Page images

It is not surprising, therefore, that with this method of treating the problem it has proved difficult to identify 3MgO.Fe,0,.3Si02. I have, however, found one occurrence, namely in an analysis of a deep brownish red garnet from a garnet-diopside nodule from the Jagersfontein diamond mine in South Africa, by the late Percy A. Wagner.' Dr. Wagner's analysis is as follows:

[merged small][merged small][merged small][ocr errors][merged small][merged small][merged small][merged small][merged small][merged small][ocr errors][merged small][merged small][merged small][merged small][merged small][merged small][merged small][merged small][ocr errors][merged small][merged small][merged small][merged small][merged small][merged small][merged small][merged small][merged small][merged small][merged small]

The closeness of this ratio 3.06 :1:2.91 to the standard ratio for garnet, namely 3:1:3, is a testimony to the purity of the garnet and the accuracy of the analysis. There is no reason, therefore, for not accepting the result of the interpretation of this analysis in terms of garnet ‘molecules ', which is as follows :

[ocr errors][ocr errors][merged small]

Per cent.

50-68 19.06 13.14 10.36 3.36 1.92 1.17



per garnet.

[ocr errors]

3:09 per cent.




1. Die Diamantführenden Gesteine Südafrikas ', Berlin, p. 402, (1909).

We are compelled, therefore, to accept as proved the existence of the garnet molecule’ 3MgO.Fe,09.3Si0, already deduced as present in the garnets from which chondrules have on my hypothesis been derived.1 This new garnet needs a name.

I should have liked to name it after the analyst, who did such brilliant work in South African geology before his early demise. This course is, however, prevented by another wagnerite, a monoclinic magnesium fluophosphate named as long ago as 1821 after an earlier, Oberbergrath, Wagner (see Dana). It is suitable that instead we should call this new garnet khoharite, after the meteorite for which it was first proposed.


a. Names of garnet' molecules '. As given in the usual text books of mineralogy there are six known garnet molecules conforming to the general formula 3R"0. R",03.3Si0, In four of these garnets R,0, is alumina, in one of them it is ferric oxide, and in the sixth it is chromic oxide. There seems to be no reason based on the atomic dimensions of Al", Fe", Cr" and Mn" why there should not also be four ferric, four chromic, and four manganese garnets. In the past I have shown the existence of two additional ferric garnets (skiagite and calderite) and one manganic garnet (blythite); with the identification of khobarite it seems desirable to take stock of the position. The garnet 'molecules' now known , arranging them in order of increasing ionic radii of divalent and trivalent ions are : Alumina garnets--3MgO.Al,0.3Si0,

Pyrope. 3Fe0.A1,0,.28i0,

Almandite. 3MnO.Al,0, 3Si0,

Spessartite. 3Ca0.A1,0,.3Si0,

Grossularite. Chromic garnets-3CaO.Cr,0.3Si0,

Uvarovite. Ferric garnets3MgO.Fe,0.3Si0,

Khoharite. 3Fe0.Fe 03.3Si0,

Skiagite. 3MnO.Fe,0.3Si0,

Calderite. 3Ca0.Fe 03.3Si0,

Andradite. Manganic yarnets 3Mno. Mn, 09.3Si0,

Blythite. 1 The alternative is to assume an error of 2.72 per cent. surplus in the deter. mination of magnesia and the presence of 13:46 per cent. of impurities or included minerals instead of 3.09 per cent.

2. On the Composition of some Indian Garnets ', Rec. Geol. Surv. Ind., LIX, pp. 191-207, (1926): p. 202 for skiagite and p. 204 for calderite and blythite.




It will be seen from the foregoing statement that there are three manganic and three chromic garnet ‘molecules' still undiscovered, but all possible, judging from the atomic structures of garnets. Whether the facts of Nature provide for their occurrence anywhere is another matter.

b. Names for compound garnets. In speaking of garnet 'molecules' and of R20, groups I have, of course, been using the old nomenclature, and have been referring to the composition of garnets in terms of the results of their chemical analysis. X-ray studies of garnets show, however, that they do not contain the R,0, group as such, and, moreover, that the complete unit cell of garnet is eight times the traditional formula or ' molecule ', so that strictly speaking the formula of garnet should be written as

8[R",R"2(SiO2)3], the unit cell of garnet containing 24 atoms of divalent metals, 16 atoms of trivalent metals, 24 atoms of silicon, and 96 atoms of oxygen, 160 atoms in all. The silicon and oxygen

oxygen are combined in groups of SiO4, because X-ray analysis shows that each silicon atom is the centre of a tetrahedron of oxygen atoms, the other elements being linked to these oxygen atoms.

