Components ol an Igneous Rock — Williams 45. 
suits obtained by Thoulet's solution and by calculation from 
fhe chemical composition agree in a general way and their 
being the only ones based primarily on specific gravity makes 
their accordance a strongly significant fact in considering 
which of the methods most nearly approximates the truth. 
The chemical analysis of a comparatively simple rock whose 
minerals components are of definite molecular proportions 
cannot but give a close indication of the actual or 'modal' 
composition. It is of course more significant if, as with 
the rock in hand, certain of the elements determined in the 
chemical analysis are limited to a single mineral. For ex- 
ample, potash occurs in orthoclase only and since the ortho- 
close molecule is a fixed one, this oxide, K^O., determines 
beyond question the percentage of orthoclase in the rock. 
When one or more of the chemical constituents are to be 
apportioned among two or more minerals, and especially 
when these minerals are not of fixed composition, the accu- 
racy of the results becomes less. 
The Westerly granite represents a fairly simple type of 
composition and its calculated mineral analysis may be 
taken therefore as being correspondingly accurate. The 
general agreement of the heavy solution analysis with the 
calculated composition certainly suggests an approach to the 
truth, despite the unavoidable irregularities attending the use 
of fhe former. 
Of the microscopic, it is difficult to say which method 
will ordinarily furnish results of the greatest value. The use 
of photography eliminates to a degree the personal factor 
and may ])erhaps often be employed advantageously. It is 
needless to call attention further to the inconsistency of cal- 
culating percentage composition by using d^ when d is the 
measured width of the mineral grains as they appear in 
thin section. 
The closest agreement of any two members of the table 
is noted between the direct measurement of d and of area. 
Were the second calculated from the first by squaring the 
values found for d and reducing to 100 per cent, the differ- 
ence would be far greater, which may be taken as indicating 
that the d mej^sured cannot be safely assumed as the first 
dimension of a cube ; and therefore, which is evident, that 
