14$ Washington — Submarine Eruptions of 1831 and 1891. 



Apart from the higher silica and alumina of the earlier one, 

 the basaltic lavas of the two submarine eruptions are closely 

 similar, I being slightly higher in MgO and Na 3 and lower 

 in FeO, CaO, and P 2 5 . The variation in TiO a is also consider- 

 able, but in both rocks the amount of this constituent is higher 

 than usual among basaltic rocks. The distinctly more femic 

 character of the later eruption is in harmony with the decidedly 

 more siliceous and salic character of the material of the early 

 phase of the Graham Island eruption, as shown by Abich's 

 analysis to which attention was called above, and there is thus 

 indicated a progressive change in composition with increase 

 in femic constituents for the general magma of the region. A 

 change in the same general sense has already been noted as 

 having occurred on Pantelleria, and a similar relation is observed 

 on Sardinia, where the large volcanoes of Monti Ferru and Arci 

 poured forth respectively phonolitic trachytes and rhyolites in 

 the early stages of activity, followed by extensive outflows of 

 basalt. On this island too all the small cones of the most recent 

 date are basaltic, though varying somewhat in composition, 

 and in general decidedly more salic than are those of Pantelleria, 

 Linosa, and the two submarine eruptions. - The basalts of 

 Catalonia also show great analogies with those of Pantelleria 

 and the neighboring volcanic vents, as has been elsewhere 

 pointed out. 



Aside from the very high Ti0 2 of all these basalts, one of 

 their most striking characters is the predominance of ferrous 



over ferric oxide. This ratio _ ~ reaches a maximum in the 



. Fe A 

 basalts of the two submarine eruptions and in that of the 



dike on Pantelleria, which are highly vesicular, and shows 



a minimum in the also very vesicular lava of La G-arrinada, 



near Olot. Lavas showing ratios between these extremes 



are either only slightly vesicular or are solid flows. 



In a former paper* the difference observed at the Catalan 

 volcanoes was ascribed to the more favorable conditions for 

 oxidation of the ferrous iron when the lava issued in a highly 

 vesicular condition. Since, however, we find that oxidation is 

 at a minimum in the highly vesicular basalts described in the 

 present paper, some discussion of the matter will be pertinent. 



The ready oxidizability of the ferrous iron in rocks on heat- 

 ing in air, even when in the solid though powdered form, is 

 well known to all who have had to make rock analyses, and 

 the researches of Gautier f have shown that steam alone' also 

 exerts a powerful oxidizing action on ferrous silicates. On the 

 other hand, further researches of Gautier and of Hiittner showed 



* H. S. Washington, this Journal, vol. xxiv, p. 240, 1907. 



fCf. F. W. Clarke, Bull. U. S. Geol. Surv., No. 330, 1908, p. 228. 



