THE SENSORY CANAL SYSTEM OF FISHES. 503 
necessarily ensue. And here let me remark in passing that 
the system adopted by some authors—particularly Garman 
(82) and Allis (1)—of naming every separate branch a canal is 
one which I cannot assent to, for even were there a separate 
source of innervation to each and all of these branches, it is 
very questionable whether any purpose is served by adding a 
burdensome nomenclature; and further, when we come to 
review the sensory canal system in fishes generally, we find 
that the conclusions arrived at do not support such a method. 
In the most generalised forms the canals are simple open 
grooves, or grooves covered by scales, e.g. Heptanchus. 
There are a few exceptions to this in the deep-sea Teleosts 
such as Cottus bathybius, Liparis micropus, Lycodes 
murena, &c. The gradual specialisation to dermal canals, 
then canals partly dermal and partly borne by short drainpipe- 
like canal bones, then bony channels, and finally passing 
through the cranial and other bones, is readily traced in the 
Ganoids and Physostomous Teleosts. 
From what little we know of the development of these canals, 
the gradual evolution of the system is further borne out, for 
we find that they make their appearance first as isolated 
grooves, which coalesce and form canals, and these again join 
with others to form a network more or less distributed over the 
head. 
The whole question of the origin and evolution of this 
system I hope to discuss in some detail in a later paper. For 
the present, therefore, I think it advisable to treat the whole 
series of sensory organs, canals, pits, pores, &c., as one system, 
and to divide what have been termed canals into a series of 
branches. We must not lose sight of the fact that, whatever 
their origin or function, this system is of great importance to 
the fish, and is consequently subject to endless modifications, 
and therefore, in the present state of our knowledge, it would, 
I think, be unwise to lay down any hard-and-fast rule re- 
specting this subject of nomenclature. The scheme I have 
tabulated below, I think most morphologists will agree, is one 
based on broad and general principles, and will greatly facili- 
