122 . BULLETIN OF THE 
one case we were dealing with an animal which had lost two cone cells 
or in the other, with one which had gained two; in other words, it 
would be impossible to determine which of the two conditions was the 
primitive one. The importance of embryological evidence in determin- 
ing this question must therefore be apparent. But evidence from even 
this source might not be conclusive. Thus in the development of the 
lobster I have traced in detail the steps by which the ommatidia are 
formed, and although in this Crustacean the considerable number of 
cells in each ommatidinm would warrant one in expecting some evidence 
of increase by division, the division of the cells in the retina is entirely 
accomplished some time before these elements show any grouping into 
ommatidia. Hence, the exact method of origin of the cells of the om- 
matidium cannot at present be given. I have observed that the same 
is also true in Gammarus ; cell division is completed before the cells are 
grouped into ommatidia. Perhaps in the development of some other 
Crustaceans evidence of the kind which I have sought may be obtained, 
but in the few species which thus far have been studied the evidence 
has not been produced. 
Although the supposition that ommatidia may increase the number 
of their cells by the division of those which they already possess is not 
supported by any direct observations with which I am acquainted, there 
are some facts recorded which are indirectly confirmatory of it. Thus, 
in Phyllopods, an increase in the number of cone cells appears to accom- 
pany a progressive differentiation of the retina itself. In this group, as 
I have already pointed out, the simplest condition of the retina is found 
in Branchipus and Apus. From the retina of Apus that of the Estheride 
can be easily derived, and the retina in the Estheride represents a con- 
dition from which the retina of the Cladocera may have arisen. That 
this series of retinas, from Apus through the Estheridz to the Cladocera, 
is a natural one is abundantly proved by the course taken in the develop- 
ment of the eye in these groups. If we regard the condition of the 
cones in these Crustaceans, we shall find that in the most primitive 
retina, that of either Branchipus or Apus, they consist of four cells ; 
that in the more complex retina of the Estheride they are usually com- 
posed of five cells, although cones of four cells are not unfrequent occur- 
rences; and finally, that in the Cladocera they are always composed of 
five cells. Apparently in this series the development of the retina is 
paralleled by a corresponding development in the cones, whereby one 
composed of four cells is ultimately converted into one with five cells. 
Since the resemblance between any two of the cells in a cone composed 
