¹ There may, however, be an indication of semantic fit, if we consider the semantic field that this word could have originally belonged to. It has already been noted that we consider that the base here was ‘ride.’ Now a verēdus (or a *uorēdo-) was a post-horse; that is, it was used in a sort of support or auxiliary rôle presumably to relieve other horses which had completed the prior stage of the communications or transport circuit, or to be available for hire. Perhaps we may see an extension of this semantic value of ‘auxiliary’ in the later use of the compound paraverēdus as ‘reinforcement, reserve horse,’ and the use in Old French as ‘cheval de marche’ as opposed to ‘cheval de combat.’ In other words, a verēdus may basically have been ‘a riding support.’

Such a meaning would dovetail with the morphology of other related verbal formations. OIr. rethid ‘runs’ Welsh rhedeg (in modern spelling) forms the compound fo-reith ‘helps’ (pret, furráith) Mediaeval Welsh gwa-ret (pret. gwa-rawt). The identical morphology in preverb and in preterite formation (with *-ā- vocalism) within Celtic and the exact match in preverb and semantics with Latin suc-currō make this formation one of obvious antiquity in Celtic. I believe furthermore that we may see a near synonym with identical formation in Celtic: OIr. techid ‘flees’ OWelsh tebet (Med. Welsh techaf, Bret. tec’h ‘fuite’) appears in a compound Med. Welsh go-dep ‘shelter, refuge.’ It seems to me that we see the same compound as first element in the name attested in the well known “bilingual” Ogam VOTECORIGAS = Voteporigis (see Thurneysen Grammar 571, §920), i.e. *uo-tek^wo + rīg-. Since the root *tek^w - (labio-velar established by the British Celtic forms) is broadly attested in Indo-European, this name might have meant ‘succouring king’ as well as ‘king of refuge.’ To these we then add *uo-rēdo-, We seem to have a semantic structure *uo + verb of going ‘support, help’; the pairing of ‘ride’ with ‘run, etc’ is semantically plausible (differing by just one feature) and reminiscent of Slavic ‘going (on foot); (on conveyance)’.

These verbs of going would fit in a larger picture with uo-: OIr. fo-gní ‘serves, provides’ (di-gní- ‘do, make’), fos(s)uidur ‘I support’ (:fossad ‘firm,’ Welsh gwastad), fo-sisiur ‘I acknowledge’ (: sessam ‘standing’), and fo-loi-n-g ‘supports’ (*legh- ‘lie’ → *log-- → lo-n-g-; cf. C. Watkins, Indo-European Origins of the Celtic Verb, Dublin 1962, 117, fn. 8). Perhaps fo-ling ‘anticipates’ (: lingid ‘leaps’) also belongs with the verbs of going.

An interesting divergence from the above, however, and one that is not to be brushed aside, is OIr. for·tét ‘helps,’ verbal noun fortacht (f.). This is composed of the basic, central, and highly irregular (hence old) verb ‘go’ plus the preverb *u(p)or. Obviously this involves different semantics from the other verbs just discussed; I cannot at present offer any parallels to for·tét. It is just possible that the ancestor of this compound in Gaulish was the idiom that formed the model for the revision which I propose below of *uorēdo- to verēdus.

We may also speculate that the source idioms for the structure *uo + verb (of going) ‘support, help’ were *uo + run (fast motion) and *uo + verb of position ‘support.’ This would help to account for the divergence of for-tét.

² I see that Warren Cowgill (Indo-European and Indo-Europeans, Philadelphia 1970, 117 §10) accepts the assumption that wer became *wor under the influence of its antonym wo.

³ On the phonetic explanation of guar guor- see now my forthcoming article in Studia Celtica.

⁴ The French is attested from the Chanson de Roland (1080) as palefreid, while the Old Provençal palafren has been crossed with fren ‘frein.’ The phonology of these Romance forms, with fr, points to an origin via Germanic, and not to a direct Latin ancestry. It is even more surprising that the Larousse Nouveau dictionnaire étymologique (2e édition revue et corrigée, 1970), p. 525, has Pferd as a borrowing from Lat. veredus!