This paper will apply the idea of “reconstructing from below”: what protolanguage would we arrive at if we set out exclusively from the attested languages and did not make any use of what we think we know about Proto-Indo-European? So far as Iranian is concerned, the results would probably not differ too much from our usual reconstruction, exactly because reconstructed PIE heavily relies on Greek and Indo-Iranian. I will thus transfer the exercise to a subgroup of Iranian, called “Central Iranian” for these purposes. The languages chosen are Bactrian and Parthian, which by various criteria are very close, although belonging to the Eastern and Western branches, respectively. Crucially, Parthian and Bactrian share a number of interesting and non-trivial morphological innovations in addition to a series of phonological changes and a shared nominal system. The experiment will attempt to reconstruct the protolanguage of Parthian and Bactrian with the aim to see what the result amounts to and whether it looks like a robust hypothesis of an intermediary protolanguage.