356 
the pseudocones (fig. 3, 0. s.). This spine-bearing cornea is soon shed 
and a faceted one formed, each facet, which is in the main the product 
of two newly formed corneagen cells, often containing in the centre 
the remnants of an ommatidial spine! The pseudocone disappears 
soon after the spine-bearing cornea is shed. Just outside of the eye 
of Vespa are many hair-bearing cells over which, in the first pupal 
cuticula, temporary spines are formed resembling those over the omma- 
tidia. This cuticle is apparently pushed off by the growth of perma- 
nent hairs, which at first are protoplasmic outgrowths or secretions 
which resemble in almost every particular the pseudocone of Vespa! 
In Aphis similar hair cells are found in abundance between the 
ommatidia. 
Now if the ommatidia are hair-bearing sense buds, we ought to 
find (1) some resemblance between isolated hair cells and retino- 
phorae, (2) we ought to find isolated hair cells acting as rudimentary 
ommatidia. Such is really the case, for, as regards the first point, 
the isolated hair cells of Vespa are beyond all question double 
cells and contain a coiled canal, which for several reasons I believe 
to be continuous with a nerve tube (fig. 4 C.). After the first pupal 
moult, the larger component cell forms a long, protoplasmic process 
which is finally converted bodily into one of the bristles so abundant 
on the forehead and about the eyes. Briefly stated, these double hair 
cells resemble the retinophorae of Molluscs and Arthropods: (1) in their 
axial nerve canals, (2) in the imperfect union of their twisted, com- 
ponent cells, and (3) in the position, size, and color of their nuclei 
(fig. 4 C.). As regards the second point, I have found hair cells 
between the ommatidia in the convex eyes of Aphis, Vespa 
and Belostoma; in the last two cases, the protruding 
hairs were absent or very rudimentary, and in all 
three, the cells were surrounded by a layer of pigment 
so that they bore a striking resemblance to very 
simple ommatidia and probably functioned as such! 
The development and permanent condition of the corneagen bears 
directly on the views here advanced; and I consider it an important 
fact that I am now able to show, after the examination of new ma- 
terial, that the corneagen of Vespa is not formed by the union of 
two folds over the whole eye, as I formerly suspected, but by the 
growth of two cells over the distal end of each ommatidium (fig. 3, 
c.g. and #.). Even more important is the fact that in the oldest stages 
of Vespa there are minute openings in the corneagen, through which 
protrude hair-like projections of the cone cells (fig. 3 E.), making a 
