THE ELECTROMAGNET -IN- PETROGRAPHY a1 
4. Lievrite, hedenbergite, limonite. 
5. Augite (15 to 20 per cent. of iron oxides), pleonaste, arfvedsonite. 
6. Hornblende, augite (poor in iron), epidote, pyrope. 
7. Tourmaline, bronzite, idocrase. 
8. Staurolite, actinolite. 
g. Olivine, pyrite, vivianite. 
10. Biotite, chlorite, rutile. 
11. Haiiyne, diopside, muscovite. 
12. Nepheline, leucite, dolomite. 
It has been already mentioned that, owing to variations in 
composition exhibited by many minerals (especially in point 
of iron percentage), such lists ought not to be regarded as 
being accurate. 
Perhaps the most noteworthy feature of Doelter’s work with 
the electromagnet, was that of his experiments on crushed 
rocks. He attempted the separation of the mineral constituents 
as far as possible by the electromagnet alone, and stated the 
results quantitatively. But the fact that he operated on the 
dry crushings detracts seriously from the merit of the results 
he obtained. The character and extent of his separations may 
be seen from the following instances compiled from his results. 
Nepheline Syenite, Antao Island 
Magnetite : ; : : ‘ 5 per cent. 
Augite and hornblende . : : : : » 30 . 3 
Nepheline and felspar . : : ; ; . 65 = 
Phonolite of Praya 
Magnetite ; ; : : ; : : ; 4 per cent. 
Augite : 4 J : ; : : POREEY Ves 
Felspar and nepheline . ; ‘ ‘ ‘ a1 85 50 
Compact Basalt, Santiago Island 
Magnetite : : : ; : - ; . 1 GU per cent, 
Augite ; i - : F : : 7 3946 ” 
Olivine. : ; ; a : : a = [GTO ns 
Plagioclase F : 3 : ‘ ; a Geena Ci 
Of the above results, we may take that for the nepheline 
syenite as possibly accurate. In the cases of the phonolite 
and basalt, however, we may reasonably inquire what has 
become of the undifferentiated ground of the rock. Indeed, 
it is fairly obvious that accurate results cannot be got by 
physical methods when dealing with hemicrystalline or very 
