1. The electrical pathways that couple the rods and that link the outer segments of the rods to the coupled network, were evaluated. Two separate micro-electrodes were inserted into the inner or outer segments of the same or neighbouring rods under visual control. Current was passed through one electrode, and the resulting potential recorded with the other. 2. The input resistance, measured at the inner or outer segment in a rod in the network, is strongly outward rectifying. It is typically near 40 Momega when the membrane is hyperpolarized 10 mV or more by extrinsic current, less than 10 Momega when the membrane is depolarized by 5 mV or more, and near 30 Momega at the no-current level. 3. When current is injected into the outer segment, the response in the inner segment is nearly identical with that at the outer segment, suggesting that the resistance coupling the segments is not high relative to the input resistance of the rod in the network. 4. Under voltage clamp the light response current for a rod in the network is of constant magnitude for potential levels between -80 and -20 mV. This suggests that there is little or no measurable light elicited conductance change associated with the response, possibly a consequence of coupling between rods. 5. The rod response increases with increasing diameter of a concentric test flash up to about 200 micron, or about 16 rod diameters. 6. When current is injected into one rod, the response in its immediate neighbours is between a quarter and one tenth that recorded in the injected rod for all potential levels in the injected rod. 7. The membrane time constant, measured in a rod in the network, is proportional to the voltage-dependent input resistance at 0.16 msec/Momega. With assumptions about the geometry of the rod network this represents a membrane capacitance of 1.5 muF/cm2. 8. The data can be approximated by a network model of square array. The model predicts that: the outer segment contributes less than half the current for the total rod response, the membrane resistance of an individual rod is greater than twice the measured input resistance for the rod in the network, near 60 Momega, and the coupling resistance for each arm of the network is about 4 times the individual rod resistance, near 240 Momega.