The noteworthy features in analysis A, as compared with an undoubted grossularite from California (analysis C.) are the low percentage of SiO₂ and the important amount of water given off at a temperature greater than 105† C., and these features are reflected in the physical properties. Both the refractive index (n) and the specific gravity of this hydrous mineral are very much lower than those usually quoted for calcium garnet. It seems clearly indicated then that the mineral belongs to the group of crystalline compounds termed hydrogarnets by Flint, McMurdie and Wells (1941, p. 14) or garnetoids by McConnell (1942, p. 452). These hydrogarnets or garnetoids are those compounds that have structures similar to that of true garnets. Closely related minerals are hibschite (Belyankin and Petrov, 1941) and plazolite described by Foshag (1920) and Pabst (1937). (For comparison see Table 2.) To date, only these two compounds are known to occur as natural minerals, although work by Flint, McMurdie and Wells (1941), on the reactions of the glass phase of portland cement clinker with water, has shown that complete isomorphism, or more correctly isotypism (McConnell, 1943, p. 98) exists between isometric tricalcium aluminate hexa-hydrate (3CaO Al₂O₃ 6H₂O), the corresponding ferrite (3CaO Fe₂O₃ 6H₂O), grossularite garnet (3CaO Al₂O₃ 3SiO₂), and andradite (3CaO Fe₂O₃ 3SiO₂). This new mineral then is a member of the 3CaO Al₂O₃ 6H₂O—3CaO Al₂O₃ 3SiO₂ solid solution series, to which plazolite and hibschite also belong.

A. Hydrogrossular, Champion Creek, Waimea S.D.

B. Hibschite, Nikortzminda, Georgia, U.S.A. (Belyankin and Petrov, 1941, p. 451, analysis No. 1).

C. and D. Plazolite, Crestmore, Riverside County, California (Foshag, 1920, 1924). In connection with the CO2 reported in these analyses, it should be noted that in a personal communication to Flint, McMurdie and Wells (1914, p. 19), Foshag states that small amounts of calcite account completely for CO₂.

Naturally in view of the large amount of water and the low percentage of silica relative to lime and alumina, a recalculation of the analysis on the basis of X₃Y₂(ZO₄)₃ is not in the least satisfactory. Faced with a similar problem as a result of work on griphite,

McConnell (1942) has shown that a variation of the usual garnet formula is necessary in order to accommodate minerals such as plazolite, hibschite, griphite, etc. Thus the formula X₃Y₂ (ZO₄)₃–m (OH)_4m has been derived as a variation of the normal expression X₃Y₂ (ZO₄)₃. The analysis of hydrogrossular is, therefore, recalculated on this basis (Table 3); it should be understood that members of the garnet-hydrogarnet series all have eight molecules to the unit cell.

The formula thus derived for hydrogrossular is:—

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

8 [Ca_{2 9} (Al.Fe”'.Fe”.Mg.)_2.0 (SiO₄)_2.5 (OH)_2.25]

This formula, though devised first by McConnell (1942, p. 456) to explain the composition of a hydrophosphate garnetoid, certainly seems to be satisfactory in this case, although the (OH) group is slightly too high for the number of silicon ions present.

Chemical tests on a number of hydrogrossulars of slightly different refractive index indicate that they are all completely soluble in acids. In one specific case 0.25 gms. of the analysed hydrogrossular was completely dissolved by 50% HCl after two hours on the water-bath. HNO₃ acted similarly, but with H₂SO₄ action was fairly slow and incomplete, due probably to the relatively insoluble nature of the CaSO₄ formed by the reaction.

The work of Flint, McMurdie and Wells (1941) on the structure of the 3CaO Al₂O₃ 6H₂O—3CaO Al₂O₃ 3SiO₂ series and that of Pabst (1937) on plazolite in particular, have shown that all members of this isomorphous series are isometric, while the former authors have found that the cube-size varies regularly from 11.84A for grossularite to 12.56A for tricalcic aluminate hexahydrate. For plazolite, a member of the series with a composition approximately grossularite 66⅔%, and tricalcic aluminate hexahydrate 33⅓%, Pabst has shown that the cell dimension is 12.14Å=0.01Å, that is slightly larger than in grossularite. The dimensions of the cell side of a molecule of the analysed mineral may be calculated if the usual formula is employed, viz.:—

[The section below cannot be correctly rendered as it contains complex formatting. See the image of the page for a more accurate rendering.]

A³₀=ZM 1.65 × 10^-24./P

Where Z = number of molecules in the unit cell (8 for the hydrogarnet series), M = molecular weight of mineral (437.41 in the case of the analysed hydrogrossular) and P = density of the crystal (3.35) (Bragg, 1937, p. 24). The value thus obtained is 11.99å, which is approximately the figure that would be expected from the data given by Flint, McMurdie and Wells (1941, p. 30).

With the small amount of information available an attempt has been made to correlate refractive index with chemical composition. It is to be regretted that Flint, McMurdie and Wells were unable to give optical data with the analyses of the iron-free members of the hydrogarnet series as was done in the case of the hydrated calcium alumino-ferrite solid solutions (Flint, etc., 1941, p. 16). By plotting silica percentage in the abscissa and refractive indices in the ordinate a curve is then drawn through the points representing tricalcic aluminate hexahydrate, compound 1 (Flint, etc., 1941, p. 15, Table 1), hibschite (plazolite), Dun Mt. hydrogrossular, and grossular (Text Fig. 1). In drawing this curve SiO2 and H2O only have been considered, but it is perfectly clear from the work of Flint, McMurdie and Wells that a slight replacement of alumina by ferric oxide leads to a rise in the refractive index, as would be expected. The hydrogrossular used in this work contained only 1.46% of total iron oxides and 2.07% of magnesia and these have been neglected entirely as far as their effect on refractive index is concerned. The (OH) ions appear to be very strongly held in the molecule, for on heating the fine powder of the analysed hydrogrossular to dull redness for a period of one hour, the refractive index rose only approximately 0.01, viz., from 1.7021 to 1.7115.

