*kummo-

Based on the evidence from Luwian kumma/i-, Lyc. kuma- and derivatives and Pal. aš-kummāwa-, one can reconstruct a root with a word initial voiceless velar obstruent, a labial glide and another sound that would have assimilated to m before another m in both Luwic and Palaic. Without any Indo-European comparative evidence, one could posit a root of the shape *keu̯H- or *ku̯eH-. Instead of the root-final laryngeal, one could reconstruct a nasal m or n instead, e.g. *ku̯em- or *ku̯en-, the latter of which would likewise assimilate.
In the literature so far, Luw. kumma- was reconstructed with a root-initial palatal *k̂ because of an etymology with av. spəṇ-ta- (Schindler apud Watkins 1987b:401 still followed by Melchert 2012a:211). However, if one does not believe the hypothesis of Melchert (ibid.) that a palatal *k̂ develops to k in Luwian before /o/, for which there is no convincing evidence, the etymology should be abandoned. Furthermore, the Luwian cognate of Hitt. kunna- is zunni-^(ti) ‘to make alright, to rectify (?)’, which the expected outcome of *k̂, i.e. z-, can be seen; cf. under lemma. For more information on the reconstruction of kumma/i- see the Luwian derivative kumma-.
Regarding the stem, an adjectival o-stem is tentatively reconstructed based on the possible presence of a Luwian adjective kumma/i- ‘pure, sacred(?)’ (though only attested as neuter plural or collective kumma). The only secure stem that can be reconstructed is the one in *-eh₂-; see PAnat *kummeh₂-. Luw. kumma/i- has been taken to be the equivalent of Hitt. šuppi- ‘pure’ (Watkins 1987b:399-401), which remains possible, although it is not entirely proven; cf. Laroche 1979c:108f. for conjectural evidence with personal names. For a list of place names possibly containing the word kumma/i- see Watkins ibid.

[D.S.]