THE KATMAI REGION, ALASKA 583 



but descend like a hot avalanche So much solid matter was suspended 



in the cloud, that it became too dense to surmount obstacles and behaved 

 rather like a liquid. 



Though one may find in the detailed descriptions of these sand- 

 avalanches certain differences from the results seen in the area of 

 the valley flow, they seem to be of degree rather than of kind, and 

 the analogies are striking. 



Some of the pieces of pumice in this deposit show a banded or 

 variegated structure, such as is illustrated in Figure 11. The 

 difference of composition of adjacent bands is easily apparent, and 

 in the specimen figured, determinations of silica by Dr. Allen have 

 shown 74 . 70 per cent in a white band and 60 . 40 per cent in a 

 dark band. It is believed that these structures are due to a 

 process of partial solution of basic rock in the new siliceous magma. 

 The very limited degree of mixing of solutions shown by these 

 specimens hardly permits us to suppose that the solvent action was 

 long continued; therefore we must look for the source in matter 

 which became involved in the magma when it was near or at the 

 surface and just prior to its foaming-up into pumice. There are 

 several possibilities that should be considered. We might suppose 

 that the sedimentary series beneath the valley had been previously 

 injected by rocks of this description and that these were encountered 

 by the new magma and material absorbed from them; or that the 

 floor of the valley was composed of an old lava flow of basic com- 

 position, which contributed material; but the supposition that, for 

 a number of reasons, appears to me the most probable is that the 

 source to which we should look is the deposit of lava bowlders of 

 glacial origin that covered the floor of the valley to a great thick- 

 ness. The fragments of undissolved andesite found with the ash 

 are probably of the same origin, while the pieces of shale that are 

 quite common in places were doubtless derived from the under- 

 lying Naknek sediments. All of these may be duplicated in the 

 lava and ejecta of Novarupta. 



THE rUMAROLES 



The fumaroles, which are now the most active volcanic features 

 of the region, usually find vent through the unconsolidated deposits 



