438 
DRS. T. ANDERSON AND J. S. FLETT ON THE ERUPTIONS OF THE 
the lake of mud behind them, and a large part of the contents of one of the lateral 
valleys was suddenly discharged into the river. In some of the rawnes these lakes of 
thick, black mud were quite half a mile long, and if heavy rains should cause rapid 
erosion by the main stream, flows of mud-lava may yet take place on a large scale. 
The hot sand in the Rabaka Valley does not readily form mud. It is too hot for the 
rain to penetrate far below the surface, and the water evaporates rapidly. The mud 
lakes had been fllled with the cold material from the slopes above, which had been 
washed into the depressions by the heavy rainfall. 
The terraces and other features due to .steam erosion exhibited on the Wallibu were 
repeated in all their essential features in the valley of the Rabaka Dry River, onR 
here the process was far less advanced, owing to the .smaller body of water which 
characterised this stream. The new cut gorge was not over 20 or 30 feet deep, the 
terraces on the wdiole less numerous and closer together (see Plate 32). Two 
or three of them could often be seen, 3 or 4 feet apart, and as the valley was broader 
than that of the Wallibu, the terraces were better preserved and more extensive. 
But much of the deposit retained its original rolling surface, and only a part of the 
slieet of sand in the valley had been remodelled by running water. The firces of the 
terraces and the banks overlooking the .stream often slipped and the hot grey ash 
was then exposed. It showed only a rudimentary and very imperfect stratification, 
and at a depth of a foot or so the sand was grey and quite dry, while the upper part 
was moist and of a darker colour. 
Only after rains did water occupy the stream channel; usually it was empty, and 
the moist, stony mud on the bottom continuously steamed. But after a heavy shower 
great steam explosions rose from the whole upper course of the river, and probably 
they were no less violent than those on the Wallibu (see Plate 31, fig. 1). As the 
banks were lower, the landslides of hot ash were smaller, and great rolling sudden 
steam-jets were not so frequent as on tlie leeward side. Sometimes the river flowed 
hot, black, and thick witli mud, right down to the sea, and on one occasion we had 
to ride our horses through the surf, as a current of mud, several inches deep, was flowing- 
in the river, and was too hot for our horses to cross it. This was only after severe 
rains, as in most cases the water was completely evaporated in passing through the 
hot sand in the upper part of the channel. 
In more than one place in the valley of the Rabaka Dry River we observed large, 
circular, flat-bottomed depressions, some of which were perhaps 20 yards in diameter, 
while others were not more than 8 or 10 feet (see Plate 33, fig. 2). At most they 
were 8 feet deep, and often they occupied the apex of a low flat cone, tlie diminutive 
size of which, as compared with the broad, flat central basin, reminded one of the 
lunar craters. Very commonlv the surface of the ash fields around such a crater 
was covered with scattered blocks in considerable numbers. They were most frequent 
* As this proof is lieiiig corrected for the press, ve learn that such mud-hovs have actually taken place. 
‘Sentry’ newspaper, Kingstown, November 28, 1902. 
