A scale of the perceptual differences among Japanese consonants has been constructed by the AA. The present task was to find whether the relative differences among pairs of consonants has any effect on the retrieval of consonant-vowel (CV) syllables in either primary or secondary memory. Pairs of CV syllables were formed using 17 consonants and the single vowel /a/. Each consonant was paired with all others - and with the absence of a consonant - in all, 18 beginnings of syllables. The pairs of syllables were assembled in sets of six. Each pair (slide) was presented on a screen for 1.5 sees. Groups of 3-5 Ss read the syllables aloud as they saw them. After the sixth pair was shown the Ss turned one of 26 pages on a pad of response forms and tried to write the six missing syllables, one member of each pair. The mean inter-consonantal difference between the error responses and the targets was computed. This, averaged for 52 Ss - 26 who supplied one member of the pair and 26 who supplied the other -, provided a measure of primary memory for each of six sequential experiences. The procedure was replicated with another 52 subjects. They were required to spend 10 secs. counting backward by 3's after viewing the sixth slide and before responding. This was taken as a measure of secondary memory. In both instances the error responses were more similar to the targets in terms of the previously derived scale values than would be expected by chance.