One can easily surmise that the u-forms have secondarily emerged under the influence of b, at least some of them are quite ancient and may reflect a secondary lexicalization of the two variants as early as in PS (perhaps with semantic specification “well” vs. “pit”). Note that for Fronzaroli 1971, 611 *buˀr- is diachronically primary with respect to a more innovative *biˀr-, which is hard to explain in phonological terms.