EFFECT OF STREAM WORK 421 



water until the stream accumulated force enough to overcome its barriers, 

 when it rushed down in pulsations to the sea. By the time of my second 

 visit to Saint Vincent, in March, 1903, the immediate gorge of the 

 Wallibu had been mostly cleared of its filling of ash, except for material 

 retained in protecting curves of the walls, and a considerable amount of 

 material had been washed off from the hill slopes. At that time hot, dry 

 dust-flows were still carrying material out from the sides into the bottom 

 of the gorge (see plate 44, figure 2), showing one method by which the 

 dry season contributed to the removal of the fresh material. The general 

 history of events has been that dust-flows during dry weather and the 

 moderate showers of all seasons have brought material down into the 

 gorge which the torrents due to heavy rains have carried out into the sea. 

 My thiid visit, in the latter part of June, 1908, was at the end of the 

 dry season, when the effects of this filling process were quite evident. An 

 immense amount of volcanic gravel and sand formed a comparatively high 

 floodplain in the mouth of the gorge, and for half a kilometer or more 

 inland from the line of coastal bluffs, ready to be carried out when the 

 rainfalls should provide sufficient water for the purpose. One afternoon 

 during this visit there was a downpour of rain lasting about two hours. 

 This brought the river down on the surface of its bed to the sea in pulsa- 

 tions overloaded with debris, and easily rolling along bowlders 30 or more 

 centimeters in diameter. The stream flowed in the form of waves 30 to 60 

 centimeters high, whose crests passed the point of observation every 10 

 to 12 seconds. Following each crest was slack water, with consequent 

 deposition of sediment tending to change the course of the next crest and 

 break it up into rivulets distributed over the broad expanse of the river 

 bed. These pulsations were exactly like those noted in the Wallibu and 

 Rabaka in May, 1902, 6 which evidently were not a function of either the 

 primary or the secondary eruptions. Neither can temporary dams thrown 

 across the streams by landslides, as postulated by Anderson and Flett, 

 be considered their sole or even principal cause. The explanation ad- 

 vanced independently by Professor Eussell and me received ample con- 

 firmation in June, 1908, when frequent showers brought the Wallibu and 

 Rabaka rivers down in pulsating floods, though the volcano had been 

 quiet for years ; the thick beds of hot ash had largely or quite disappeared, 

 and the gorge had widened so that landslides, become comparatively in- 

 frequent, could not possibly have any regularly intermittent effect on the 



6 E. O. Hovey : Bull. American Museum of Natural History, vol. xvi, 1902, p. 344. 

 I. C. Russell : National Geographic Magazine, vol. xiii, 1902, p. 276. 

 Anderson and Flett : Philosophical Transactions of the Royal Society of London, 

 series A, vol. 200, 1903, p. 430. 



