When the nasal or temporal half of a developing eye in Xenopus larvae between stages 32 and 38 was removed, the remaining fragment became rounded up and subsequently developed into a normal eye by mid-larval stages. The contralateral retinotectal projection was normal in approximately twothirds of the animals, when mapped in later tadpole life. The remaining animals showed either some form of mirror reduplication in the visual map or else some degree of lack of order, as compared with a map taken from a normal tadpole of equivalent age. We conclude that the regulative powers of the larval Xenopus eye extend to much later stages (stage 38) than was previously thought; and a possible mechanism for the production of reduplicated maps from half-eyes is considered.In 1957 Szekely produced half-eyes by surgical operation on Triturus vulgaris embryos. The poles of the cut edge of the eye were apposed until healing took place, and the small, rounded, reconstructed eyes were then transferred to a host embryo. Operations were performed at a time when the naso-temporal axis was known to be determined, whereas the dorsoventral axis was as yet undetermined. Nasal and temporal half-eyes were transplanted with reversed nasotemporal polarity: the original nasal pole of the nasal half-eye was placed temporally and the original temporal pole of the temporal half-eye was placed nasally. When tested later by food-taking and optokinetic reflexes, the original nasal or temporal pole of the half-eye was found to have retained its original positional characteristics (or neuronal specificity) and the rest of the eye developed so as to form a normally organized eye, polarized dorsoventrally in proper relation to the orbit and polarized nasotemporally in accordance with the position of the original nasal or temporal pole.Straznicky, Gaze and Keating (unpublished observations) made half-eyes in Xenopus embryos at a time when both nasotemporal and dorsoventral axes had been established (Jacobson, '68). In these experiments of Straznicky et al., the edges of the remaining half-eye were surgically apposed (that is, the eye was rounded up) at the time of operation. The polarity J. COMP. NEUR., 162: 1s22 and connections of these eyes were later assessed by mapping the retinotectal projections and the results have been summarized by Gaze ('70). Various patterns of connection were found. Some of the eyes showed a normal projection. Some showed reduplication of the field projection resembling that seen in surgically produced double-nasal "compound" eyes (Gaze et al., '63, '65, '70). Other animals gave projections from the operated eyes which showed curling of the visual field contours (Gaze, '70). If the formation of the retinotectal projection is discussed in terms of the hypothesis of neuronal specificity (Sperry, '43, '45, '51, '65), we can say that these curling projection contours are what might be expected had the position-related specificities of the retinal ganglion cells remained unaltered after the operation. In this simplest case...