The rat retina is uniform and contains almost exclusively rods. Therefore the rat eye, when uniformly illuminated, produces a gross electroretinogram (ERG) which is simply related to the activity of the individual retinal sources of the ERG. Characteristics of ERG's are shown on an intensity scale of the average number of quanta absorbed per rod per stimulus flash obtained by direct accurate measurement of all quantities involved. An independent check on the accuracy of these measurements is applied to pigment-bleaching data reported by Dowling (1963). When ERG characteristics are placed on this scale it is found that: (a) The b-wave can usually be observed when fewer than one out of two hundred rods absorbs a quantum, the threshold being determined by the noise of the preparation. (b) Near threshold the b-wave amplitude is proportional to intensity. (c) The a-wave appears when there are more than two to four absorptions per rod per flash. (d) The b-wave latency decreases with intensity, and the amplitude becomes proportional to the logarithm of intensity when fewer than one out of ten rods absorbs a quantum. This implies that the b-wave sources must combine excitation from more than one rod (probably more than seven). Therefore the b-wave cannot arise from independent rods or rod-bipolar synapses, but probably reflects activity of entire inner nuclear layer cells.