No. I.] EYES OF ARTHROPODS. 



203 



layer become sickle-shaped and arrange themselves in pairs, a 

 single cell on either side of a calyx. These cells grow inward as 

 far as the neck of the calyx, where they terminate in a rounded 

 swelling containing a large nucleus. Their inner ends soon be- 

 come deeply pigmented, and appear to form a part of each omma- 

 tidium. On surface views of the eye at this time we can dis- 

 tinguish the small, faint nuclei of the corneagen and the 

 sharp outlines of the calyx, on each side of which is a large 

 semicircular cell derived from the middle layer. Surrounding 

 the sickle-shaped cells are the ends of a circle of 18 more cells; 

 in some cases I could only count 16 or 17. Each ommatidium, 

 therefore, including the four retinophorse, but not the two middle 

 layer-cells, is composed of 22 cells. 



The inner wall of the ganglion-fold, towards the close of 

 the pupal stage, breaks up into short strings of unipolar 

 ganglion-cells, similar to those of the optic ganglion, which 

 become continuous with the layer of ganglion-cells covering 

 the brain. 



The inner wall decreases in size until, at the close of the 

 pupal stage, it is no longer visible. 



The fibres of the nerve-bundle layer are not single fibres, 

 although they appear to be at first sight. They are composed 

 of a bundle of fibrillae which can only be recognized as such at 

 certain places. One of these fibres may be followed with cer- 

 tainty from the basal membrane through the nerve-bundle layer 

 and the neurilemma, to the retinal ganglion, where it is reenforced 

 by the fibrillae arising from the ganglion-cells of the outer layer ; 

 immediately afterwards it expands into a nerve-spindle of the inner 

 layer. The nerve-spindle is not, so far as I can observe, a " neuro- 

 spongium," such as described by Hickson (16), but merely a 

 point where its constituent fibrillae become swollen and refractive, 

 and less compactly arranged. From the retinal ganglion the 

 fibre, which can no longer be resolved into its elements, is con- 

 tinued to the convex surface of the outer medulla, where it turns 

 at right angles into the fibrous mass of the medulla, just below 

 the outer surface of which it expands again into a short nerve- 

 spindle, and is then continued straight onwards to the opposite 

 surface. On tangential section of the outer surface of the me- 

 dulla the nerve-spindles are seen in cross-sections as dark 

 points arranged with perfect regularity, and probably corre- 



