Intracellular recordings were made of the slow hyperpolarizing light responses of single rods in the retina of the snapping turtle. Physiological criteria used to identify rods were verified by intracellular injections of Procion Yellow. 2. The amplitudes of the responses elicited by fixed intensity flashes increased as the stimulus was enlarged to a diameter of 300 mum. Scattered light was found incapable of accounting for this effect, which must result from summative interaction of rods with neighbouring receptors. Effects of summative interaction were observed even at stimulus intensities that produced maximal responses. Enlarging the diameter of the higher intensity stimuli from 100 to 300 mum increased the peak response amplitude by almost 50%; it also produced a distinct initial peak of the response which we term overshoot. The amplitude of this overshoot was graded with stimulus size. 3. Complete intensity-response relationships were determined using stimulus diameters of 100 and 750 mum for each rod. With the smaller stimulus the intensity response range was 4-5 log units, and with the larger stimulus this was increased to 5-0 log units. For intensities below about 60 quanta/mum2 per flash (514 nm) the amplitudes elicited by the large stimulus followed a sigmoid-shaped curve. However, at higher intensities an additional lobe appeared on the intensity-response relationship. The appearance of this lobe correlated with the emergence of the overshoot on the response wave form. 4. Determinations of rod flash sensitivity (mV per quantum per mum2) showed that it increased with stimulus size up to a stimulus diameter of about 300 mum. With diameters between 50 and 150 mum, a linear relationship existed between the flash sensitivity and stimulus area. Absolute quantal sensitivities increased with stimulus area by a factor of 26, from a value of 28 muV per photoisomerization per rod with a stimulus 25 mum in diameter, to 720 muV per photoisomerization per rod with a stimulus 300 mum in diameter. 5. By comparison, red-sensitive cones showed increased sensitivity as a function of stimulus size only up to a stimulus diameter of 120 mum. Their over-all sensitivity was lower than that of rods and proved linear with stimulus diameter rather than with stimulus area. 6. Simultaneous recordings were made from rod-cone pairs to determine whether the overshoot, and hence the lobe on the amplitude-intensity function, could result from a cone input to the rod response. The time course of the cone response proved much too rapid to fit the overshoot of the rod response. 7. The spectral sensitivity of the dark-adapted rod response closely followed the difference spectrum of the rod photopigment for wave-lengths greater than 450 nm. This was true throughout the intensity range of the response, including low intensities where response averaging was necessary. 8. At low response amplitudes (approximately 1 mV), about 70% of the 40 rods tested showed responses to long wave-length stimuli consisting of two components...
