According to Christol (1996: 809ff.), some Latin words in -ss- (allegedly with “expressive gemination”) could receive a principled explanation as systematic exceptions to rhotacism. At the time of rhotacism, conservative dialects might have remained at the stage with intervocalic /-z-/ (as in Oscan). Then, speakers in rhotacizing dialects (or using a rhotacizing phonostyle) would identify the intervocalic [z] with [ss], at that time the only intervocalic sibilant in rhotacizing speech; cf. Lat. 〈ss〉 rendering Gk. 〈ζ〉 (= [z]). Christol’s own examples are not strongly supportive; but Lat. crassus ‘thick, fat’, grossus ‘unripe fig’ (perhaps also grossus ‘thick, coarse’), and classis ‘levy’ could provide good examples of the process. The etymologically obscure crassus and grossus can thus be given satisfying accounts based on attested PIE s-stems, s-presents, and/or s-enlarged forms. (The discussion of crassus includes analysis of the neglected form crassundia ‘sausages’, which may support an old s-present.) For classis: the root etymology was clear (cf. Lat. calāre ‘call, summon’, U. kařetu/carsitu ‘he shall call’, Gk. καλέω ‘call’, etc.), but the -ss-formation was not; it can now be interpreted, given Christol’s theory, via s-forms attested within Latin (e. g. Calābra) and elsewhere (cf. the s-present in Hitt. kalliš- zi ‘call, summon’).