140 IDDINGS. 
In that from Pantelleria it is apparent that the alumina is 
much less than the alkalies, and that a large part of the 
alkali molecules must enter into a non-aluminous mineral 
molecule. 
The chemical characteristics of these groups of rocks are 
also expressed by the behavior of the other essential oxides. 
The reciprocal nature of the two oxides of iron is noticeable 
in all of the groups and has been discussed in another con¬ 
nection.* In the third diagram of each series the iron is 
represented as ferrous oxide. 
In the group from Electric peak and Sepulchre mountain 
the similarity in the variations of the lime, magnesia, and 
iron oxide is very marked, there being a great decrease of 
the molecules of these elements from the less silicious to the 
more silicious end of the series. The magnesia is slightly 
more variable than the other two, and the ferrous oxide is 
less abundant than the lime, though this relation is varied 
in the rhyolitic lavas at the end of the diagram. In general, 
these oxides are in nearly equal proportions as indicated by 
their close grouping in the diagram. In the Crandall group 
the relation between these oxides is like that just mentioned, 
the iron is slightly less than the lime, and the magnesia is 
still more variable, especially at the basic end of the series, 
and this is even more pronounced in the Absaroka group. 
In the first group the average of all of these oxides is higher 
than in the second and third groups for rocks of like per¬ 
centage of silica. The generally uniform variation and 
diminution of these molecules, together with the highly 
variable behavior of the magnesia, are characteristic of these 
groups of rocks. 
In the group from Vesuvius and vicinity the variations 
are greater and the relative proportions are not so nearly the 
same. This is sho'wn by the greater spread of the lines. The 
lime is the most variable and is considerably greater than 
* Papers on the rocks of Electric peak and Sepulchre mountain already 
cited. Twelfth Annual Report U. S. Geol. Survey, p. 630. Bull. Phil. Soc. 
Washington, vol. 11, p. 212. 
