154 THE JOURNAL OF GEOLOGY. 
in reality segregations and not enclosures. In accordance with 
the theory of the differentiation of magmas the large masses of 
basic andesites as well as the smaller segregation masses are the 
products of such a differentiation process, so that the later ejected 
acid products would carry up with them parts of the later more 
basic differentiation products (the segregations) still remaining 
behind, while the main andesitic masses were ejected before these 
last were formed. That such is the true explanation is indicated 
by the fact that the segregations are, relatively to the neighbor- 
ing basic rocks, richer in lime and magnesia, poorer in alkalies, 
and containing at the same time either exactly the same percent- 
age of silica or even less (cf. Anals. 4 with 11, 6 with 8 and 
\ 
14). 
Again, if the masses in question were merely fragments of 
earlier now buried lava outflows, it is difficult to understand why 
the later basic streams of Oros, and the rest did not carry up 
fragments when such contiguous outflows as those of Kako- 
perato or Mt. Gaiapha did do so. 
As to the specific gravities it will be seen that the more 
basic pyroxene-andesites are considerably denser than the horn- 
blende-andesites or the dacites, and that the enclosures have 
likewise a decidedly higher specific gravity than their enclosing 
rocks. 
Order of eruptions—It has already been mentioned that, 
though the evidence at present available is scanty, the eruptive 
center of Atgina shifted as time went on from the south to the 
north. -If, as seems probable, this be true, we find that the erup- 
tions began with pyroxene-andesites and ended with more highly 
acid hornblende-andesites, the order of eruptions being, on the 
whole, one of increasing acidity. The establishment of this rela- 
tion is of some importance, and we shall see whether the other 
occurrences of the region bear out the above conclusion or not. 
The most striking and conclusive case is met with in the 
Oros district, where the certainly later flank eruption of Anzeiou 
and Kakoperato are much more acid than the earlier main out- 
flows forming Mt. Oros. A similiar relation also obtains in the 
