AN ESTIMATE OF THE GEOLOGICAL AGE OP THE EARTH. 267 



This WO may compare with the results of averaging the rock analyses 

 selected b}" H, Rosenbusch in his ''Eleuiente der Gesteinslehre.''' 



These may be tabulated as follows (the references are to the pages 

 of the work referred to) : 



KoO. 



Na,0. 



Eruptive rucks: 



Mean of 19 Lipa rites, Traohytcs, Basalts, etc. (p. 35) . 

 Plutonic rocks: 

 Mean of — 



18 Granites (p. 78) 



46 Syenites (pp. 106-126) 



25 Diorites (pp. 140, 141) 



6 Essexites (p. 172) 



8 Theralites and Shonkinites (p. 176) 



14 Gabbros (p. 151 ) 



Moan of eruptive and plutonic rocks 



Al.so thi' following dynamically altered eruptive rocks: 

 Mean of — 



5 Porphyroids and Serecite gneisses (p. 440) 



12 Mica gneisses (p. 468) 



4 Amphibole gneisses (p. 484) 



6 Pyroxene and Augite gnel.sses (p. 4S6) 



13 Hiilleflinte gnei.sse.s (p. 493) 



10 Amphibolites (p. 515) 



6 Eklogites (p. 520) 



Mean of metamorphic eruptive rocks 



Mean of all 



4.26 

 .5.60 

 l.SG 

 2.49 

 3. 86 

 0.77 



3.80 

 6.80 

 3.82 

 4.68 

 4.98 

 2.62 



With reference to the last division of rocks — the metamorphic 

 eruptives — Rosenbusch admits the old standing- difficulty of distin- 

 guishing between the altered sediments and the altered eruptives. 

 Thus gneisses derivable from sediments give just such chemical 

 proportions as appear in those referred to eruptive origin. There 

 exists no sure criterion for classing a gneiss according to its origin 

 (p. 472, loc. cit.). 



This is a well-known difficulty, and has been the subject of much 

 research and speculation. We dare not do more than suggest here 

 that in doubtful cases the general law of the alkali ratios of eruptives 

 and sedimentaries, where this admits of application, should carry 

 weight with petrologists. 



The uncertainty referred to may, of course, affect the estimated 

 composition of the crust rock and of the siliceous sedimentaries. 

 Examination will, however, show that the uncertainty being confined 

 to a couple of groups only, can probabl}' afi'ect the final averages but 

 little in either case. 



1 Stuttgart, 1898. 



