Professor S. J. Shand — A System of Petrography. 465 



and 62 - 5 per cent. It has to be recognized that strict application of 

 these limits is difficult in some cases, as for example where the 

 felspar is soda-orthoclase or raicroperthite, but we ought not to be 

 contented with any lower degree of precision than that which these 

 three subdivisions afford. To be strictly logical we ought to use 

 the Or-Ab ratio only for leucocratic rocks, employing a lime- 

 magnesia-iron ratio in the same way for melanocratic rocks, but this 

 would involve a great departure from existing nomenclature and is 

 better left alone at present. 



A new point is introduced here by Holmes' proposal to calculate 

 felspathoids and micas into their equivalents of orthoclase and albite 

 for the purpose of classification. I am unable to accept this proposal, 

 partly because it takes us away from the actual composition of the 

 rock and partly because in many cases it must be impracticable and 

 its results misleading. That this is not too severe a view the following 

 considerations will, I think, show : — 



(1) Natural leucite always contains a variable quantity of soda. 



(2) Natural nephelite, as recently shown by Bowen, 1 holds from 

 10 to 35 per cent of a plagioclase molecule and from 15 to 30 per 

 cent of kaliophilite. 



(3) Rock-forming micas may hold lithia and soda as well as 

 potash, and alumina may be replaced by iron. 



(4) The proportion of a mica is extremely difficult to estimate 

 accurately, whether geometrical methods o A - heavy liquids are 

 employed. The scales tear away in making sections, and they do 

 not give a clean separation with liquids. 



I cannot see how the first three of these difficulties can be overcome 

 unless every mineral is analysed as well as weighed. 



It is simpler, and it avoids all possibility of misrepresentation, to 

 use the ratio of 



(orthoclase -\- leucite) : (albite + nephelite + sodalite, etc.) 



without any modification, and with the usual limiting values of 

 -f and -f-. The difference between this procedure and that of Holmes 

 is important. Holmes aims at an expression of the ratio of potash to 

 soda ; but for the reasons already advanced his statement will rarely 

 be quite true, and may be sei'iously untrue, unless each mineral is 

 analysed separately. I, on the other hand, am content to state the 

 ratio of the essentially potassic minerals orthoclase and leucite (with 

 whatever dissolved molecules they may hold) to the essentially 

 sodic minerals albite, nepheline, sodalite, etc. (with their dissolved 

 molecules). Holmes uses an unreliable norm : I prefer the plain 

 truthfulness of the mode. 



3. As regards the albite-anorthite ratio which is to be crossed 

 with the above, custom has drawn loose lines of division between 

 albite and oligoclase and between andesine and labradorite. Holmes 

 makes the matter more precise by drawing the lines at 15 and 50 

 per cent of anorthite. Here again, while recognizing the difficulty 

 of adhering strictly to these limits (as, for example, when the 



1 Amer. Journ. Sci., February, 1917. 

 DECADE VI. — VOL. IV. — NO. X. 30 



