Germanic *dail- 'portion; to divide, apportion, deal' is reflected in Runic Norse _da[i]liþun_ '(they) divided'; Gothic _dails_, Old Norse _deill_, Old English _dæ:l_, Old High German _teil_ 'portion'; etc. Pokorny (IEW 176) considered it likely that *dai-lo- 'portion' was borrowed into Germanic (following the Lautverschiebung) from Veneto-Illyrian, citing the South Illyrian personal name Daetor as a reflex of Indo-European *d@..., zero-grade of the extended root *da:-i. His form of the root was updated by Mallory-Adams (OxIntPIE 469) as *deh2/4(i)- 'to cut up, divide'. Pokorny dismissed the idea of an aspirated IE by-form *dh@... beside *d@...
However, to explain two other groups of Gmc. words, Pokorny (IEW 152-3) did postulate an IE by-form *bheugh- beside *bheug-. The first group includes Go. _us-baugjan_ 'to sweep out' and a few others against Pa:li _pari-bhuñjati_ 'cleans, sweeps out'. The second includes Go. _biugan_ 'to bend' and many others against Sanskrit _bhujáti_ 'bends', _bhugnás_ 'bent', etc. Pokorny regarded the two roots, his *bheug(h)-(2) and *bheug(h)-(3), as probably identical. This is reasonable enough, since a primary sense 'to turn' could be continued on the one hand as 'to bend', on the other 'to turn out, sweep out, cleanse, save, liberate'.
My objection is to the double standard. If an IE by-form with aspirated root-anlaut is to be rejected in favor of borrowing, then the same procedure should apply to root-auslaut. That is, rather than invoking IE *bHeugH- beside *bHeug-, we should be looking at Germanic borrowing from a language which reflected the latter as *beug-. Under his *bheug(h)-(2) Pokorny cites the Illyrian personal name Buctor along with Venetic Fuctor as equivalents of Avestan _baoxtar-_ 'liberator'. The consonantism of the presumed borrowing is thus of the Illyrian type, which along with the timing (after the Lautverschiebung) agrees with Kuhn's Nordwestblock language. (Torsten kindly made available Kuhn's most important NWB papers in the Cybalist Files.)
In addition to having /g/, OE _bu:gan_ 'to bend oneself, bow' has /u:/. This is reminiscent of _du:fan_ 'to dive', whose vocalism I attempted to explain in terms of NWB borrowing (message #65864, 16 Feb 2010). In a reply (#65686, same date) Piotr pointed out that the IE root is more likely *dHeubH- than *dHeub- (endorsed by Mallory-Adams 472, and promoted in my post). I have pasted the relevant parts of these posts for further remarks:
[DGK]
"Class II strong verbs with /u:/-presents do not form an etymologically
homogeneous group, so no simple explanation suffices for all. A thorough
treatment requires examining these verbs one at a time, which is a job for
another day. Nevertheless the fact that this subclass is significantly better
represented in Dutch than in High German, along with the peculiar consonantism
of some of the verbs, suggests that borrowing from Kuhn's Nordwestblock may be
involved in many of them. In this posting I will focus on the verb appearing in
Old English as <du:fan>. This verb and associated words illustrate the
difficulties of this subclass quite well. Furthermore, since Schrijver uses
some of the associated words as parade-examples for his Language of Geminates,
the choice of <du:fan> enables us to gauge whether Schrijver's approach or an
extension of Kuhn's approach is better suited to the problems at hand.
Old English <du:fan> 'to go underwater, dive, sink' and its weak causative
<dy:fan> 'to submerge, immerse, dip' continued into Middle English as <du:ven>
and <dy:ven>, which became confounded in the 12th century; eventually the strong
verb disappeared and the weak one replaced it as an intransitive. Modern
English <dive> has acquired a strong preterit <dove> by analogy with Class I
(<drive>, <drove>), though <dived> is still in use. Old Norse <du:fa> and
<dy:fa> correspond to OE <du:fan> and <dy:fan>, and ON <deyfa> is cited as a
synonym of <dy:fa>. In Middle Low German we have <bedu:ven> 'to be drenched
with water' and in Middle Dutch <bedu:ven> 'to sink into water', which continues
into Modern Dutch as <beduiven>.
