It is the long pigment cells that are the puzzling element. Since there can be little doubt but that these cells can project and retract their pigmented parts (as already described), it would seem that a part of their function is to check the diffusion of light in the vitreous body when exposed to strong light. This function would be quite analogous to that of the pigmented cells of the vertebrate retina, which in light become projected between the rods and cones. Similar observations have also been made on the compound eyes of arthropods by Herrick[10] and by Parker[7], who find that the distal retinula cells of Palæmonites project themselves distad in the dark, thus surrounding the vitreous cones with a cylinder of pigment, while (Parker) the pigment of the proximal retinula cells migrates centrad and the accessory cells move distad; in light the reverse takes place. Other observations of this kind are not wanting for crustacea, insects and arachnids. To my knowledge, the pigment changes that I have described are the first of their kind for medusæ. I suggested while describing the capsule, that the lens might be adjustable. That the fibers of the long pigment cells extend to the lens is my principal reason for this. May these cells not represent ganglion cells and their distad fibers nerve fibers? That they are not sensory (_i. e._ are stimulated by light waves) seems to be suggested by their not having any axial fiber and in having several centrad processes. These facts suggest that they are not sensory but the center of a reflex mechanism.[h] When the sensory cells proper are stimulated, the impulses are conducted centrad into some nerve center (it may be the nerve tissue underlying the retina, or other nerve centers such as the two groups of ganglion cells in the upper part of the club, or the radial ganglia) from which center, again, impulses return over fibers leading to the long pigment cells causing them to project their pigment, and conducting the impulse to the lens, to produce a change in its adjustment. Since these cells are not so numerous as the prism and pyramid cells taken together, but in turn have a number of processes continued centrad (the sum of which processes approximates the number of sensory cells, prism and pyramid cells) it appears that these cells are admirably adapted to function in just such a mechanism as I have described,--each long pigment cell serving a number of its immediate neighbors.