From: dgkilday57
Message: 71662
Date: 2013-12-20
---In cybalist@{{emailDomain}}, <dgkilday57@...> wrote:
[...]
I believe Henning was correct in connecting Avestan _as^ti-_ 'palm as unit of measure, breadth of four fingers' with (Younger) Av. _as^ta_ 'eight'. However, I disagree with the scholars who infer that 'eight' must be the dual of an old singular meaning 'four', and who conclude that 'four' was the base of PIE counting before 'ten' was adopted.
Greek _ókris_ 'protruding point, crag, top, corner', Latin _ocris_ 'mons confragosus', Umbrian _ocar_ 'citadel', etc. are usually explained as /o/-grade derivatives of PIE *h2ek^- 'sharp' (e.g. Ionic _ákris_ 'hilltop, peak'). However, the words with /o/ could also be /e/-grade derivatives of a root *h3ek^- 'extend, protrude, point, point out'. One can point out a direction with the index finger, or at a greater distance from the seeker, with all four long fingers held together. When the four-finger pointing gesture is used, the thumb is usually folded against the palm to avoid confusion with other gestures. Hence the whole set of four fingers becomes a pointer.
Thus we could be dealing with two derivatives of *h3ek^- referring to the same thing at different times. First *h3ek^-tó- 'extended' is applied to the protruding palm, then it comes to mean 'set of four (fingers)'. Its paral *h3ek^téh3(w) 'both protruding palms' becomes the dual 'two sets of four (fingers)', works its way into the numeral system, and *h3ek^tó- itself becomes obsolete. A new noun *h3ék^-ti- 'set of pointers' is applied to the protruding palm, and this survives into Iranian as the name of a measure. But at no time is it necessary to suppose that *h3ek^t- functioned within the numeral system as 'four'.
(Since I believe Kluge's Law affected Italic as well as Germanic and Celtic (under the name Stokes' Law), I suspect Latin _occa_ 'harrow' is inherited from PIE *h3ek^-néh2, not the result of ad-hoc metathesis (thus Hirt) or ad-hoc dialectal development (thus de Vaan) for expected *ocita.)
Non-Anatolian IE *kWet-wor- 'four' likely had a similar development from a different root *kWet- 'spread, extend, protrude'. Schmid has identified this root in Lithuanian _kê.sti_ 'to spread out'. I would further consider Gaulish *pettia 'piece (of land)' to be the collective of *petto-, Celtic *kWetto-, by Stokes' Law from PIE *kWet-nó- 'spread, extended' (cf. English _spread_ 'piece of land, homestead'). Also, Hittite _kutruwan-_ 'witness' could be derived from *kWet- as 'one who points out' the guilty party or the facts of a contract. Pace Eichner, it makes little sense to derive it from 'fourth', since Anatolian uses a completely different stem for 'four', and a witness is not the fourth but the third party in a matter, as we see from Latin _testis_, Oscan _trstus_.
[Addendum: PIE *kWet- is also recognizable in Latin _triquetrus_ 'having three protrusions, triangular'. Sanskrit _caturas'ri-_ 'four-cornered' is of similar structure. Given the semantic parallelism between *kWet- and *h3ek^-, I would argue that Skt. _as'ri-_ 'corner' is probably derived from *h3ek^- and exactly cognate with Greek _ókris_ and Latin _ocris_.]
Beekes has argued that PIE *kWetwó:r underlying Gothic _fidwor_ (with Anlautwechsel resulting from counting-sandhi as explained in a previous post) and Latin _quattuor_ (with contamination from etymologically unrelated _quadrum_ 'whetstone' > 'square' to be explained in detail later) must have originally been a singular referring to "a group of four things belonging together". Thus he posited an inflected masculine sg. beside the plural forms. I find it implausible that a masc. sg. would become indeclinable in Latin, so I regard *kWetwó:r as originally neuter (or inanimate, if it was created before there was a feminine gender in Non-Anatolian IE).
The formation of this apparent compound is archaic and difficult to break down. The second element *wó:r is presumably a root-noun parallel to *dó:m 'house' (Homeric _dô_, Armenian _tun_), from the simple root *dem- 'build' without the extension *-h2. The meaning of *wó:r is perhaps 'inventory' from a simple root *wer- 'find, take' unextended by *-h1 (cf. Arm. _gerem_ 'I take (prisoner)', Grk. _heúre:ka_ 'I have found', Old Irish _-fri:th_ 'has been found'). Greek _despóte:s_ 'master' continues *dems-póti- 'master of the house', shifted from an /i/-stem to the first declension, with *déms the gen. sg. of *dó:m. Sanskrit has _pátir dán_ and _dámpatih._ where _dán_, _dám-_ can be formally understood as the endingless locative *dém of *dó:m, but since the master is not always inside the house, the endingless form is less specific than the true locative. Greek _déndre(w)on_ may contain the same preposed form giving the sense 'building tree, tree suitable for use in building' (certainly not 'tree in the house'). Since Go. _fidwor_ demands an oxytone protoform *kWetwó:r, we must assume that if it reflects the same sort of preposing as _dámpati-_ and _déndreon_, the recessive accents of the latter two words are not original.
If this assumption is made, PIE *kWetwó:r can be tentatively analyzed as *wó:r 'inventory' preposed by the combining form of *kWo:t 'extension, protrusion' vel sim., thus meaning 'inventory of protrusion, set of protruders' vel sim., which is certainly applicable to the natural group of four long fingers. In the sense 'group of four' this inanimate singular would have clashed with animate objects in apposition with it, motivating the creation of new animate plural adjectives, as well as special feminine forms involving *-srés like those for 'three'. But since the numerals from 'five' to 'ten', excepting the dual 'eight', were originally adverbial in syntax and thus indeclinable, analogical pressure would have eliminated the primary declension of 'eight' and the neuter sg. 'four', allowing the latter to survive into historical times in Latin and Gothic.
[...]