BRANNER: THE STONE REEFS OF BRAZIL. 263 
would therefore be regarded as fringing reefs. This relation has usually 
been brought about, however, by small geographic changes, such as the 
drifting of the shore sands behind the outer reefs. The channel between 
the coral reefs and the shore is deepest between the Abrolhos and the 
Parcel das Paredes. It is there only eleven fathoms. For the most part 
the depth is considerably less, and too small to admit the entry of ordi- 
nary sailing vessels. 
With the exceptions noted below the coral reefs of Brazil have no 
apparent connection with eruptive phenomena. There are probably in- 
Cipient reefs about the volcanic island of Fernando de Noronha, and it is 
possible or even probable that the Rocas reefs are built upon an erup- 
tive base. But the Rocas reef is two hundred and twenty-four kilo- 
metres from the Brazilian mainland, with a channel of over two thousand 
fathoms separating it from the barrier reefs of the coast. The island of 
Santo Aleixo, just south of Cape Santo Agostinho, is likewise eruptive, 
but it is only two or three kilometres from the shore, and therefore has 
no relation with the coral reefs other than that of the sedimentary rocks 
along the same coast. There are igneous rocks also in the Abrolhos 
Islands, but they are sheets and dikes in the rocks that form the group. 
Tun CHEMICAL Composition or THE BRAZILIAN CORALS. 
I have had analyses made of a few of the skeletons of living corals 
the results of which are given in table (A) on page 264. 
All samples were washed with boiling distilled water to remove sea 
salt, and the complete removal was verified. The washed samples were 
dried, and of these the analyses were made. All the specimens contained 
considerable organic matter. 
It should be noted that these analyses are of skeletons of polyps, the 
upper portions of which were living when the samples were collected. 
Samples were also taken of the dead reef rock at Ponta do Mangue, 
State of Alagoas, one hundred kilometres northeast of Maceio, and of 
this rock an analysis has been made with the results as given in table (B) 
on the next page. 
If all the calcium present were there as carbonate, and the rest of the 
carbon dioxide as magnesium carbonate, it would be equivalent to 
Jarbonate of lime (COCO) е + а . . 8219 
Jarbonate of magnesium (МЕСО) +. . . 12.08 
It should be especially noted that the specimens of the reef rock rep- 
resented by the last analyses were taken from an old reef that has long 
been dead. 