The refractive indices and densities of a number of hydrogarnets from different localities have been determined and not one has a value as high as grossularite (see Table 4). The material that was analysed was selected on account of homogeneity and low refractive index; this did not represent material of the lowest refractive index, however, for a specimen, No. 1580 (Otago University collection), showed a curious heterogeneous hydrogarnet. This consisted of what appeared to be partial and rare reaction rims surrounding material with n=1.723, while the rims had n = 1.693 and n=1.678; the latter material was far too rare to attempt any separation but must certainly have a composition very close to that of plazolite.

An examination of the finely ground powders of these minerals indicates in many cases that they are not truly isotropic but instead a combination of very low birefringence and moderate dispersion was observed. The strength of double refraction was certainly less than 0.001, while the minerals appeared to be isotropic in red light but very slightly anisotropic in blue.

It appears, therefore, as if true grossularite is not the usual constituent of the altered gabbros and related basic intrusive rocks, but instead, a member of the garnet-hydrogarnet solid solution series, usually with a composition near the grossularite end, takes its place. Examples as hydrous as plazolite appear, at least from refractive data, to be rare, although further study may reveal more hydrous types. As the hydrogarnets occurring in these rocks are members of an isomorphous series, it is not considered that new names should be proposed for each mineral species with a slightly different refractive index from the others as this would lead to a needless multiplicity of terms; but instead it is proposed to introduce the single name hydrogrossular for all members of the garnet-hydrogarnet series with a composition intermediate between plazolite and grossularite. It should be noted that hydrogrossular is not synonymous with the term grossularoid introduced by Belyankin and Petrov (1941), apparently to include the two similar minerals plazolite and hibschite only.

The cubic tricalcium aluminate hexahydrate end-member is one of the well-known products of hydration in portland cement but otherwise it is unknown in rocks. It has been held that the only naturally occurring members of this group of compounds are plazolite, hibschite, both 3CaO Al₂O₃ 2SiO₂ 2H₂O, and grossularite, the anhydrous end-member. Work by the author, however, indicates that the hydrous members are more widely developed than has been believed, and compounds ranging from plazolite (or very close to it) have been observed in rodingites and altered gabbroic dykes. It would be interesting to observe in similar rocks just how frequently the determination of grossularite is not strictly an accurate one. In one case Pabst (1936, pp. 9—10) describes undoubted grossularite associated with vesuvianite in veins in serpentinites, for that author gives both an analysis and the refractive index of the garnet.

Murgoci (1900), Marshall (1911), Benson (1914), Grange (1927) and Turner (1933) have all described rodingite dykes and garnetized gabbros intimately associated with basic and ultrabasic intrusions, but concerning the origin of the rodingites and garnet gabbros there has been diversity of opinions. Benson (1914, 1926), Grange (1927), and Turner (1933), however, have shown quite clearly that the “garnet” and other lime silicates such as vesuvianite and prehnite, owe their origin to the action of late magmatic waters on calcic minerals in the basic igneous rocks, and that primary magmatic crystallization cannot explain the features that are observed. No doubt during these secondary changes lime will be derived from diallage or by serpentinization of wehrlites and lherzolites as suggested by Grange (1927). The composition of hydrogrossular is, it is the writer's view, very much in support of the theory that these basic rocks have arisen under the influence of magmatic waters. Flint,

McMurdie and Wells (1941) have shown that any member of the garnet-hydrogarnet series may be prepared synthetically in solution at moderate temperatures, but under considerable pressures, up to 600 atmospheres being employed in some experiments; tricalcic aluminate hexahydrate may even form at room temperatures according to Thorvaldson (1939, p. 258). Therefore it is suggested that the development of hydrogrossular and similar compounds is the result of a rather high concentration of lime in late hydrothermal solutions containing both silica and alumina, or in an environment from which both these oxides may be obtained. They may be derived from the breakdown of anorthite or augite; on the other hand, we may look to the hydrothermal solutions that cause serpentinization, as these are generally considered to be somewhat siliceous (Benson, 1918). No doubt the presence of silica in solution prohibits the crystallization of the basic hydrated calcium aluminates for, as far as the writer is aware, the only hydrated calcium aluminate occurring naturally is hydrocalumite described by Tilley, Megaw and Hey (1934). In addition the hydrated calcium silicates, such as crestmoreite, riversideite, etc., are also very rare and so far have not been recognised in rodingites.

Text Figure 1.
Variation of retractive index with chemical composition in the garnet-hydrogarnet series.

Possibly we find in the development of hydrogrossular, or grossularite for that matter, an interesting parallel with the ability of cement to react with water, and to harden. To explain the latter phenomenon, Brandenberger (in Bussem, 1939, pp. 144–145) advanced the theory that this instability was due to the dual part played by the Ca-ion. It is reasonable to suppose that the instability of the anorthite molecule which was formed at high temperature is due to