Br, Shand — Saturated & Unsaturated Igneous Rocks. 513 



like Dr. Hatch ^ and Mr. Mennell,* who tie their faith to silica 

 percentages, would instead employ the mineralogical dividing lines 

 which separate the oversaturated from the saturated rocks, and these 

 again from those which are undersaturated (1) with regard to leuco- 

 cratic constituents, (2) with regard to melanocratic constituents, and 

 from (3) those which are undersaturated or unsaturated in all their 

 constituents, they would find themselves in possession of a much 

 more ' natural ' classification of rocks, and one which would be vastly 

 simpler to use than any classification which is based upon silica per- 

 centages. The connexion between mineral and chemical composition 

 need not be obscured thereby, but rather the reverse, especially if, as 

 could easily be done, the relative degrees of oversaturation or under- 

 saturation were introduced as subordinate factors in the classification. 

 The fact that, under difi'erent conditions of temperature and pressure, 

 one and the same magma may give rise to rocks of different degrees 

 of saturation, is an argument in favour of my contention, not against 

 it. The classical experiment of Fouque and Michel-Levy, in which 

 a melt of orthoclase and biotite yielded leucite and olivine on cooling, 

 illustrates this point. The silica percentage, being the same in the 

 product as in the educt, fails to express the difference between these 

 two very different mineral associations ; yet the former is saturated, 

 the latter unsaturated, and by making degree of saturation a criterion 

 of systematic position we effect a separation of things formed under 

 unlike conditions, which is at least as important as the separation of 

 things of unlike ultimate composition. 



If Dr. Hatch and Mr. Mennell had considered the natural criterion 

 of undersaturation, in place of the artificial one of silica percentage, 

 they would not have been led into the paradoxical positions of 

 classifying borolanite, a rock which typically contains neither plagio- 

 clase nor augite, with " alkali-gabbro " and " dolerite " respectively. 

 In my opinion both these gentlemen have erred by committing 

 themselves to classifications which are neither definitely chemical 

 nor definitely mineralogical. A chemical classification, to be of any 

 value, must have regard to all the molecules present in a rock, not 

 to one alone. If a classification is to be mineralogical, let it be 

 consistently so. I believe that both chemical and mineralogical 

 classifications are necessary, but of hybrid classifications I would go 

 so far as to say, as Harker says of hybrid rocks, "like other hybrids, 

 they are barren." 



The recognition of the essential difference between saturated and 

 undersaturated rocks ought to lead to the avoidance of such loosely- 

 used terras as shonkinite, monzonite, and essexite. We read in 

 Dr. Hatch's textbook, for example, that " nepheline and leucite occur 

 occasionally in monzonites " ; that essexites contain a variable quantity 

 of nepheline; and that nepheline and sodalite "may be present in 

 small quantities " in shonkinite. The older petrographers always 

 drew a sharp distinction between the syenites and the nepheline- 

 syenites ; however small the proportion of nepheline, it was considered 

 sufficient to justify the removal of the rock from the former to the 



^ Textbook of Petrology, 1909. ^ Mamial of Petrology, 1913. 



DECADE V. — VOL. X. — NO. XI. 33 



