I'KNTTI ESKOLA. M.-N. Kl. 



In Ii.t4. I llic molecular percentages of ilie liiiie-er)ni|)oiinds in all these 

 garnets ha\e been ])l()tte(i on the r)rdinates, the abscissas representing the 

 ratio I'Y'O : (MgO + CaO), as for the variation curve. ! 1ère the average 

 percentages of the CaO-compound ami their \ariatir)n uitli the KeO : (MgO 

 + CaO) ratios for the various rocks are shown by curves. 



The pyropes in peridotites, serpentines, etc. have a very constant 

 amount of lime, ^lliis is only due to a limited variation in the original 

 material. 



In the eclogite-garnets the amount of lime is more widely variable 

 than in any other, owing to variable original material and specific physical 

 conditions during their genesis. Therefore very little can be said with 

 certainty about an eclogite-garnet on the basis of the refractive index. In 

 the Norwegian eclogites, howexer, the variation is no greater than in the 

 two following groups'. 



The garnets of all other I'ocks contain a prevailing amount of iron. 

 The average amount of lime is largest in the garnets of the amphibolites 

 and much smaller in those of the gabbros, apparently varying with the 

 conditions obtaining during their formation. The still smaller quantities of 

 lime found in gaiMiets from die cordierite-bearing rocks and the granites 

 may be ascribed to the 'composition of material a\'ailable. 



The group of gneiss and mica-schist again shows a greater average 

 and maximum discrepancy from the a\'erage amount of lime than the other 

 groups, excepting the eclogite. In this case this may simply be ascribed 

 to the fact that this group is less homogeneous, comprising sedimentogenous 

 and eruptivogenous schists and gneisses of various composition and faciès 

 development. 



Pracdcally, the diagrams in tig. i ma}' be used in the following- 

 way. After having determined the index of refraction of a garnet either by 

 the prism method or by the immersion method with liquids or melts, one 

 finds the point on the curve corresponding to the (F'e,MnlO : (MgO + CaO) 

 ratio and, going downwards along the ordinates and selecting the proper 

 diagram, the probable molecular percentage of CaO. The possible error 

 in such determinations may be estimated on the basis of the above table 

 and the diagrams. It must be noted that, in exceptional cases, the error 

 may be greater than the maximum discrepancies named in the table. 



All the other kinds of garnets but almandite-pyrope must, of course, 

 be excluded. Spessartite garnets are always associated with other nian- 

 ganiferous minerals and lime-garnets with rocks rich in lime. In dubious 

 cases chemical tests must be consulted. I preferably use the determination 



' I have therefore, in my studies, used the average lime-percerttage of 6 analyses of 

 Norwegian eclogite-garnet, being i8 mol. ^/o, the average discrepancy being 2.5 and 

 tlie maximum discrepancy 7.0. 



