MINKEALO(iV; AVITiï A CLASSIFICATION OF SILICATES. 43 



and phlogopite, -while towards tlie porsilicaie end they hohl but litth» protox^'d, as seen 

 in indicolite, rubellite, the mnscoA-ites and piuite. 



§ 44. It follows from what has been already set forth that the more or less arbitrary 

 ratios generally assigned by chemists to various silicates, and deduced from empirical for- 

 mulas which in many cases represent ])ut approximately the results of chemical analysis, 

 are not always to be regarded as exact. Thus in examining the various formulas hitherto 

 devised for protosilicates, we find that for the whole succession from chondrodite to 

 apophyllite, the atomic proportions between the bases and the silica may be represented 

 by some twelve simple ratios between 4 : 3 and 1:7. It is probable, in view of the com- 

 plex constitution, involving from twenty to thirty atoms of l)ase, which we have assigned 

 to these polysilicates, that, while some of these ratios are exact, others represent but 

 approximations to the truth. The same remark applies withecjual force to the persilicates, 

 where a like number of similar ratios is made to include all of the known species. In the 

 present state of our knowledge of their composition, however, the usually accepted atomic 

 ratios of oxvds and silica are retained for the formulas and the tal)ular views of protosili- 

 cates and persilicates given further on. 



§ 45. In the study of both of these suborders we note many mineralogical resem- 

 blances between species which differ widely in atomic ratios, and similar resemblances 

 become still more apparent when we examine the larger and more complex group of the 

 protopersilicates. Here, for example, in the families of the zeolites, the feldspars and the 

 scapolites, we find physically similar and homœomorphous species in which, with a con- 

 stant ratio of protoxyds and alumina, the silica is variable. The same thing may be 

 observed among the micas of this suborder ; as when corundophilite is compared with 

 biotite, or margarite with certain muscovites. These latter also present another type of 

 variations in composition, for in different analyses of muscovite, while the ratio of sesqtii- 

 oxyd and silica remains unchanged, that of the protoxyds, represented by alkalies, is 

 variable. Both of these modes appear in the tourmaline group, where species physically 

 similar present at the same time variations in the ratio of protoxyd bases to alumina and 

 in that of ahimina to silica. These divergences in chemical composition, without sensi- 

 l)le changes in external characters, afford strong arguments in favor of the necessity of the 

 aid of chemistry in a true mineralogical classification. 



§ 46. To distinguish in general formulas between protoxyd metals on the one hand, 

 and aluminium and its replacing elements on the other, several devices have been proposed 

 by chemists, of which the most simple is perhaps the employment of Eomau and Italic 

 letters respectively. Accordingly, in these pages m is used for an atom of aluminium, ferri- 

 cum, mauganicum, chromicum or zirconium, while m represents an atom of any other 

 metal, and (the presence of oxygen being understood) the atomic ratios of the protosili- 

 cates are given under the head of m : si, those of the persilicates, m : si, and those of the 

 protopersilicates, m : m : si. 



§ 4*7. Having shown the wide chemical differences existing between the three great 

 divisions of the order Silicate, we proceed to consider those differences, alike chemical and 

 physical, which are found between species often having identical or similar centesimal 

 composition. Physical characters, irrespective of chemical composition, constitute, in the 

 language of Mobs, the characteristic of mineral species, and served as the basis of his sys- 

 tem of classificatiou. We propose to show that, by a re-examination of these characters in the 



