INVERTEBRATE PHOTORECEPTORS 651 



in this animal. Graham and Hartline measured the photometric inten- 

 sity of Hght stimulus required in various wave-length bands to elicit the 

 same nerve-impulse response to a flash of fixed duration. For any single 

 ommatidium the maximum sensitivity fell between 520 and 530 m^, and 

 no Purkinje shift could be demonstrated by varying the intensity over a 

 range of 1-100. Each ommatidium was remarkably constant in the 

 spectral-sensitivity values obtained in repeated trials, but each omma- 

 tidium also differed markedly from others. When the several curves for 

 a number of ommatidial action spectra were matched at the point of 

 maximum sensitivity, it became clear that some ommatidia show more 

 response than others to shorter wave lengths and correspondingly less 

 response to longer wave lengths. Even two different ommatidial-response 

 systems would make possible the discrimination of violet from red light, 

 and three would permit full color vision — if the central nervous system 

 were organized to utilize their unlike spectral sensitivities. In Limulus 

 the central nervous system seems to lack this necessary organization. 

 But a relatively minor change at the nervous level could provide for 

 color vision in an arthropod sensory system. 



The dimensions and divergence of ommatidia are subject to wide vari- 

 ations. Ommatidia facing downward commonly are relatively shorter 

 and have larger lenses than those facing upward. Usually they diverge 

 from one another more strongly. To a degree this is evident from a close 

 inspection of the large compound eye of a dragonfly, as was pointed out 

 by Ashton as long ago as 1840. But when sections are cut through the 

 eye so that ommatidial axes can be plotted carefully [as by Baumgaertner 

 (1928) for the honeybee eyel, it is seen that in particular directions diver- 

 gence between ommatidia reaches a minimum. For Apis this permits 

 maximum acuity to lie in a plane inclined 65° to the sagittal, and in this 

 plane only in an arc from 47° behind the anterior limit of the eye to 49° 

 ahead of the posterior limit of the eye. Not only do the ommatidia 

 diverge to different amounts, but each ommatidium appears to be com- 

 pressed in its pigment-cell sheath — not in the corneal lenses or crystalline 

 cones. Hence the angle between ommatidia is regularly greater in the 

 transverse plane of the head than in the frontal plane, with a ratio of 

 difference reaching about 2/1. Yet the radius of curvature of the bee 

 eye is smaller in the transverse plane than in the frontal, with a ratio 

 near 2.5/1. The whole compound eye then shows an astigmatic char- 

 acter to its visual-acuity possibilities. These morphological features are 

 paralleled by differences in behavior depending on the direction in which 

 the compound eyes are stimulated. 



Antrum (1949) has generalized to the effect that, in all insects that fly 

 well, the angle of view of each ommatidium in the horizontal direction is 

 about twice that in the vertical. Hence a target remains for a longer 

 time within the visual field of an ommatidium if it is moving horizontally; 



