98 APPENDIX T. 



existing east of it, impinge on this side and undermine the bank, which is made vertical for a height of 3 feet. 

 Thus the trees fall on to the Ijeach, with their matted roots forming a mass a yard or more in diameter, 

 and enclosing from 6 to 12 inches in thickness of sand, pumice, bits of coral and breccia; while the 

 matted roots of the other vegetation of the surface in some places project as a shelf, till this also liecomes 

 too much undermined to hold together, and then they also break awaJ^ In that way this side of the 

 island is being eroded all along and carried into the lagoon, thus determining the eastern outline of the island. 



Close to the west side of the island, and between it and INIotungie, a channel has been eroded through 

 which the tide commences to flow to the lagoon soon after it has covered the tidal platform, and that has 

 determined the western outline. Its present length, therefore, simply indicates a portion of what was once 

 its width before erosion by the waves had cut a channel on either side of it, increased by the addition of 

 the foraminiferal sand accumulation at its north or lagoon end. 



That this is so, is shown by the eroded long scarps previously mentioned in the base of what was the 

 breccia sheet, which here and there roughly, but fairly, indicate its oceanward outline. These scarps, which 

 sweep round from the south of Avalau in more or less of a curve, are continued by some further outliers 

 to near the point of breccia south of this island, and then on to the island beyond (I\Iotuloa), the only 

 interruptions to this line being where the breccia curves in to form the boundary of the breaches and 

 channels which are mentioned above. On the lagoon side also, if a concave curve be drawn to the lagoon 

 side of the eastern three-quarters of Motuloa and extended to Motungie, that will pass quite close to the 

 north end of this island; this further suggests that there is nothing unusual or ei'ratic in the circumstance 

 of the longer axis of Nukusavalivali crossing the I'cef, as it merely indicates a portion of what its width 

 has been at an earlier period. 



To the east of this island between it and Motuloa the wide breach ])efore mentioned is actively corroding 

 the base of the still opposing l)reccia : through and over this the sea pours and rushes with the incoming 

 tide, into a rather deep pool (which has probably been hollowed out by this force). It lies on the lagoon 

 side of the l)reccia, but within the line of curve defining the nortlicin boundary of Motungie, Nukusa- 

 valivali and the major portion of Motuloa. Near the side of this pool and close to Nukusa\alivali, in the 

 breccia sheet, from which it has not l)cen fully relieved, I observed one of the large mammilated corals, 

 Montipora (? species) almost horizontally beddeil and nearly as perfect as when it grew ; being markedly 

 distinct, in this respect, from the ordinary included masses of broken and worn coral rubble and reef 

 fragments. A few paces to the north of this occurs a coarse sandstone dipping north 10" cast, at 10', and 

 undergoing erosion ; the strike of this sandstone would conform to the local curve of the atoll rim as is 

 the case with the sandstones at Avalau. The existence of these sandstones indicates that conditions were 

 once favourable for their deposition : such as (1) the former presence of an extension along the reef of the 

 land now occupj-ing positions on these rocks, and (2) a lagoon Ijoundary of a form suitable foi' them to 

 al)Ut on, and agreeing with theii' o1)served present strike. 



The pool close b}' this has, I think, received its general form before the present breach occurred, and 

 was probably in part at least filled again when the sandstones were laid down. These have all been 

 excavated again hy the inrush of waters, which would also have removed the strip of_sandstone referred to, 

 but that it has almost certainly been protected since the time of its deposition, till within no very distant 

 time, by the overlying material similar to that composing the upper portion of the adjacent island. Then 

 this super-imposed material was remo\ed 1)y the erosion proceeding along the cast side similar to that 

 which now brings down the trees of Nukusavalivali. At that time pr()lial)ly nnuh of the erosion of l)rcccia 

 on the lagoon side of this and succeeding islands took place. The l)reccia has subsequently been covered 

 over by the advance of the island matei ial lagoonwards, as is shown by the observed northward extension 

 of Nukusavalivali. 



