*1 Cf. PPN *maqoha "cooked".
*4 POC *ma-osak "ready to be eaten (because ripe or cooked) (LPO I:157).
*6 PMP *(ma-)qesak "ripe, cooked, ready to eat" (ACD).
*8 This set of reflexes is in almost perfect complementary distribution with those of PN *maqoha, and there is no recurrent semantic differentiation. *moso is therefore presumed to derive from an earlier *maosa, a doublet of *maqoha. However it remains unclear just how the present distribution of the two variants came about.

The following note by BB explains an earlier attempt to derive all forms from a single original of the *maqoha form:
Note. At some point all non-Tongic languages lost *h initially and between unlike vowels, but (? some) SO languages retained *h, as s, between like vowels (most) of the time. If *moho was innovated from *ma-oha, (the two forms are in complementary distribution), then, to account for the retention of the *h in SO languages, the assimilation of the final *a must have occurred first. But our subgrouping theory would not allow for this (because EP languages appear to have retained *a) unless we suppose that *h was lost twice (totally in EP and partially in SO). As has often been noted, SO may not be a subgroup.