294 The National Geographic Magazine 



St Vincent, and through the U. S. 

 "Weather Bureau the U. S. Geological 

 Survey received the material for study. 

 As described in Science * it differs from 

 that of Barbados only in the smaller 

 size of particles and a relatively larger 

 proportion of vesicular glassy to crystal 

 fragments. The greater the distance 

 the more important relatively becomes 

 the vesicular glassy particles which rep- 

 resent the molten material of the erup- 

 tion of La Soufriere, although even at 

 a distance of 275 miles the crystalline 

 matter is still in excess of the glassy. 

 For the purpose of comparison, Mr 

 Steiger's chemical analysis is quoted 

 here with a hypersthene augite andesite 

 of Crater Lake, Oregon, in the follow- 

 ing table. 



In an endeavor, if possible, to get a 

 notion of the destructive gases which 

 wrought such havoc at St Pierre, this 

 dust was treated with a larte amount of 

 water for two hours on a water bath. 

 An analysis of the solution thus ob- 

 tained by Mr Steiger is given in column 

 7 in the table of analysis, indicating that 

 the substances dissolved were CaS0 4 and 

 NaCl. These are common substances 

 in sea water, and it might be argued 

 that they prove that sea water played 

 an important part in the eruption. As 

 the CaS0 4 is much in excess of the 

 NaCl, they cannot be attributed directly 

 to sea water, but more likely to the 

 presence of the acids HC1 and S0 2 in 

 the presence of steam acting upon the 

 lime-soda feldspar and forming the com- 

 pounds recognized in solution. 



The analyses of the insoluble portion 

 given in column 6 of the table shows a 

 considerable percentage of S present, 

 but apparently not in a free state. It 

 may be in the form of pyrrhotite, as 

 suggested by Mr Hillebrand, but we 

 were unable to prove it. Sulphides are 

 rare in fresh volcanic rocks, although 

 they may become common in altered 



* Science, June 13, 1902, p. 947. 



forms. For purposes of comparison an 

 analysis of a hypersthene augite ande- 

 site from Crater Lake, Oregon, is given 

 in column 8. 



In comparing these analyses it must 

 be borne in mind that, while Nos. 1 

 and 4 represent the composition of the 

 molten magma in Mont Pelee and La 

 Soufriere respectively, the sand and 

 dust are made up chiefly of comminuted 

 older rocks, mixed with a portion of the 

 lately erupted magma, and would not 

 give necessarily reliable results as to the 

 composition of the magma. However, 

 it so happens that, the older rocks being 

 of approximately the same chemical 

 composition as that of the latel}' active 

 magma, the composition averages nearl}- 

 the same. 



Dust spread so far and wide from the 

 great eruption of Krakatoa was com- 

 posed almost wholly of material from 

 the molten magma of eruption. It was 

 composed almost wholly of fragments 

 of pumiceous glass, with a few associ- 

 ated cn^stals. Comminuted older lavas 

 formed scarcely an appreciable part of 

 Krakatoan dust, while at Bandai-san, 

 Japan, in 1888, according to Y. Kiku- 

 chi,* "of lava or pumice there is no 

 trace. ' ' All of the material ejected was 

 decomposed and comminuted lavas of 

 earlier eruption. The dust from the 

 late eruption of Martinique and La Sou- 

 friere, collected at a great distance from 

 its source, is nearly midway between 

 these two extremes. 



SUCCESSION OF LAVAS 



Among the volcanoes of the western 

 United States, especially in the Cascade 

 Range, the succession of lava has been 

 determined at a number of places. The 

 earlier eruptions from the large vents 

 are uniformly andesite of one form or 

 another, but generally hyperstnene an- 



* Journal of the College of Science, Imperial 

 University, Japan, vol. iii, part ii, p. 141. 



