354 
The main conclusion is that the convex eye of Arthro- 
pods is a group of hair-bearing sense buds. 
Each ommatidium of Belostoma contains one very large cell, 
whose pointed outer end terminates in a double rod lying in the centre 
of the retinula (fig. 1, g!). In some cases this cell is double, like 
the retinophorae in the ocelli of Acilius; and there are indications 
that the six peripheral retinula cells are built on the same plan. 
There are at least three other cells in the centre of each ommatidium. 
The largest one is spindle-shaped and its outer end may appear either 
fibrous or granular, or as a slender, hyaline rod that lies side of the 
double one belonging to the large retinula cell described above (fig. 1, 
g? and A‘). Of the two remaining cells to which no rods are 
attached, one is for the most part a coarse fibre. Its nucleus, if still 
within the limits of the eye, is probably one of the small red bodies 
often seen near the base of the ommatidium. Each of these nine or 
ten retinula cells is continuous with a bundle of nerve fibrillae 
(1891,22); 
In Tabanus, at the base of the circle of seven, equal, retinula 
cells, is an eighth cell, which is very short, but nevertheless provided 
with a distinct rod and nucleus (fig. 2, band ce). Within the circle is 
a bundle of axial nerve fibres which can be traced through the base- 
ment membrane into the optic nerves (fig. 2, b and c). 
The cone cells in Belostoma are very small and their rods 
are only slightly thickened and refractive in the place where the cone 
in other Arthropods is present. Each cone cell terminates in 
a hair-like process which probably abuts against the cornea 
(fig. 1, r. h.). 
The inner ends of the cone cells extend, between the retinula 
cells, as four delicate fibres along the outer surface of the rhabdom; 
I could not trace them inwards as far as the basement membrane. 
Essentially the same condition seems to prevail in Tabanus and in 
Vespa during the pupal stages. In Homarus also, according to 
PARKER, the ends of the cone cells pass around the spindle. Thus 
in these cases, contrary to my former observations on Penaeus, 
the cone cells are not continuous with the rhabdoms. 
My observations on Belostoma and Vespa were made independ- 
ently and some time before the appearance of PArker’s paper. Publi- 
cation was delayed in order to study earlier stages, but the necessary 
material could not be obtained. 
In Tabanus, the distal ends of the cone cells form a hya- 
line rod, the pseudocone, which is expanded and hollow at the 
