Stonry—Limitation of Insect Vision. 241 
from the primary surface to the inner surface. To make it 
possible to do this the glass hemisphere which we have 
used to represent the primary surface may now be removed; 
it is no longer required, since its position is sufficiently 
indicated by the points of junction of the indiarubber funnels 
and the glass threads. The outer surface of our model, which 
carries the lenses, should be a stiff, immovable arch, but 
the inner surface is to be made of some material which is 
capable of slightly contracting. If, after constructing the model 
in this way, its inner surface is made to shrink a little,! this will 
pull the glass threads inwards and elongate all the indiarubber 
cones. In this way the narrow ends of the cones are brought 
farther from their lenses, into the position where the image of a 
near object would be formed. The model now represents the 
insect’s eye when accommodated for the vision of near objects. 
In this model the image of the outer world is formed either at 
the primary surface or at the inner surface, for a speck of light 
falling on the upper end of one of the glass threads will travel 
lengthwise along the thread and emerge from its lower end, being 
kept from escaping laterally by total reflections. Now, it seems 
probable that something of this kind actually occurs in the insect’s 
eye. In fact the apparatus corresponding to the rods and cones 
of our eyes seems, so far as I can make out, to be situated, not at 
the primary surface where the image is first formed, nor even at 
the inner surface where the image may be reproduced in the way 
described above, but in a deeper situation with which the inner 
surface communicates only through curved transparent threadlets. 
Each of these threadlets seems to have a thin transparent core, 
and if this core be of sufficiently highly refractive material, it 
would, although curved, be competent to carry the light forward 
by total internal reflections, from the lower end of one of the 
glass threads to one of the pieces of the apparatus which corre- 
sponds to the layer of rods and cones in our eyes.” 
1 A diminution of the radius of the inner surface of the model to the extent of about 
one millimetre would effect a sufficient range of accommodation. The motion in the 
insect’s eye may need to be more than in proportion to this, since the filaments, as 
well as the funnels of its eye, are probably extensible, which is not the case in the 
model. 
2 The light is probably carried forward most effectually where, as in the dragon- 
fly, the threadlets are less than a micron in section, 7.e. not much more than the 
S 2 
