1. Electrical measurements indicate that there is a local barrier to extracellular current flow in the outer part of the locust compound eye, and that neurons must interact amongst each other because of this structure (Shaw 1975). An attempt to confirm this structure independently is described here, visually with dye tracers, and more quantitatively with radioisotopes. 2. Dyes injected into the blood (haemolymph) rapidly appear in a parallel array of capillaries which transfix the neural layers anterio-posteriorly just below the basement membrane of the retina, but dyes cannot be detected within the retina or the optic lobe neuropile even after many hours. 3. Procion dyes infiltrated through a hole in the cornea spread throughout the retina, but appear to travel no further centrally than the basement membrane, although pigmented glial cells extending down to the capillaries take up the stain. In this region below the basement membrane, dye movement both into and out of the eye seems to be retarded, in agreement with the electrical measurements. 4. 3H-inulin and3H-polyethylene glycol (PEG) both diffuse only very slowly from the blood into the eye, and finally indicate an exchangeable space of about 3% of the eye volume, believed to be extracellular. The electrical measurements require a similar space. 5. Theoretical uptake curves for the unrestricted diffusion of inulin and PEG have half times of about 9 and 4 min, respectively. The half times of the curves best fitting the actual measurements are 4.5 and 1.6 h, indicating that diffusion is much retarded. If as believed, this retardation occurs only in the basal retina, the diffusion coefficients of the two tracers in this region have to be reduced by about a hundred to fit the uptake curves. 6. The smaller tracers sorbitol, mannitol, sucrose and sulphate all indicate an unrealistically large space in the eye and (except sulphate) leg muscle, a space which does not equilibrate with the blood (except with sulphate). This suggests extensive penetration of the cells in these tissues by the tracers or their metabolites. Autoradiography of3H-sucrose in the retina shows this too. 7. The uptake of sucrose into the eye is similar to that predicted for diffusion followed by removal of tracer by cellular absorption. The uptake rate measured for sorbitol, and that estimated for saccharides to explain the observed metabolic rate, are much too large to be explained by diffusion then uptake. This suggests that a rapid bypass route exists for some compounds across the barrier, to handle metabolic needs. The basal pigmented glial cells are properly positioned, and take up both sucrose and Procion Yellow. 8. Taken overall, the results confirm the electrical measurements in suggesting that the retina is sealed off from the blood by a basal barrier, which greatly retards diffusional access to the eye. The isolation of the retina from the lamina synaptic region should be several times more severe.