Jastrow's note "(=gammīdā)" is not clear since no such form is quoted in his dictionary
Note also gurmaiza 'fist' compared in [KB 196], but thought to be an Iranian loanword in [DM]. If cognate, -z requires explanation
Likely a semantic shift; for a similar shift, cf. UGR yd and HBR yād 'hand' and 'penis'
A phonetic diversity throughout SEM can be explained by (1) low compatibility of a voiced and voiceless stop in HBR and ARM, and (2) change of *-m- > -n- in contact with -d (for a similar phonetic evolution cf. *p/ban/md-(aw-)at-, No. ).
Cf. AKK gamatu 'a measure' SB in LL [CAD g 32], [AHw 278]; one wonders whether it can be explained from *gamattu < *gamad-t-u or from *gamadu, with -ad- mistaken for the -at- suffix, developed from *kamadu because of low compatibility of k and d.
Cf. ARM *garmīdā, with -r- inserted probably by contamination with some similar term; could it be ARM and SEM *garm- 'bone' (No. )?
The AMH and GAF examples were alternatively identified by Leslau with AMH kǝrn, TNA kʷǝrnaʕ 'elbow' [LGur 347] (see *kʷirnāʕ-, No. ), which is also possible. A more obscure case is END h_ǝtnä, hǝtnä 'arm, cubit, arm below the elbow' [ibid.]; it is hard to say whether this may derive from the present root or from *kʷirnāʕ-, in either case with a series of unusual phonetic transformations.
Note a derived pattern and a meaning shift in MSA.
[KB 196]: HBR, ARM (-r-); [Brock 133]: SYR, ARM, HBR, GEZ (gʷǝnd 'baculus').
Comparison with ETH and MSA forms denoting 'stick, wood' (TNA gʷämäd, JIB gend) suggested in [Leslau 15] is unconvincing
