SKETCH OF 4GINA AND METHANA. I51r 
type show the molecular proportions of the several oxides and 
will be referred to further on. 
From these analyses, which include all the prominent varieties 
of the region except the Poros andesite, it is seen that we have 
representatives of almost all the members of the andesite and 
dacite families, and that on the whole they are quite normal in 
composition. Leaving the segregations out of account for the 
present it is seen that the range of silica percentage is almost the 
complete one for the two groups, varying quite gradually from 
54.53 (No. 5) to 67.34 (No. 15), or even 68.91 in the dacite of 
Kalamaki.t* | The line between the dacites and the andesites is 
quite sharply drawn, though the dacite of Anzeiou shows an 
abnormally low percentage of silica (61.29). The other dacites, 
though, show quite normal amounts, and the dividing line may be 
drawn at 61 per cent. of SiO, ,? the most acid of the andesites, that 
of Poros, containing 60.21 per cent. (according to Lepsius, loc. cit.). 
The presence of chlorine in all the specimens in which it was 
tested for is of interest and analogous to results obtained elsewhere 
in volcanoes very near the sea.3_ Attention may here be called 
to the much smaller quantity of Cl. in the segregation (No. 14) 
than in its surrounding rock (No. 13). The H,O is in quite large 
amount in several, and is more than can be accounted for by 
perlitic groundmass. All the specimens, however, except No. 2, 
seemed quite fresh. 
While both of the analyses of hornblende-andesite (Nos. 1 
and 2) can be quite closely paralleled by analyses of the same 
species from other localities, yet they show znter se great differ- 
ences in their percentages of silica and alumina. The very low 
silica content of the Stavro rock is the more surprising inasmuch 
as on account of the presence of tridymite, green hornblende and 
biotite and on other grounds, a considerably higher amount of 
tLepsius in PHILIPPSON, Pelop., p. 603. The SiO, determination in the Sousaki 
dacite is not counted here as the specimen was not fresh. 
The analyses of KUcu (op. cit. p. 78) indicate about the same limit, his most 
acid andesite containing 62.26 per cent. SiOz. 
3 Vesuvius, Anals. of Haughton, in Rou. Beitr. z. Petr., 1884, p. lvi. Etna, Anals, 
of Ricciardi, /ézd., p. lxxvi. 
