The retinal fiber layer and the juxtaretinal portion of the optic nerve of goldfish have been studied with light and electron microscopy in order to determine whether the age-related order of fibers in the nerve originates in the retina.In the retina, no patent spaces (channels) were noted. The fibers ran in fascicles and consisted of two classes: nonmyelinated fibers, which ran superficially (close to the vitreal surface), and "myelinated" fibers, which ran more deeply and were loosely wrapped by processes presumed to be glial. The myelinated fibers were larger and presumably older. The nonmyelinated fibers are believed to be the young ones, from the peripheral, more recently generated, ganglion cells, for the following reasons.(1) Their size and cytoskeletal elements were typical of young axons. (2) They were the only axons in peripheral retina. (3) They were continuous with the nonmyelinated fibers in the nerve, previously shown to be the young ones. (4) When retinal axons were cut peripherally, the degenerating axons were in the superficial part of the fiber layer. (5) Growth cones, presumably from the newest ganglion cells, were always observed at the most superficial position in the fiber layer, in direct contact with the basal lamina of the inner limiting membrane superficially and nonmyelinated fibers deeply. The nonmyelinated fibers always clustered together in the retinal fiber layer and occupied the most central portion in the cross-section of the optic nerve head. Thus, the age-related organization of fibers in the nerve is established in the retina.These results are discussed in the context of growth, with the aim of evaluating the relative importance of four factors that might influence the intraretinal course of the growth cone. Its interactions with other fibers and with the basal lamina of the inner limiting membrane seem to be more important than interactions with the glial end feet or guidance into open, preformed channels.

The selectivity of axonal regeneration to the extraocular muscles in teleosts has been reinvestigated by mapping, with retrogradely transported HRP, the motor pools of the muscles innervated by the oculomotor nerve. In normal goldfish, the motoneurons of the superior rectus, inferior rectus, and inferior oblique muscles formed discrete, nonoverlapping motor pools; the motor pool of the medial rectus muscle overlapped with those of the inferior oblique and inferior rectus muscles. In fish whose oculomotor nerve had regenerated (after intracranial transection), in contrast, many motoneurons in other, inappropriate motor pools reinnervated the superior rectus and inferior oblique muscles (the only muscles examined in lesioned animals). Furthermore, these inappropriate motoneurons continued to project to these muscles for at least 1 year. The oculomotor nerve and its molecular branches were examined by light and electron microscopy to determine the pathway by which axons regenerated to their muscles. Axons regenerated within the basal laminae of Schwann cells, which persisted in the distal nerve-stump after a lesion. After labeling the inferior oblique nerve with HRP in regenerated nerves, there were labeled axons in all of the muscular branches; this indicates that regenerating axons branched, which was confirmed by finding an increased number of myelinated axons in other, regenerated inferior oblique nerves. Thus, different branches of the same axons sometimes reinnervated different muscles. These results demonstrate that regenerating axons in the oculomotor nerve are misdirected to inappropriate muscles, and do not selectively reinnervate individual muscles, as had been previously suggested (Sperry and Arora, 1965).