94 A NATURALIST IN THE PACIFIC chap. 



prospect of finding rocks exposed on its upper part, its slopes 

 being densely covered with tall reeds, my examination was con- 

 fined to the lower portion during a walk around the island. On 

 its east and north sides occur rocks of much the same character as 

 those exposed in the neighbouring low promontory to the east of 

 it. In addition to agglomerates and basaltic andesites occurred a 

 rubbly pitchstone composed of fragments, up to a centimetre in 

 size, of an opaque brown glass displaying a few phenocrysts of 

 plagioclase and pyroxene, the interstices being filled with crushed 

 fragments of the phenocrysts and finer glass debris. This rock is 

 allied to the " crush-tuffs " described on page 334. It may be added 

 that the basic tuffs are more frequent on the west and south sides 

 of the island. 



The low island of Na Vatu in the midst of the Soni-soni inlet 

 is about 250 feet across and only 3 or 4 feet above the ordinary 

 high-tide level. In 1898, when I visited it, this tiny island pos- 

 sessed about 20 houses and a population of 60 or 70 persons, and 

 I gather from Hazlewood's account of these islands that Na Vatu 

 was crowded with houses more than half a century ago. It was 

 apparently in the first place a sand-key, and is protected against 

 the wash of the waves by a low sea-wall formed of large blocks of 

 stone. 



An interesting exposure of bedded tuffs and clays is displayed 

 at Ravi-ravi on the west side of the peninsula. A broad shore- 

 flat has been formed by the marine erosion of a line of coast 

 composed of these deposits. The strike is well exhibited, the dip 

 being about 30 degrees N. by W. Here there are alternating beds, 

 a few inches thick, of coarse and fine tufaceous sandstones, some- 

 times calcareous, with marls or calcareous clays. The mineral 

 fragments of the coarser rocks are composed of plagioclase, augite 

 .and rhombic pyroxene, the last being abundant and giving a more 

 acid character to these deposits. The calcareous fragments appear 

 to be principally shell debris. The marl is in part composed of 

 much finer detritus of the same minerals. The other materials of 

 these deposits are derived from the degradation of basic andesitic 

 rocks, and include also a little palagonite. To the westward of 

 Ravi-ravi these beds show signs of disturbance, being steeply 

 tilted to the N.W. Agglomerates also occur in the disturbed 

 area. 



The history of the Kumbulau peninsula is evidently the history 

 of the eruptive phases of a number of more or less submerged 

 small vents and of the periods of great marine erosion that 



