A series of forty-eight stimulus patterns were painted and converted to sound by the pattern playback. At the Haskins Laboratories, New York, experiments concerned with resonant consonants have been carried out in which the approach was somewhat different from that employed in the experiments to be described in this paper. The basic pattern used in the Haskins experiments was probably designed to compensate for the absence of third format, since an addition of it to the Haskins pattern made alteration necessary to gain a satisfactory approximation to the speech sound desired.