Apetl 25, 1902.] 



SCIENCE. 



673 



average was 492, and the median 50. It is 

 interesting to note that the ratio of the medi- 

 an estimates in the two sets is approximately 

 the same. The number of bills actually re- 

 quired is a little less than seven. 



J. Franklin Messenger. 

 Columbia University. 



REGENT PROGRESS IN PETROLOGY. 

 CHEMICAL CLASSIFICATION OF ERUPTIVE ROCKS. 



OsANN, in a recent paper (Tschermak's Min. 

 u. petroc/raphische Miith., Bd. XX., pp. 399- 

 588, 1901), has carried out with reference to 

 the effusive rocks the principles of classifica- 

 tion which in an earlier publication {Ihid., 

 Bd. XIX., pp. 361^70, 1900) he applied to 

 the plutonic rocks. It is his avowed intention 

 to discuss in a third contribution the applica- 

 tion of the same principles to the dike rocks. 



The chemical compositions of the rocks are 

 expressed by the general formula 



s„a,„Crf n^ 

 where 



s = the molecule SiO,. 



a^the atomic group (NaK)jAljO^. 



c = the atomic group CaAl^O^. 



f^the atomic group (FeMnMgSrBaCa)O. 



;i =; the proportion of soda molecules when the 

 relative numbers of soda and potash 

 molecules in the rock are calculated to 

 a sum of 10. 



V = the number of molecules of silica when 

 the ordinary molecular ratios of the 

 rock analysis are calculated to a sum of 

 100. 



w, X and y = the proportions of each of their 

 respective atomic groups, when all three 

 are calculated to a sum of 20. 



2^ the numerical value of the soda propor- 

 tion n. 



In these two papers 207 analyses of plutonic 

 and 403 analyses of effusive rocks are consid- 

 ered and the corresponding rock formulas cal- 

 culated. Prom these formulas the rocks are 

 plotted upon triangular projection paper, the 

 elements of the projection being a, c and f. 



The carrying out of this plan has involved 

 much labor, and if the result is somewhat dis- 

 appointing it has at least the full value of 

 recording a careful and sustained experinaent. 

 It is to be regretted that the author has 

 modestly restricted his attempt at classifica- 

 tion to setting up types within the groups and 

 families of the Rosenbusch classification. It 

 is partly owing to this acceptance of a scheme 

 which has grave objections and which is based 

 on principles little in common with those on 

 trial in this essay, that the latter falls short 

 of more conclusive results. For example, it 

 is seen that the formula of the 'Klausen type' 

 of the diorites is identical with that of a 

 granite intermediate between the 'Syene type' 

 and 'Kamm type,' and similar cases are found 

 among the formulas of the basaltic andesitic 

 and allied rocks of the effusives. 



Inspection of the diagrams fails to show any 

 grouping of the effusive rocks upon which 

 classification might be based. In the plutonic 

 rocks the anorthosites alone show some tend- 

 ency in the graphic projection to form a dis- 

 tinct family. The silica does not appear in 

 the method of plotting here used, and the 

 result is hardly so graphic and satisfactory as 

 that enaployed by Brogger in his 'Ganggefolge 

 des Laurdalits.' 



GNEISSES OF THE SCHWAKZWALD. 



In continuation of his studies of the crystal- 

 line metamorphic rocks of Baden, Rosenbusch 

 (Mitth. der Grossh. Badischen Geologischen 

 Landesanstalt, IV., pp. 367-395, 1901) gives 

 detailed petrographical descriptions and chem- 

 ical analyses of the para-augite and para- 

 amphibole gneisses of the Schwarzwald, the 

 prefix para signifying their derivation from 

 former sediments. The augitic gneisses range 

 from quartzose or psamitic types, to those free 

 from quartz. It is concluded on chemical 

 grounds, supported by geological relationships, 

 that these gneisses have been formed by the 

 metamorphism of calcareous sandstones, dolo- 

 mitic calcareous shales and clayey marls. The 

 hornblendic gneisses were derived from a fer- 

 ruginous dolomitic marl containing quartz and 

 rutile. 



These interesting studies, which recall those 



