148 STEPPES AND DESERTS. 



Considering that our present (it is true, very limited) 

 geological knowledge of the mountainous parts of North 

 Africa does not make us acquainted with any trace of vol- 

 canic eruptions within historic times, it is very remarka- 

 ble to find among the ancients so many indications of a 

 belief in the existence of this class of phenomena,, in the 

 Western Atlas, and in the neighbouring west coast of the 

 continent. The streams of fire, so often mentioned in 

 Hanno's ship-journal, may indeed have only been strips of 

 burning grass, or signal fires kindled by the wild inhabitants 

 of the coasts to give to. each other notice of the danger 

 threatened by the appearance of the hostile vessels. The 

 lofty flame-enlightened summit of the " chariot of the gods" 

 (Qe&v oxrifjia), may recall obscurely the Peak of Teneriffe ; 

 but farther on Harmo describes a singular conformation of 

 ground. He finds in the Gulf near the Western Horn, a 

 large island, and in it a salt lake which again contains a 

 smaller island. South of the bay of the Gorilla Apes, the 

 same conformation is repeated. Is this a description of 

 coral productions, of "lagoon islands, (Atolls)" or volcanic 

 "crater lakes" in the middle of which a cone has been 

 upheaved ? The Triton lake was not in the neighbourhood 

 of the lesser Syrtis, but near the Atlantic coast. (Asie 

 Cent. T. i. p. 179.) The lake disappeared in consequence 

 of earthquakes which were accompanied by great out- 

 bursts of fire. Diodorus (Lib. iii. 53, 55) says expressly, 

 ?ri/poe K-0vr/juara jueTaXa. But the most wonderful con 

 formation is ascribed to the " hollow Atlas" in a passage 

 hitherto little noticed, occurring in one of the philosophic 