The 'dive' group of words above is generally considered related to the 'deep'
group, which is well represented in Germanic. Here we have Gothic <diups>, ON
<dju:pr>, OE <de:op>, Old Frisian <dia:p>, Old Saxon <diop>, <diap>, and Old
High German <tiuf>, <tiof> 'deep', from which Gmc. *deupaz and Indo-European
*dHeub- are readily inferred. An /o/-grade Gmc. causative *daupjan 'to dip,
immerse' is reflected in Goth. <daupjan>, ON <deypa>, Faeroese <doypa>, OE
<di:epan>, OFrs <de:pa>, OS <do:pian>, and OHG <touffan>. Outside Germanic we
have Lithuanian <dubùs> 'deep, hollow', <daubà> 'gorge', and Old Church Slavic
<dUbrI> 'gorge'."
[Piotr]
"And Slavic *dUno < *dHub(H?)nom. Note that there is no secure evidence
for Winterian lengthening in this word-family in Balto-Slavic, which
would favour *dHeubH- over *dHeub-. Modern revisions of Kluge's Law
suggest nasal assimilation with degemination after a heavy nucleus, so
the 'deep' group could reflect something like *dHeuP-nó-, where P = any
labial, and in particular *dHeubH-nó- (a vr.ddhied adjectival derivative
of *dHubH-nó-). Parallel examples include *xWi:taz < (possibly) *k^weitnó-."
I have no objection to this, and the Gmc. adj. 'deep' should thus be *deuppa-.
[DGK]
"Phonetically, the 'dive' words could represent Gmc. *du:f- (thus Pokorny, Idg.
et. Wb., and de Vries, An. et. Wb.) or *du:b- (thus Watkins, AHD, and Köbler,
Ae. Wb.). Pokorny refers the 'dive' group to an IE root *dHeu-p- parallel to
*dHeu-b- 'deep, hollow', and suggests that *dHeu-g- 'to duck' may belong to the
same master root with a different extension. De Vries goes further and derives
*dHeu-r- 'door' from the same master root *dHeu- 'wickerwork'. The senses
'hollow, deep' are supposed to have been extracted from the characteristics of
wicker vessels. This is ingenious, but it smacks of hyperanalysis. I do not
deny that root-extensions were used in Proto-Indo-European word-formation, but
until we can deduce reliable meanings for the extensions, we are on thin ice
trying to connect 'deep' with 'door'. Moreover the extensions had not been
productive for many centuries at the time when the Germanic languages were
diverging. The real questions here are whether Common Germanic effectively had,
beside *deup-, another root *deuf- or *deub- of similar meaning, and whether
such a parallel root is responsible for the 'dive' group. In favor of IE
*dHeup- and Gmc. *deuf-, Pokorny and de Vries cite OCS <duplU> 'hollow' and
<dupina> 'hole', other Slavic words, and OHG <tobal> 'narrow valley'. This does
seem to establish IE *dHeup- in the sense of 'hollow' or the like, with a Gmc.
reflex, and one might well separate it in sense from *dHeub- 'deep'. Due to
root-restrictions, the -p- here very likely is an extension, and I have no
serious objection to -b- as another extension. But *dHeup- fails to explain the
/u:/-vocalism of the 'dive' group. If the latter comes from an IE root parallel
to *dHeub-, it should show the same vocalism. In fact the best example, ON
<deyfa>, apparently from Gmc. *daufjan, can be regarded instead as a
contamination of <deypa> (from *daupjan, from IE *dHeub-) with the -f- of
<du:fa> and <dy:fa>. Similarly OE <a:-di:efan> can be understood as a
contamination of <di:epan> with the -f- of <du:fan> and <dy:fan>."
ON _deyfa_ and OE _a:-di:efan_ can be explained better from Gmc. *daubjan, reflecting the inherited IE /o/-grade *dHoubH-.
[DGK]
"The same objection applies to Watkins's parallel IE root *dHeubH- 'deep'.
