During this period *S was becoming PMP *h, and *s shifted to an alveolar sibilant to fill the gap created by this change, but not until a pattern of alternation with *ñ under prefixation was firmly established. This does, however, create a problem, since we have already found it necessary to reconstruct PAN *c on the basis of a different sound correspondence. If we treat *s as a palatal fricative this problem can be resolved, and the reconstruction is still consistent with the scenario Ross has proposed.