It has been demonstrated that mixtures of ‘primary’ tastes may be perceived as singular rather than as multiple (Erickson and Covey, 1980). This finding may be accounted for in two different and testable ways. First, suppression: one of the tastes in the mixture may be suppressed by the other (i.e., in a mixture of A and B, A suppresses the taste of B); the prediction which follows from this hypothesis is that a solution of A (the suppressing component) should be indistinguishable from the AB mixture. Secondly, synthesis: the singularity of the mixture may arise from the formation of a new taste, and in this case both components should be distinguishable from their mixture. The present data indicate that the latter is true in some cases. Since the taste of these mixtures must be other than the primary tastes used to form them, these findings question the presumption that only four ‘primaries’ are adequate to describe the range of tastes.