THE VOLCANOES OF GUATEMALA. 
481 
reducing the measured thickness at the same time. A large portion of 
the rain, however, ran off, producing a feather-pattern erosion, shown in 
some of the earlier photographs, like that noticed in St. Vincent and 
Martinique.* As in those islands also, the torrents of water and mud 
flowed in new courses independent of the old ones, which had been filled 
up, and formed new channels, in many cases cutting deep into the 
soil and subjacent beds which previously existed. Thus a deep and 
narrow ravine, about 80 or 100 feet deep and perhaps 100 or 120 feet 
wide, now exists to the east of, and not many yards from, the Plaza of 
Palmar. It has been cut out of the old tuff and agglomerate, and now 
conveys the water from the river Nima, which formerly ran in quite a 
different direction into the Samala (see Plate III.). In other cases the 
floods carried away bridges and deepened the old ravines, and the mud 
brought down blocked up the river Ocosito near Ocos, and altered the 
configuration of the coast in that neighbourhood. All these changes 
strikingly recall similar ones in St. Vincent.t 
When the surface of the ash deposits had become more consolidated, 
and before denudation had had time to produce much effect, access to the 
mountain was easy, but as the rain and atmospheric agencies did their 
work, deep gullies were formed, divided by narrow ridges. In the low 
grounds, where change had been less active, the ridges, at the time of 
my visit, were generally flat-topped, where the crust had protected the 
underlying, less-consolidated material. This often weathered into almost 
vertical walls, till another somewhat harder layer was reached, which 
formed a new shelf, the whole making a succession of steps, such as to 
remind one of the tops and keyboards of a succession of pianos placed 
end to end, along the tops of which it was not difficult to walk (see 
Plate IV.). Further up the mountain, where the process was further 
advanced, intermediate blocks had often been entirely washed away, 
and this necessitated constant ascents and descents, which were 
decidedly fatiguing. Further up again, as in Plate V., the whole 
top crust had generally been removed, and the ridges were often 
reduced to mere knife-edges, which were liable to give way and pre¬ 
cipitate the traveller into the deep crevasse on either side. This plate, 
which is taken on the south-west of the mountain above Palmar and 
the site of the Baths of Sabina (from which direction the wind was 
blowing at the time of the eruption), exhibits well the comparative 
thinness of the ash on that side of the mountain, as shown by the dead 
tree-trunks, which still project through the ash. Nearer the crater 
the ash becomes much thicker and the barrancos deeper. This plate 
also exhibits the structure of the mountain, with the beds of tuff and 
agglomerate dipping conformably to the slope. This shows that they 
* Anderson and Flett, part i. plate 28 ; Sapper, 4 Ereignisse,’ Taf. vii. 
t Anderson and Flett, loc. cit. ; Anderson, Geographical Journal , March, 1003. 
