Poly (I) codes for valine even more than for glycine. Since the ratio is constant over a wide range of Mg 2+ concentrations the ambiguity appears to be inherent in the translation of I-I-I, rather than induced by the relatively high Mg 2+ concentration (or the streptomycin) required with this messenger. These findings indicate that I in codon position 2 can pair readily not only with C (complementing glycine codon GG_ ¶ ¶ Genetic code assignments are taken from Nirenberg et al. (1966) . ) but also with A (complementing valine codon GU_). I in anticodon position 3 is also known to pair with A (as well as with U and C). Crick's model for this pair would require a large increase in interstrand distance, rather than simply “wobble”. This stretch would be avoided in an alternative model, suggested by Sakore & Sobell (1969), in which A would rotate from the usual anti to the syn conformation. Since a large stretch would seem difficult to accommodate in the less flexible middle codon position the present findings suggest that the rotation model for I·A pairing merits further exploration.