Gabbroid Rocks of Minnesota. — Wmdiell. 375 
IV. Plagiochsyte (336E), from the summit of Carlton peak. 
Traces of PjO.? and F (or CI), COj and S occur in rare minerals of 
the rock; no BaO, nor SrO; no other elements determined. 
V. Troctolyte (514), from near Duluth; traces of CO-, occur in 
calcite (very rare); no appreciable BaO, nor SrO; no other elements 
determined. 
VI. Orthoclase gabbro (1797), from Duluth; F (or CI), and CO: 
are present in traces in apatite and calcite; no appreciable BaO, nor 
SrO; no other elements determined. 
VII. Cordierite noryte (983). from Sec. 15, 63-9, not far from 
Snowbank lake; traces of ZrO^, P2O.-,, F (or CI), S, and C (?) indi- 
cated mineralogically; no appreciable BaO nor SrO; no other ele- 
ments determined. 
VIII. Quartz gabbro (854G). from S. E. V^ S. W. ^4 Sec. 12, 64-6, 
southwest shore of small lake, near Little Saganaga lake; F (or CI) 
and ZrO: occur in rare minerals; the relative abundance of calcite in- 
dicates that CO; is in more than mere traces, and many attain i.po per 
cent; no appreciable BaO, nor SrO; no other elements determined. 
IX. Silicoferrolyte (960), from a low ridge of fifteen rods from 
the shore, S. W. ^ Sec. 24, 64-12, near Birch lake; no BaO, nor SrO. 
nor LiiO; no other elements determined. 
Of late years iniich attention has been directed to the 
chemical composition of igneous rocks, and conclusions rela- 
tive to their origin have been derived from this study. At- 
tempts have been made to coordinate magmas, to establish 
natural classifications, and to determine whether or not igne- 
ous rocks are derived from a single primitive magma. The 
writer has accepted the theory proposed by Prof. N. H. 
Winchell of the origin of all these rocks by a refusion of pre- 
existent rocks. Therefore the variations in the composition 
of these rocks are due (at least in part) to the variations in the 
rocks from which they have been derived. But disregarding 
this hypothesis for the moment let us see what would be the 
relations of these rocks considered as ordinary igneous rocks, 
and what position they would take in the classifications here- 
tofore proposed, etc. 
The relative number of molecules and atoms of each oxide 
and metal are given in the following tables, as calculated from 
the analyses given abqve. Each analysis is first calculated to 
the scale of one hundred, after the elimination of such ele- 
ments as H-O, CO^, and P=0=. The relative number of mole- 
cules of each oxide (called Z by Rosenbusch) is then obtained 
by dividing the per cent of the oxide by the weight of its mole- 
cule (see table II). From these values are readily obtained 
