206 PATTEN. [Vol. I. 



closely connected with the brain than it is with the omma- 

 teum. 



Then, again, if the retinal ganglion of Arthropods is to be 

 regarded as a part of the ommateum, both structures should 

 arise from the same cell-layer. This, however, is not the case ; 

 for I have shown that the retinal ganglion of Vespa arises as a 

 secondary specialization of the nerve-fibres and ganglion-cells 

 arising from the optic ganglion. In my opinion there is not 

 sufficient evidence for supposing with Hickson, that all the 

 optic ganglia, as well as the retinulae, form one complete retina, 

 for we have seen that the -optic ganglion develops as a part 

 distinguishable from, but continuous with, the brain ; while the 

 layer to which the retinulae belong is an entirely independent 

 formation. The retinal ganglion cannot be regarded, I believe, 

 as distinct, functionally or morphologically, from the -rest of the 

 optic ganglion. The similarity in structure of the two parts 

 points towards this conclusion. I regard the shape of the 

 optic ganglia, and the, at first sight, marvellously intricate 

 arrangement of nerve-fibres in the medulla, as the resultant of two 

 tendencies: the one is to increase the number of ganglionic cells, 

 necessitating an extension of the cortical layer; the other is to 

 arrange the ganglionic cells, and the fibres arisyig from them, 

 in the least possible space. As soon as the cortical layer of 

 the optic ganglion has increased in extent so as to form a 

 nearly spherical mass, with the nerve-fibres in" the centre, 

 it appears to have reached a limit to its advantageous growth 

 in that manner. The further increase of ganglion-cells is 

 obtained by the formation of a new centre of growth, — the 

 retinal ganglion. This process might go on until several 

 ganglia had been formed. 



Although I have not as yet arrived at any conclusion 

 regarding the arrangement of the nerve-fibres in the medulla, 

 it seems probable that the successive layers of fibres running 

 at right angles to each other have been formed by alternating 

 growths of the ganglion-cells and nerve-fibres on different 

 sides of the optic ganglion. If we had a mass of fibres shaped 

 like an hour-glass, and composed of layers of spirally-wound 

 fibres, those in alternating layers being at any point nearly 

 at right angles with each other, we should have some such 

 arrangement as that in the medullae. A median longitudinal sec- 



