248 Cincinnati Society of Natural History. 
structure. When a transverse section of a pellet is immersed 
for a moment in carmine ink and then washed in water, the 
cortical layers will be seen to have readily absorbed the ink, 
while the inner layers are not affected by the staining fluid. 
Still, upon crushing, even the cortical layers present a consid- 
erable degree of hardness. 
The pellets are of a strikingly uniform appearance. With 
the exception of the smallest ones, on which the incrustation 
has just commenced, and which therefore still present the 
irregular form of the nucleus, they are in most cases nearly 
spherical, rarely spheroidal or discoidal. 
About one-third, in bulk, of the whole quantity consists of 
broken pellets, the remaining two-thirds being unbroken, and 
the total number of these is 456, varying in size, weight and 
form as follows: 
25 pellets, diameter 2.0 to 4.0 mm, form irregular nodular; cortical 
layer thin, or none; ends of tubuli often plainly marked on the sur- 
face, but rarely with an open lumen. Weight of the smallest specimen 
.09 gm. 
37 pellets, diameter 4.0 to 4.5 mm, form approximately spherical. 
Average weight .22 gm. 
384 pellets, diameter 4.5 to 5.0 mm, nearly spherical. Average weight 
.24 gm. 
10 pellets, diameter more than 5 mm in one direction, not over 
4.5 mim in another direction. These 10 specimens are either prolate or 
oblate spheroidal in shape. The largest one, an oblate spheroid, hay- 
ing diameters of 4 and 7.5 mm, weighs .40 gm. 
The largest fragment of the coral (Fig. 1) measures 8 mm. 
in width, but only 3 mm in thickness, and weighs .15 gm; 
another, 6 mm. in width, has a thickness of 3.5mm. ‘The 
others a still smaller. 
In order to account for the uniformity of the round pellets 
and the scarcity of the flat ones, we may suppose that, in tap- 
ping the underground repository of these peculiar formations, 
the upward stream of water carried only those pellets which 
did not exceed a certain maximum weight. As to the flat 
pieces, whose weight exceeds that limit, it seems that the 
4 
