76 DR. H. B. GUPPY. 
tirely occupied the banks, the third stage, that of the horse-shoe 
island, is reached, In some instances, there is yet a further stage, 
when, during a long continuance of westerly winds, another bank 
is thrown up across the mouth of the horse-shoe, and a small atoll 
with a shallow lagoonlet is produced. Thus the currents are the 
principal agencies in forming the horse-shoe islands of Keeling 
Atoll. In large atolls, where more open-sea conditions prevail in 
the lagoon, and especially where, as in the Maldives, there are two 
opposite sets of winds and surface-currents, each prevailing in its 
own half of the year, we should expect to find the horse-shoe island 
replaced by an atollon. Keeling Atoll, however, les for eleven 
months out of the twelve within the region of the constant trade- 
wind and westerly drift current, so that the situation is only one 
favouring the formation of horse-shoe islands facing to the south- 
ward and eastward. The protected character of the lagoon, also, is 
not a condition that would assist the growth of a circular island or 
atollon. 
Another important feature in this atoll is to be found in the 
existence outside the seaweed edge of the present reef of a series 
of submerged lines of growing corals separated from each other 
by sandy intervals. Unfortunately, I was not able to examine 
these to the extent I desired, since it can only be satisfactorily 
done later in the year, when the sea is sufficiently smooth to allow 
boats to approach the breaker edge of the reef. This feature, how- 
ever, is familiar to the residents, who have supplied me with infor- 
mation on the subject. It would seem that all around the circum- 
ference of this atoll there is a space outside the present edge of the 
reef varying from 200 to 500 or 600 yards in width, where ships 
have anchored, and where boats in the calm season go with fishing 
parties. Here the submarine slope slopes gradually down to 20 or 
30 fathoms ; but beyond this the descent is precipitous. It is on 
this gradual slope that the lines of growing coral occur, separated 
by sandy intervals from each other. There may be two or three of 
these lines, the innermost covered by 4 or 5 fathoms, and the outer 
by from 20 to 30 fathoms. 
We are thus able to perceive that the outward extension of the 
reef is effected, not so much by the seaward growth of the present 
edge of the reef, as by the formation outside it of a line of growing 
corals, which, when it reaches the surface reclaims, so to speak, the 
space inside it, which is soon filled up with sand and reef débris. 
The evidence, in fact, goes to show that a reef grows seaward rather 
