SUJ^FACE AND GEOLOGICAL STRUCTURE 
21 
than the time of the lower greensand, at a period that 
cannot be more exactly indicated. 
Concerning the formation of the marine chalk deposits 
in Western Madagascar, the same author writes as fol¬ 
lows: ‘Wt the time of the Middle and Upper Chalk a 
wide open sea stretched from Europe over the desert, 
over Arabia, perhaps over Somaliland, (that Somaliland 
belonged to the once existing Ethiopian middle sea has 
been since shown by Keller to be indisputable) and 
South European Chalk deposits penetrate into the in¬ 
terior of the East Indian tableland in the Valley of 
the Nerbudda.” This chalk sea apparently reached 
Madagascar, but only the Western side. During the 
eocene period it was still spread over a great part of 
East Africa and North West Madagascar. 
Then followed a subsidence of the coast line. 
Among other formations coming down to the present 
time the coral reefs must be mentioned. The informa¬ 
tion hitherto available is based on the earlier accounts 
of Captain Owen and Dr. Allan; these accounts have 
passed into the well-known works of Darwin, but in 
many points they do not appear to me to be well 
established. Thus, I find indicated on the maps a long 
fringing reef which is supposed to stretch without inter¬ 
ruption from Tamatave to Cape Amber in the north of 
the island, and then to go on and fringe the island of 
Nossi-Be. It is impossible, considering the mode of growth 
of the coral, from first to last, not to doubt the existence 
of so continuous a reef. The coral polype is emphatic¬ 
ally a salt-water creature which cannot endure a dilu¬ 
tion by fresh water; now as numerous streams descend 
from the eastern slope, these must prevent the formation 
of coral at their mouths. 
This continuous reef does not in fact exist, though 
there are isolated reefs. Point Hastie near Tamatave 
is fringed by a mighty coral bank, but near it, to the 
