THE DEVELOPMENT OF THE COMPOUND EYE. 553 



During the third stage of the development of the dioptron, 

 the optic stalk and neural portion of the disc expand and their 

 tissue becomes spongy. The fibrous structure of these parts 

 becomes indistinct, and its elements widely separated from 

 each other. They eventually disappear entirely and the space 

 which they occupied is seen to be filled with delicate retiform 

 tissue. 



In the meantime the sub-dioptric space becomes larger, and 

 the mesoblastic cells which occupy it become more numerous 

 and increase rapidly in size. These give off processes which 

 grow into the dioptron, and are gradually converted into 

 tracheal vessels. 



In the fourth stage of development the retina has extended 

 into the cavity of the neural portion of the disc, taking the 

 place of the retiform tissue and eventually filling the whole 

 space bounded by the peritoneal coat of the expanded optic 

 stalk and its neural disc. 



During the last three or four days of the pupa stage the sub- 

 dioptric space gradually disappears, and the pre-retinal mem- 

 brane comes into immediate relation with the basilar mem- 

 brane. 



The Development of the Rstinal End Organs. These appear to 

 be developed from the epithelial elements of the retinal disc, 

 but I have been unable to follow the whole process. It ap- 

 pears to me that the cells from which they originate undergo 

 considerable elongation, and that their nuclei are displaced 

 inwards so that they lie beneath the end organs, between them 

 and the optic nerve fibres. 



The non-inversion of the terminal elements in the sense in 

 which they are said to be inverted in the Vertebrate eye, has 

 already been used as an objection to the views which I have 

 expressed ; but there is no involution of the neural epiblast 

 from the commencement, and if the roof of the vesicle of the 

 cerebron originates by delamination from the hypodermis, as I 

 have suggested, no inversion of the retinal elements would 

 occur. Nor do I think that the importance of the position of 

 the rodlets is of so much importance as has been ascribed to 



