The pioneer work of Granit (1942) showed that a single ganglion cell in the frog's retina was not always connected to one type of receptor. The studies described in our three preceding papers (Donner & Rushton, 1959a, b;Donner, 1959) were undertaken to explore further the connexion between ganglion cells and the different types of receptor. It was clear that if in any given state only one type of photoreceptor was connected, stimulation by substitution would exhibit three characteristics. (a) A silent substitution could always be achieved if the intensity of the new light was correctly chosen, the correct intensity (in quanta/sec) for each wave-length varying inversely as the absorption at that wave-length by the visual pigment involved. (b) The Fechner fraction, that is the logarithmic range of intensities over which the substitution remained silent, must be constant whatever wavelength was involved. (c) Adaptation to any colour which did not alter the connexion between receptor and ganglion would leave the spectral sensitivity of silent substitution unchanged.The results of our first paper (Donner & Rushton, 1959 a) showed that some ganglion cells did not exhibit silent substitution, and hence must have been connected to more than one type of receptor at the same time. This was a complexity which we wished to avoid, and so we chose only those cells which did respond usually by silence when the substituted intensity was properly adjusted. Even these cells failed in a restricted range of intensity just below the level where the photopic dominator curve is obtained. But in the photopic range the results were consistent with the idea that ganglion cells are connected to a single class of receptors (cones) whose visual pigment absorbs according to Granit's (1942) photopic dominator curve: in full dark adaptation they are connected to other cells (rods) whose sensitivity corresponds to rhodopsin, and in changing from dark to mesopic adaptation there is a gradual change from the pure rhodopsin sensitivity to one with 'humps' in the green and in the blue, but with the characteristics (a, b, c) above still maintained. A formal explanation could be that some inert screening pigment 22PHYSIO. CXLIX