AND THE MOKPHOLOGY OF THE EYE IN INSECTS. 411 



insects (fig. 57) I have been able to make out only two layers — a nuclear layer (nuc) and 

 i layer of large fusiform cells (cl). The latter are connected by a number of nearly parallel 

 aerve-fibres {no 2 ) with a deep ganglionic layer, which consists of small stellate cells (g). 



III. On the Development of the Compound Eye. 



The dioptron and neuron are developed from two distinct sources : the former origi- 

 nates from the hypodermis, the latter from a solid outgrowth of the cephalic ganglia ; 

 50 far, therefore, there is ground for a morphological comparison between the nervous 

 and dioptric structures of the vertebrate and compound eye. The dioptron is comparable 

 with the crystalline lens, whilst the neuron, so far at least, is homologous to the retina. 



The researches of Dr. Weismann * have shown that the dioptron in the Fly is formed 

 from a single layer of cells, and my own observations verify this in the case of the 

 Lepidoptera aud the Dragon-flies. 



It is well known that the outer facets of the eye in the Crustacea are developed later 

 than the more central facets, the eye increasing in magnitude with each successive 

 ecdysis. But I do not know that it has been remarked that the peripheral portion of 

 the eye in insects is less developed than the more central part. I have found this to be 

 the case even in the fully formed imago, whilst a section taken through the eye of an 

 immature imago often throws much light on the manner in which the eye is developed. 



In larvae in which the compound eye is functionally active such sections are still more 

 instructive. 



Perhaps it is by the gradual addition of fresh facets that the segregate retina takes its 

 origin. In the Dragon-flies, at least, the continuous retina of the imago is not fully 

 formed until after the final ecdysis, when it gradually replaces the segregate retina of 

 the larva. 



The dioptron in its earliest stage of development consists of a single layer of columnar 

 cells, which cannot be distinguished from the ordinary hypodermis of the insect. I have 

 observed this condition in various Lepidopterous larvse. The columuar cells lie imme- 

 diately beneath a continuous non-facetted cornea. 



At this stage numerous tracheal vessels and stellate cells lie beneath the columnar 

 layer, and processes from the columnar cells pass into the deeper layer, where they com- 

 municate with the stellate cells, which belong to the ordinary connective tissue of the 

 insect. 



In this respect the columnar cells closely resemble the cells of the sensory epithelium 

 in the Medusae and Mollusca. The researches of the brothers Hertwig f show that in 

 the MedusaB the cells of the sensory epithelium, in the region of the rudimentary eye- 

 spot, become the terminal organs of the nerve. This may possibly be regarded as au 

 indication in favour of the hitherto accepted view that the sensory termi nations of the 

 nerve in the Arthropod are developed from the hypoderm. Such a conclusion is, how- 

 ever, in my estimation, no more justified than if it were extended to the case of the 

 crystalline lens of a vertebrate. In the Medusa the epithelial cells, from which the 



* Zeitsclir. f. w. Zool. Bd. xiv. 



t 0. und R. Hertwig, Das Nervcnsystem und Siunesorgane dos Medusen. 



