SUVADIVA ATOLL. 413 



50 yards broad. The lagoon islets in the neighbourhood are all sand heaps covered with bushes, 

 their shores giving signs of more or less considerable loss. 



From Gadu we went almost due W. to Nadale, putting down 25 soundings on the way. 

 All these showed an increased depth as compared with the chart, varying from 1 to 6 fathoms. 

 They contrast with those of the E. side, where the depths were all about the same as recorded. 

 Indeed, there appeared to have been a real difference, an actual average increase of about 2 fathoms 

 in depth along our line. The bottom was everywhere coated with mud, except in the immediate 

 proximity of shoals. The latter along our course were all located, and appeared to be smaller, but 

 no accurate observations or comparisons were possible. The lagoon islands all showed washing away, 

 but the process evidently is not very rapid. With regard to the southern encircling reefs there 

 is a marked tendenc}' to velu formation, but most of the enclosed pools are stated by tlie natives 

 to have formed one or more communications with the lagoon of the atoll. 



The islands immediately to the S. of Nadale show varying degrees of washing away on their 

 lagoon sides. To seaward all run out into points, formed of coral rock, which are separated by 

 some breadth of sandy boat channel from a continuous boulder zone. All have towards one another 

 points of mixed sand and stone, apparently growing out to join the whole line together. Mahota 

 thus is almost a rhombus in shape, the two acute-angled points to seaward and lagoonward, and 

 the two other angles towards Nadale and the S. Erosion is taking place on Nadale both towards 

 the sea and the lagoon, but the former is very slight, the beach being formed of coral rock. The 

 latter forms a belt 100 yards broad, but does not seem to have ever extended out as far as the 

 boulder zone, the reef-flat having been formed bj' the outgrowth of the reef edge. On the lagoon 

 side the water has broken through the sand into a kuli, which formerly seems to have occupied 

 the centre of the island. At its entrance is a small sand islet with three coconut trees. 



On its lagoon side Nadale has a sand-flat, noticeable for its enormous quantities of sand-feeding 

 Holothurians and Polychaets. It is about 400 yards broad, and ends abruptly with large clumps 

 of Mndrepora against a velu full of coral patches, 8 — 12 fathoms deep. These masses increase in 

 size towards the lagoon, and are characterised by their perpendicular, or overhanging, sides, built up 

 of such massive genera as Goniastraea and Porites. Some are so large that their centres are bare, or 

 even slightlj' hollowed out, and quite dead. The E. reef is well-defined and likewise arises perpen- 

 dicularly from the velu, but towards the atoll lagoon the bottom is visible for 150 yards, covered 

 with great areas of Madrepora and Lohophi/tum. The velu appears to stretch S. as far as Kaudu, 

 sending small bays out towards the sea between the different islands, and communicating freely with 

 the lagoon. To the N. the velu continues for about 2 miles, but then its enclosing reef gives place 

 to a series of isolated shoals, a small velu only appearing again to the N. of Huadu. This change 

 is evidently due to the breaking up of the reef, the original line being clearly shown by the regu- 

 larity of the line of shoals. 



The islands from Nadale to Hondedu have all sandy beaches on their lagoon sides, exhibiting 

 little or no signs of erosion. Their seaward ends have rocky shores, and reach generally to the 

 boulder zone, which is well-marked ; change is all in the direction of loss rather than gain. Between 

 the islands on, or inside, the boulder zone is a series of coral masses awash, which indicates that 

 the land formerly extended in an unbroken line from Hondedu to at least Kuramati. The latter 

 island is interesting from its dumb-bell shape, two heads against the lagoon and sea, joined by a narrow 

 bar, either side of which is strewn with timber. This is peculiarly interesting, as showing how the 

 removal of the land may take place from the middle of a reef outwards. It bears also on the 

 meaning of the presence of a double line of islands on the encircling reef, as found further N. 

 by Hondedu, and on the E. side between Gadu and Huluwarolu. Sandstone masses lie quite on 

 the lagoon side, indeed almost on the edge, of the reef by Kuramati, Hondedu, and the other islands 

 to the N. It is accordingly probable that land once covered the whole of the lagoon reef in this 



53—2 



