T-O bond lengths near 1.603(1) A and two near 1.667(1) Å. B has a greater tendency to order than do Si and Al. There are one relatively Li-rich octahedron and two relatively Al-rich octahedra in each layer with mean M-O,OH bond lengths of 1.997(1), 1.960(1), and 1.955(1) Å, respectively, in layer 1 and 2.014(1), 1.941(1), and 1.956(1) Å, respectively, in layer 2. It is the partial ordering of the octahedral cations that decreases the ideal P63 symmetry to P1 (or C1, as used for refinement). Tetrahedral rotations of 18.3° are required to match the larger lateral dimensions of the tetrahedral sheet with those of the octahedral sheet, coupled with basal O corrugations because of the different sizes of the octahedra. The H + protons of the six surface OH molecules point directly toward their acceptor basal 0 atoms to give H bond contacts between 2.647 and 2.773 A.

Gillulyite, Tl 2 (As 3+ ,Sb 3+ ) 8 S 13 ,is monoclinic with space group P2/n and a = 9.584(3), b = 5.679(2), c = 21.501(6) Å, β = 100.07(2)°, V = 1152.2(7) Å 3 , and Z = 2, The average structure was determined using direct methods and refined to a final residual of 0.046 using 930 reflections. The structure consists of Tl-As-S- and As-S-bearing sheets, each present statistically 50% of the time. Four sheets, linked together by T1-S, As-S, and S-S bonds, are stacked parallel to (00 I), yielding the 21-Å c-axis repeat of the average unit cell. The Tll atom within the TI-As-S sheet is coordinated by six S nearest neighbors, which form a distorted trigonal prism. The partially occupied T12 position between the sheets is split symmetrically about the twofold axis with a separation of 1.28 Aand equal but partial occupancies of 25%. The disordering of the T12 position and extreme distortion of its coordination polyhedron is probably the result of the T1 + 6s 2 lone pair effect. The mean bond distances for the AsS 3 pyramids range from 2.263 to 2.319 Å. However, four of five pyramids have a fourth S neighbor at distances ranging from 2.943 to 3.001 Å, which may be rationalized in terms of the As 3+ 4s 2 lone pair effect. Edge sharing among the AsS 3 pyramids results in an As-As distance of 2.562 Å, comparable with that observed in elemental As. The presence of S-S and As-As bonding suggests that gillulyite might more appropriately be classified as a sulfide rather than a sulfosalt.

Pararealgar, a polymorph of realgar (α-As 4 S 4 ), crystallizes in the monoclinic space group P2 1 /c, with a = 9.909(2), b = 9.655(1), c = 8.502(1) Å, β = 97.29(1)°, V = 806.8(2) A 3 . A single crystal of pararealgar was obtained by light exposure of realgar. The structure was determined by direct methods and refined by least-squares technique to a final R index of 3.39%. The structure consists of discrete covalently bonded As.S. molecules, which are held together by van der Waals forces. In each molecule, one As atom links 2As + IS, another links 3S, and the other two link 1As + 2S. Therefore this molecule differs from that of α-As 4 S 4 but is the same as that found in As 4 S 4 (II). The difference between the structure of pararealgar and As 4 S 4 (II) is due to different molecular packing.