eit PR rROGRAPHICAL PROVINCE OF ESSEX CO., 
MASS] “GENERA DISCUSSION: AND CONCLU- 
SIONS] iV. 
Analytical methods.—As far as was possible with the means 
at my disposal the methods advocated by Hillebrand’ were fol- 
lowed, some slight modifications being necessary owing to lack 
of certain facilities in my laboratory. A word must be said in 
recognition of the high character of the work of the chemists 
of the U. S. Geological Survey. Petrologists generally are 
deeply indebted to them for the service they have rendered the 
science, not only by the investigation of methods and the very 
large amount of excellent work which they have done, but also 
for the high standard of excellence which they have set for 
other analysts to follow. 
Ignition H,O was determined in dry CO,. It will be seen 
from the generally low summation of the rocks high in FeO 
that this did not entirely prevent oxidation under the condi- 
tions obtaining, but the error will not be high. The precipita- 
tion with ammonia was always made twice, and three times in 
the case of the basic rocks, in the presence of sufficient NH,Cl. 
This is of the utmost importance, as pointed out by Pirsson? 
and Hillebrand,3 on account of the tendency of MgO to be 
coprecipitated with the Al and Fe hydrates. Neglect of these 
precautions has rendered useless many analyses, but it is a point 
which is often overlooked. With one exception, MnO was not 
determined, since its amount was apparently small and its deter- 
mination would have extended considerably the time necessary 
for an analysis, and hence lessened their number. Alkalis 
were, of course, determined by the Lawrence Smith method. 
*CLARKE and HILLEBRAND. Analyses of Rocks. Bull. U. S. Geol. Surv., 1897. 
2 PIRSSON, JOUR. GEOL., Vol. IV, 688, 1896. 
3 HILLEBRAND, Of. cit, p. 39. 
463 