As no pure garnet has ever been found, it must be very rarely, if ever, in Nature that a unit cell of garnet contains only one divalent or only one trivalent element. A garnet that on analysis is found to contain say 50 per cent. of pyrope and 50 per cent. of almandite must not be pictured as a mixture of unit cells, some with only Mg and some with only Fe" as the divalent metal; instead, it must be pictured as composed of unit cells each of which contains (on the average) 12 Mg atoms and 12 Fe" atoms. Consequently, when it is desired to express by one formula the composition of such a garnet it is desirable to use some such form as

Mg12Fe" 12A116(SiO2)24) the ratios of the divalent elements being referred to 24 and of the trivalent elements to 16.

In my Memoir on the Manganese-Ore Deposits of India published in 19091 I proposed the term spandite (as an abbreviation

1 Mem. Geol. Surv. Ind., XXXVII, pp. 163-165.



of spessart-andradite) for a

for a garnet intermediate in composition between spessartite and andradite ; and grandite for a garnet intermediate between grossularite and andradite. Later, in the 1926 paper to which reference has been made above, I used the terms pyralmandite and spalmandite for garnets intermediate between pyrope and almandite and spessartite and almandite respectively. In following this course it seems that I was adopting one that the facts of crystal structure have since proved to be justified. Thus spandite, instead of being a mixture of spessartite and andradite molecules in solid solution, must be a garnet in which the unit cell contains both Mn and Ca as divalent atoms and Al and Fe"! trivalent atoms : and so on for the other garnets named.

The above course was advocated only when each of the two garnet ‘molecules' included in the name was in large proportion. To indicate the presence of constituents in less important proportion I suggested the of chemical prefixes as in mangangrandite, mangan-almandite, ferro-spessartite, calc-spessartite and magnesia-blythite? We see now that such term as ferro-spessartite means a spessartite with some of the Mn" (or Al"') atoms replaced by Fe” (or Fe"') in the unit cell, and that such a term must correspond to a real entity. Similarly mangan-grandite is a garnet with a proportion of the divalent Ca atoms in one unit cell replaced by Mn atoms and with both Al and Fe" occupying the spaces for trivalent elements, .

On page 205 of the 1936 paper cited above I give the formulæ of five analysed garnets in terms of percentages of constituent garnet molecules. Thus spandite is given as

Py,SP23An 47Sky Ca17?



the five symbols standing for pyrope, spessartite, andradite, skiagite, and calderite respectively. The information this was intended to convey was that this garnet consisted of five garnet 'molecules isomorphously mixed in the proportions shown. The revised method of looking at this garnet is that it consists of unit cells each with the following approximate composition :

CaMn 10 Mg,Fe", Fe" 11A1, (SiO2)24,

1 Mem. Geol. Surv. Ind., XXXVII, p. 164, (1909). 2 Rec. Geol. Surv. Ind., LIX, pp. 202-206, (1926).


and this is the approximate composition of the spandite analysed. The extent to which the percentage proportions of the garnet

molecules' cannot be adjusted to integral fractions of 24 for the divalent atoms and of 16 for the trivalent atoms must be regarded, in so far as it is not due to experimental error, as a

measure of slight variation from the composition of one unit cell of garnet to another in the same specimen.

Although the second method of representing the analysis of spandite given above shows a closer relationship to the intimate facts of Nature than the first, yet this first method will no doubt still often be used in order to show the statistical proportions of the imaginary constituent molecules', which are in all probability rarely present !

My method of conjoining portions of the names of garnets to represent a compound garnet has been adopted by A. N. Winchelli. Making use of the work of Boeke, who showed that, based

a study of analysed garnets, there appear to be limits to the mutual solubilities of the six garnets commonly recognised, so that they can be grouped into two sets or types, each mutually soluble, Winchell applies the term pyralspite to the set pyrope, almandite and spessartite, and ugrandite to the set uvarovite, grossularite and andradite. Within each set there is supposed to be complete mutual solubility. The data do indicate that this is the case for spessartite and almandite and probably for pyrope and almandite; but there are no analytical data indicating the complete mutual solubility of spessartite and pyrope, and Winchell's pyralspite appears to be a statistical amalgamation of my pyralmandite and spalmandite. His ugrandite is composed mainly of grossularite and andradite, as is recognised in the compositional diagram reproduced from Boeke where the term grandite' is wrongly used3

Whether this grouping of garnet into two sets under the terms pyralspite and ugrandite is useful depends really upon whether we need names to described groups or series of garnets instead of actual garnets, and whether it is desirable to take any

Elements of Optical Mineralogy', 2nd Edit., Part II, pp. 257-265, (1927). 2 ' Die Granatgruppe ', Zeit. Kryst., LIII, pp. 149-157, (1913).

3. Grandite' was proposed by me to describe garnets containing important proportions of grossularite and andradite, and is not therefore, available for use as a name for either grossularite or andradite alone. There seems, however, to be no objection to referring to the three garnets grossularite, grandite, and andradite, as the grandite series, any more than there would be in referring to spessarite, spandite, and andradite, as the spandite series.

« PreviousContinue »