TRUE VOLCANOES. 385 



the margin of the crater fall away externally with a gentle 

 slope. 78 



The island of Amsterdam which lies 50' of latitude farther 

 towards the north (37 48') consists, according to Valentyn's 

 representation, of a single, well-wooded, somewhat rounded 

 mountain, from the highest ridge of which rises a small 

 cubical rock, almost the same as at the Cofre de Perote on 

 the higher plains of Mexico. During the expedition of D'En- 

 trecasteaux (March 1792), the island was seen for two whole 

 days entirely enveloped in flames and smoke. The smell of 

 the smoke seemed to indicate the combustion of wood and 

 earth ; columns of vapour were, indeed, thought to rise here 

 and there from the ground near the shore, but the natural- 

 ists who accompanied the expedition were decidedly of 

 opinion that the mysterious phenomenon could by no means 

 be ascribed to an eruption 79 of the high mountain, like that 



73 Valentyn, Beschryving van Oud en Nieuw Oost Indien, Deel iii, 

 (1726), p. 70 ; Het Eyland St. Paulo. (Compare Lyell, Princ. p. 446). 



79 "\Ve were unable," says D'Entrecasteaux, " to form any conjecture 

 as to the cause of the burning on the island of Amsterdam. The 

 island was in flames throughout its whole extent, and we recognized 

 distinctly the smell of burnt wood and earth. We had felt nothing to 

 lead us to suppose that the fire was the effect of a volcano" (t. i, 

 p. 45). A few pages before, he says, " We remarked, however, as we 

 sailed along the coast, from which the flames were rather distant, 

 little puffs of smoke which seemed to come from the earth like jets ; 

 yet we could not distinguish the least trace of fire around them, 

 though we were very close to the laud. These jets of smoke which 

 appeared at intervals, were considered by the naturalists of the expedi- 

 tion as certain proofs of subterranean fire." Are we to conclude from 

 this that there were actual combustions of earth, conflagrations of 

 lignite, the beds of which, covered with basalt and tufa, occur in such 

 abundance on volcanic islands (as Bourbon, Kerguelen-land, and Ice- 

 land) ? The Surtarbrand, on the latter island, derives its name from 

 the Scandinavian myth of the fire-giant Surtr causing the conflagra- 

 tion of the world. The combustion of earth, however, causes no flame 

 in general. As in modern times the names of the islands Amsterdam 

 and St. Paul are unfortunately often confounded on charts, I would 

 here observe, in order to prevent mistakes in ascribing to one observa- 

 tions which apply to the other, they being very different in formation 

 though lying almost under one and the same meridian, that originally 

 (as early as the end of the 17th century) the south island was called 

 St. Paul and the northern one Amsterdam. Vlaming, their discoverer, 

 assigned to the first the latitude of 3840', and to the second thatof 3748' 

 south of the equator. This corresponds in a remarkable manner with 

 VOL. V. "2 C 



