18 
ANALYSES OF ROOKS, IT. S. GEOLOGICAL SURVEY. 
[no. 228. 
Now, omitting the minor constituents, and considering only those 
which appear in Washing-ton's calculation, his average and mine may 
be compared as follows: 
Si0 2 . . 
A1 2 3 . 
Fe 2 3 
FeO .. 
MgO-. 
CaO . 
Na 2 . 
K 2 .. 
H 2 0- 
H 2 0+ 
Ti0 2 . , 
PA- 
MnO . 
As given. 
Clarke. Washington 
60.91 
15.28 
2.63 
3.46 
4.13 
4.88 
3. 45 
2.98 
.41 
1.49 
.73 
.26 
.10 
100. 71 
58. 239 
15.796 
3. 334 
3.874 
3. 843 
5. 221 
3. 912 
3. 161 
.363 
1.428 
1.039 
. 373 
.219 
100.802 
Reduced to 100 per cent. 
Clarke. Washington. 
60.48 
15.17 
2. 61 
3.44 
4 10 
4.84 
3.43 
2. 96 
.41 
1.48 
.72 
.26 
.10 
100. 00 
57. 78 
15.67 
3. 31 
3.84 
3.81 
5.18 
3.88 
3. 13 
.36 
1.42 
1.03 
.37 
.22 
100. 00 
One reason for the difference between my estimate and that of 
Washington is not far to seek. The two columns of reduced averages 
are not strictly comparable, for one contains many partial analyses, 
and the other relates only to those which are nominally complete. For 
the interpretation of a femic a rock, which is mineralogically complex, 
a full analysis is necessary. For the simpler salic/' rocks, determina- 
tions of silica, lime, and alkalies are usually sufficient for petrographic 
purposes. The partial analyses, therefore, represent mainly salic 
varieties, and their inclusion in the average tends to raise the percent- 
age of silica and to lower that of magnesia or iron. But the salic 
rocks are undoubtedly more abundant than those of the other class, 
and so I am inclined to regard the higher figure for silica as more 
probable than the lower. If, however, we include all the minor con- 
stituents of rocks as given in my computation, the final result assumes 
the form given on the following page: 
a The terms "salic" and "femic," proposed by Cross, Iddings, Pirsson, and Washington, are here 
used instead of the conventional "acid" and "basic." The latter terms have precise chemical mean- 
ing, and their use in another sense by petrographers is inexact and misleading. 
