Are Latin pons, pontifex and the Indo-European cognates evidence of an i stem?

Abstract

Sanskrit panthāh, Avestan pantǡ, Old Persian paθim, Latin pons (and its compound pontifex), Greek ό and ά, Armenian hun (genitive hni), Old Church Slavic pǫntƅ, Old Prussian pintis cannot easily be reconciled into one single paradigm and, consequently, the exact reconstruction is debated. It has been argued that the Indo-Iranian, Latin, Armenian and Balto-Slavic forms are evidence for an i stem, either original (Schmidt, Bezzenberger, Hirt e.a.) or of secondary and laryngeal origin (Beekes, Schrijver). Starting from the two different "i reconstructions" this article re-examines the Latin, Greek and Indo-Iranian cognates, and tries to account for the evolutions in the different languages. We agree with Beekes, Schrijver and De Vaan in that the Latin nominative is problematic and the Armenian form corresponds perfectly to a reconstruction *pontH. In addition, we believe that also the compound pontifex and the Scythian name  fit into this schema. However, we have our doubts as to the paradigm with a nominative *Hs and an accusative *eHm, find the independent innovation in Sanskrit and Avestan less likely, and consider the Old Persian form pθim not conclusive, because it is a back-formation on the nominative and especially in light of the Scythian name , which raises some questions as to the exact Iranian treatment of the Proto-Indo-Iranian cluster *nth. We therefore believe that the original reconstruction *ponteh1s (made by Pedersen in 1926) still has preference, despite the problems that it poses for Latin.*