No. 1.] EYES OF ARTHROPODS. 127 
retinophorze to which they belong. In the centre of the clear 
space is the vertical row of axial nerve fibres (Pl. X., Fig. 55). 
At the lower end of each double row of rods are two pairs of 
much larger ones belonging to the large retinophorze at the 
bottom of the retinal furrow. They are arranged in pairs like 
those of the small retinal cells, except that the clear space 
between each pair is alternately wide and narrow. 
The rods in the rows just above the gigantic ones differ from 
all others in that their free ends are united in pairs opposite the 
centre of the retinophore (Fig. 60). 
Toward the periphery of the retina, where the rods are up- 
right, the rod-mosaic is Hae of zigzag lines like those in 
Bare (Fig. 54a). 
DorsaL APPENDAGE. — Pesce the vertical, and horizontal 
retinas of eye I. there is a third part that up to this time we have 
left out of consideration. It is what in a former paper I have 
called the dorsal outgrowth of the large posterior ocellus. The 
observations recorded in this paper throw new light on that 
remarkable organ. 
We have already described how the changes seen in surface 
views gave rise to an oval sensory patch constricted off from 
the anterior, dorsal end of eye I., but with which it was still con- 
nected by means of a narrow furrow (Pl. VIII, Figs. 14-16). 
While the main portion of eye I. is being invaginated, the dor- 
sal appendage is depressed, forming a deep pit, at first round, 
then oval, and finally slit-shaped, the long axis of the depression 
being at right angles with that of the main eye (comp. Figs. 
8 and 14-16). In Fig. 79 is shown a longitudinal section of the 
main eye with a cross section through the middle of the dorsal 
appendage, during a stage corresponding to that shown in Fig. 
16. Ina similar section of a later stage, the pit is closed and 
three layers of cells are formed very much like those in eye V. 
(Fig. 80). The outer ends of the retinal cells soon lie close 
against the corneagen. The outer wall of the vesicle is flat- 
tened and probably forms an imperfect middle layer; at least I 
have seen in several cases one or two nuclei just beneath the 
corneagen, which, I believe, must be referred to the outer wall of 
the vesicle. Fig. 16 shows that the sensory cells connecting 
the ocellus with its dorsal appendage are joined to the latter at 
its anterior edge, consequently only those sections that pass 
