44 BULLETIN OF THE 



membrane the retina and ganglion is one continuous mass of cells 

 (Fig. 38). When the intercepting membrane is formed, the retina and 

 ganglion are apparently separated (Fig. 39). I am of opinion, however, 

 that this separation is only apparent, and that in reality the two struc- 

 tures are still in connection. At least in tliis stage and in stage C 

 (Fig. 41) nuclei are frequently found lying directly across the mem- 

 brane. These nuclei present so normal an appearance, and their oc- 

 currence is so frequent, that I cannot believe that their position is due 

 to accidental displacement. My only way of accounting for the place 

 which they occupy is by supposing that the basement membrane is 

 perforated where they are found. Tlie membrane was probably pro- 

 duced in the form of a net, through the meshes of which the retina and 

 ganglion retained their original connection. Either this is true, or it 

 must be admitted that tlie retina and ganglion were first connected, 

 then separated, and finally reconnected ; a supposition which seems to 

 me unnecessary as well as improbable. From stage C to the adult con- 

 dition the retina and ganglion are unquestionably connected by nerve- 

 fibres. If the conclusion arrived at concerning the origin of the optic 

 nerve is correct, it follows that the optic nerve cannot be properly 

 described as an outgrowth either from the retina or from the ganglion, 

 but it must be considered as a remnant of the original connection which 

 existed between retina and ganglion. Patten ('87, p. 196) has described 

 essentially the same method of origin for the optic nerve of Vespa. The 

 formation of the optic nerve from tissue which represents the original 

 connection of a portion of the optic ganglion with the superficial ecto- 

 derm is doubtless a reproduction of the method by which that nerve 

 arose phylogenetically. 



The fact that the fibi'es of the optic nerve are from the outset attached 

 to the proximal ends of the retinulaj is of significance in determining 

 the plan of the eye. Of the four structural types of compound eyes 

 which have been suggested, that which Kingsley has presented involves 

 the inversion of the middle or retinal layer. Mark ('87, pp. 91, 92) 

 has shown that when the retina is invei'ted, as in the anterior median 

 eyes of spiders, the nerve-fibres are at first attached to the morphologi- 

 cally deep ends of the retinal cells, and that the attachment afterwards 

 migrates toward the other ends of these cells. From the fixct that in 



parts, retina, nerve, and ganglion, are contained in the eye-stalk. The fourth part 

 is a second bundle of nerve-fibres which connect the optic and cephalic ganglia. 

 In using the term optic nerve I refer to that collection of fibres which unites the 

 retina and optic ganglion. 



