The Oscan form pui at Cp. 37.1, as Clackson points out, is not exactly attested. The sequence actually reads according to Rix's edition pụ[i: / pu]ị. The lithograph produced by Buecheler in RhM 1878 shows just the tail of something that could be an i. If the restoration is correct we have a virtual pui, but a more important issue is whether pui is the nom. sg. as I interpreted it or the dat. sg. = Lat. cui. (so Vetter p. 424). Looking at the context again (the so-called Curse of Vibia) it seems more probable to me now that pui if correctly restored is a dat. sg. If that is the case it could not be an exact morphological match for cui, but a simple remodeling of the inherited form on the basis of the thematic dat. sg. So at p. 351 pui should simply be stricken and at p. 470 the Umb. form poi or the Osc. nom. sg. fem. paí should be substituted.