From: Bhrihskwobhloukstroy
Message: 71766
Date: 2014-07-31
Latin _fu:mus_ 'smoke' has traditionally been regarded as the exact cognate of Sanskrit _dhu:máh._ 'id.' and Greek _thu:mós_ 'breath, spirit, life'. With laryngeal theory, these can be regarded as reflecting a zero-grade PIE derivative *dHuX-mó- of a laryngeal-final root *dHeuX- or *dHweX-. The laryngeal is identified as *h2 by adding Hittite _tuhhae-_ 'to cough, sigh, produce smoke' and _antuwahhas_, _antuhs-_ 'human' (i.e. *h1n-dHwáh2-o:s, h1n-dHuh2-s- 'having spirit within') to the group.
Latin _fu:mus_ now appears to disobey Dybo's Law, which should have yielded **fumus with short vowel from *dHuh2-mós. De Vaan (EDL s.v.) follows Schrijver in explaining the long vowel by analogy with _fu:li:go:_ 'soot' and *fu:je-, the presumed precursor of _suf-fi:re_ 'to fumigate', both derived from protoforms with accented *dHúh2-.
Matasovic' however (EDPC 7, fn. 4) rejects analogical lengthening and considers _fu:mus_ "a clear counter-example" to Dybo's Law, accepting only that "something like [it] also operated in Italic". In Celtic, he assumes that the law operated generally, without any restrictions of phonetic environment. For the root in question, this indeed appears to be the case. Middle Irish _dúil_ 'wish, desire, yearning' (i.e. 'emotional burning') can be referred to Proto-Celtic *du:li-, from PIE *dHúh2-li-, while MIr _dumacha_ 'mist', Ir. _dumhach_ 'misty, obscure' continue PClt *duma:ka:, *duma:kos built on *dumo- from PIE *dHuh2-mó- (cf. Walde-Hofmann, LEW s.vv. _fu:li:go:_, _fu:mus_).
Scholars then and now seem to have overlooked another explanation of _fu:mus_, namely the PIE root *dHwes- 'to smoke, breathe' reflected in Lithuanian _dvê.sti_ 'to breathe one's last, give up the ghost', _dvasià_ 'spirit', and Greek _theîon_ 'sulphur' (Homeric _théeion_, probably Proto-Greek *thwéheion 'smoky substance' from *thwéhos 'smoke', PIE *dHwés-es-; cf. Beekes, EDG s.v.). Similar in sense to *dHeuh2-/*dHwah2-, *dHwes- would have produced a zero-grade derivative *dHus-mó- yielding Old Latin *fusmos 'smoke', possibly coexisting with *fumos with its short /u/ by Dybo's Law from *dHuh2-mó-. Very old Latin texts retain -sm- (e.g. _cosmis_ on the Dvenos vase, but _dismota_ on the S.C. de Bacchanalibus and _triresmos_ on the Columna Rostrata are intentional archaisms). Plautus and Naevius already have _pri:mus_ and its derivatives, not *prism-, so Plautus' _fu:mus_ could easily continue OL *fusmos. If short-vowel *fumus continued in use for a time, it never found its way into a metrical text, and there is no trace of short-vowel by-forms in Romance reflexes of _fu:mus_ and its derivatives (Meyer-Lübke, REW 3566-73).
The upshot is that, since Lat. _fu:mus_ has an alternative derivation not identical to Skt. _dhu:máh._ and Grk. _thu:mós_, and not involving a pretonic laryngeal, it does not provide "a clear counter-example" to the operation of Dybo's Law in Italic.
Douglas G. Kilday