CHEAT CRATER OF MANINDTU. 401 



unite on one side, and leave a tongue of land project- 

 ing from the east and west sides, Eacli of tliese cir- 

 cles in a crater, and the tongues of land that project 

 from either side of the lake mark the boundai'ies be- 

 tween them, Tlie width of the larger erater at the 

 level of the lake, as given on the l^est maps I have 

 been able to consult, is three geographical miles; 

 that of the smaller crater, at the same level, two and 

 a quarter miles ; and the length of the lake, which 

 lies in a northerly and southerly dii-ection, and is ap- 

 proximately parallel to the gieat Barizan chain in 

 which it is found, is no less than six geographical 

 miles. These two craters, I believe, were not fomed 

 at the same time. The larger crater, which is on the 

 north, is older, and the smaller one to the south is 

 the later, the eruptive force which fonned the larger 

 ha\dng lost some of its power, as ^vell as having 

 slightly changed its position when it fonned the 

 smaller. This gigantic crater is the more interesting 

 to us, because it is as large as the one we supposed 

 formerly existed in the Banda Islands, when we re- 

 garded Great Banda, Fulo Pisang, and Pulo Kapal, 

 as parts of the walls of that crater, if, as "was then 

 suggested, that erater was not circular, Init nearly 

 elliptical, like this gi'eat one of Manindyu, Even the 

 famous crater of the Tenger Moimtains becomes of 

 moderate dimensions, when compared to this. 



In the western side of the larger crater is a cleft or 

 deep ravine that conducts the superfluous waters to 

 the sea. Tliis split, it may be noticed, has occm-red 

 on the side toward the sea, where, of course, the wall 

 of the crater was thinnest and weakest. This re- 

 26 



