glaber.

Uncertain. Hypothesized to be from Proto-Italic *ɣlaðros, from Proto-Indo-European *gʰl̥h₂dʰ-ro- (“smooth”), from Proto-Indo-European *gʰleh₂dʰ- (“smooth, shiny”), which is reconstructed based on the assumed cognates Proto-Germanic *gladaz (whence German glatt (“smooth”)) and Proto-Balto-Slavic *glā́ˀdus (“smooth”) (whence Lithuanian glodùs (“tool for smoothing”)). These stem from a basic root *gʰleh₂- per Schrijver 1995. However, Kroonen denies this etymology for Proto-Germanic *gladaz, preferring to derive it instead from *glōaną, which Kroonen derives from *gʰleh₁-. See Proto-Indo-European *ǵʰleh₁-. *The change of initial *gʰl- to Latin gl- may be regular (despite the contrast with the regular change of word-initial *gʰ before a vowel to Latin h-). Compare medial *-gʰl- becoming *-gl- in trāgula and figulus. (De Vaan 2008 assumes a similar evolution of *gʰr- to *gr-, citing grāmen, whereas Schrijver 1991:311 assumes based on rāvus that word-initial *gʰr- became Latin r-.) Weiss 2018 makes the alternative suggestion that g- in glaber could result from a conditioned dissimilation of roots of the form DʰlVʰD-/DʰrVʰD- to DlVʰD-/DrVʰD-, comparing trahō. In any case, there are no clearly established counterexamples (that is, no solid etymologies showing a change of *gʰl- to Latin l-), although there is a probable case of *gʷʰr- to fr- in frendō. *Schrijver 1991:191 considers it likely that PIE *CRHTC- (*CRH- followed by two consonants, the first a plosive) regularly gives Latin *CRăTC- (compare flagrum, fragrō). To explain this development, De Vaan suggests that the laryngeal may have been lost in this context "before the putative allophonic prop vowel was phonologized."