56 Canadian Record of Science. 
glabella of Triarthrus on the back, which proves its geolo- 
gical horizon. It has two specimens of Cyathophycus close 
together, nearly perfect at their bases and broken off at the 
height of about three inches. They are perfectly flattened 
and pyritized, which is also the condition of other fossils in 
these shales, with the exception of the graptolites, which 
seem to have resisted this kind of change. 
The genus Cyathophycus was originally described by Wal- 
cott from specimens obtained at Trenton, Oneida Co., New 
York. He regarded it as an alga, whence the termination 
1 Trans. Albany Instit., 1879. 
“ phycus,” but subsequently, in the American Journal of 
Science, 1881, corrected this error, and referred it to the 
sponges. Hall (35th Regents’ Report) properly places it 
with the reticulate sponges included in his family Dictyo- 
spongidae, but does not add much to Walcott’s original 
description, to which the present specimens permit some 
additions to be made. 
The specimens are perfectly flattened, but show distinct 
indications of the two sides of the originally conical form. 
The wall of the skeleton has evidently been thin and com- 
posed of slender bundles, each of a few long simple spicules, 
and increasing both by bifurcation and the introduction of 
new bundles, so as to preserve nearly the same distances in 
the wider parts of the cone. They are very regular in the 
lower part, where there are about nine principal, with 
some intermediate secondary bundles in a centimetre, but 
become more irregular toward the top. This may, how- 
ever, be an effect of decay and crushing. At the base these 
bundles become thicker, and in a specimen from the origi- 
nal New York locality, kindly lent to me by Mr. Ami, I 
have observed that they become expanded and converted 
into somewhat short clavate root spicules. This is, how- 
ever, not apparent in Mr. Miller’s specimens, which may 
have been broken off at the surface of the mud. 
The vertical bundles are crossed at right angles by hori- 
zontal spicules much less regularly arranged, but dividing 
the surface into rectangular meshes. These are slightly 
