io THE ROMAN COMAGMATIC REGION. 
decision gradually easier, so that in the final general survey of all the specimens and 
their thin sections there were very few cases where any change in the assigned posi- 
tion seemed to be justified, and few attributions which were seriously in doubt. 
CHEMICAL ANALYSES. 
As a knowledge of the chemical composition of igneous rocks is fundamental 
to their classification according to the system adopted here, especial attention was 
paid to their chemical analysis. In the selection of appropriate specimens for this 
the endeavor was made to have them, as far as possible, representative of the various 
kinds of magma found in the region those which are common as well as those which 
are of rare occurrence. Furthermore, it was attempted to have the analyses repre- 
sentative of as many as possible of the different magmas found in each district, so as 
to furnish a clue to the magmatic variations in each ; and again to have analyses 
of some, at least, of the diverse modes and textures which each magma might assume 
on solidification. These various aims could not all be completely attained, some 
magmas being represented by many analyses, while of others equally as abundant 
but few were made. At the same time, the analyses here presented are so numer- 
ous, and cover such a variety of rocks of widely diverse chemical, modal, and tex- 
tural characters, that they may be deemed an adequate basis for a discussion involv- 
ing the chemistry of the rocks. 
There exist in the literature numerous analyses of the rocks of the Roman 
Region, especially those of the Vulsinian and Campanian districts. Of these older 
analyses only a few have been found, on careful consideration, to be adequate for 
modern use either as regards accuracy or completeness. A large series of analyses 
of Vulsinian rocks are not only incomplete in the determination of minor constitu- 
ents, but, as will be shown later, are so hopelessly inaccurate, especially in the figures 
for the alkalies, that they may all be rejected without hesitation. The older anal- 
yses of the Vesuvian lavas, while in part much more accurate in the determination 
of the main constituents, are faulty on the score of completeness, in the absence of 
determinations of the minor ones, as titanium and phosphorus. In these rocks, as 
is seen elsewhere, these usually negligible constituents assume a very considerable 
importance in classification, so that we can not avail ourselves here of this otherwise 
adequate and useful older work, except in a general way. 
Most of the analyses made and published several years ago by myself have 
been incorporated, as new analyses and redeterminations have shown them to be 
reliable. In some cases they have been corrected by redeterminations of some 
of the constitutents, as well as rendered more complete by the determination of minor 
constituents in which they were lacking, especially titanium and phosphorus. Un- 
less otherwise stated, this has been carried out with material remaining from the 
original analysis. But the majority of the analyses presented in this paper have 
been made by me especially for the present investigation. In making them, as well 
as in the determinations which were needed to bring the older analyses up to modern 
standards of completeness, the methods employed were those advocated by Hille- 
