Since the discovery of Holmgren, the retinal action current has been investigated by many authors; recently Hartline on limulus, Jahn and Crescitteli on grasshopper, and Adrian on water beetle. The author studied on this subject using the compound eyes of grasshopper (Locusta danica), locust, water beetle (Cybister japonica) and fire-fly (Luciola cruciata). The experimental procedure and the results are in short, summarized as follows: Method: The retinal action current from the eye of decapitated animals was obtained with a pair of nonpolarizable electrode, placing the one on the decapitated cut end and the other on the surface of the cornea, and the current was recorded by Siemen's oscillograph. A small electric lamp from short distance was employed as stimulus. Results: 1) When the eye was stimulated by allight, a complex effect, i.e. after a small positive deflection (a-wave), soon followed by a sharp negative one (b-wave), after which a very slower one (c-wave) was observed. When the lithdrawn, there occurred also an off-effect (d wave). These waves coincide with those observed on frog's retina, indicating that three different processes occur together in the stimulated retina. 2) The magnitude of the deflection of b-wave increased with the duration of dark adaptation. 3) The magnitude of the deflection of b-wave was proportinal to the strength of the light. 4) Among amylalcohol, cocain, taurocholic acid and digitonine, when applied to the eye externally, the most effective drug, which diminishes the magnitude of the negative variation of retinal action current, was digitonine, while other drugs have little or no effect. 5) From the fact that cocain has no effect, we can conclude that the action current of retina does not depend on nervous element, but depends on the photosensitive substance which was dissolved out and disintegrated by digitonine. The substance is perhaps photosensitive substance in cone, because grasshopper retina has no rods.