Köbler postulates a Gmc. *du:ban 'to dive', along with its causative *du:bjan,
derived from IE *dHeub- without specifying the mechanism of derivation. I think
the most plausible mechanism is the borrowing of this root into West (not
Common) Germanic from the Nordwestblock, with subsequent borrowing of *du:ban
and *du:bjan by North Gmc. from WGmc. In this view the NWBlock language merged
aspirated and unaspirated IE mediae into plain mediae, as did Celtic, Baltic,
Slavic, Messapic, and Lusitanian. As for the /u:/-vocalism, it will be noticed
that Kuhn's list of NWB protoforms with initial /p/ (Anlautend P- im
Germanischen, ZMaf 28:1-31, 1961; Kl. Schr. 1:361-389) has no examples of */eu/,
but several of */u:/, */au/, and */u/. This suggests that the IE ablaut-series
*/eu/, */ou/, */u/ ended up as */u:/, */au/, */u/ in WGmc when the forms were
borrowed from NWB rather than inherited to yield the usual */eu/, */au/, */u/.
It might be objected that a language would be unlikely to monophthongize
inherited */eu/ but not */ou/. I do not think that the NWB language did so.
Instead, I think that NWB first shifted its inherited */o/ to */a/, and along
with it */ou/ to */au/, as Germanic and Baltic also did. NWB then shifted */eu/
to a secondary */ou/, yielding the ablaut-series */ou/, */au/, */u/, and this is
what the West Gmc. dialects borrowed, probably in the 2nd and 1st centuries BCE,
before secondary Gmc. */o/ had arisen from */u/ in certain environments. Now,
by van Wijk's Law (Zur relativen Chronologie urgermanischer Lautgesetze, PBB
28:243-253, 1903), when a language has more than one high close vowel, whatever
happens to one will happen simultaneously to the other(s). In particular if
/ei/ is monophthongized to /i:/, /ou/ must simultaneously go to /u:/, which as
we know happened in pre-classical Latin. The Germanic monophthongization of
inherited */ei/ to */i:/ in stressed syllables can be regarded as tautosyllabic
umlaut, and in fact coincident with the process by which stressed */e/ became
*/i/ when followed by */i/ or */j/ in the next syllable. This process in turn
can be dated to the 1st cent. CE, since Tacitus still has forms like
<Segime:rus>. I place the WGmc monophthongization of stressed */ou/ (in NWB
loanwords) to */u:/ at the same time.
Forms with Gmc. *-pp- can be explained by Kluge's Law without invoking NWB
loanwords. They could, in principle, reflect IE *dHeub- or *dHeup-, or even
*dHeubH-. The meaning however is much more compatible with *dHeub- than with
*dHeup-, and Occam's razor shaves away *dHeubH-. These forms include *duppjan
'to immerse, dip' (OE <dyppan> 'id.', OHG <tupfen> 'to wash', Low German
<düppen> 'id.'), denominal from an abstract *duppo:-, IE *dHub-ná:-, and
*duppan- 'diving bird, dipper' (OE <-doppa>), from IE *dHub-nón- 'diver'. OE
<dy:pan> 'to dip' is apparently a cross of <dyppan> with <dy:fan>, and
<du:fedoppa> 'pelican' is probably an appositional compound of <-doppa> with a
synonymous *du:fe or *du:fa 'diver' rather than 'dove-diver, Taubentaucher',
since the pelican dives after fish, not underwater doves. This noun *du:fe or
*du:fa would continue a WGmc *du:bo:n- borrowed from NWB *doubo:n-, in turn
reflecting an IE /e/-grade *dHéub-o:n- 'diver'. Obviously the WGmc strong verb
*du:ban could not have been formed from this noun, but required a separate
borrowing from NWB *doub- 'to go deep, dive'. It is not necessary to suppose
that all three ablaut-grades of the verb were borrowed. Strong Class II already
existed with verbs having */eu/ in the present, so it is not implausible that
once the present stem *doub- had been established in WGmc (probably in the 2nd
or 1st cent. BCE), the preterit singular *daub- and plural/participal *dub-
would follow by analogy, since */ou/ would have sounded closer to */eu/ than to
other present-stem nuclei. After monophthongization of */ou/ to */u:/ (prob.
1st cent. CE), the ablaut-grades would be *du:b-, *daub-, *dub-, just what we
see in OE <du:fan>, <de:af>, <dufon>, <dofen>.
Presumably the motivation in forming the OE tautological compound <du:fedoppa>
was to avoid confusion with *du:fe 'dove', unattested in OE but represented in
ME <do(u)ve> and in other Gmc. languages. This brings up another point. This
bird-name is found in WGmc, in ON <du:fa>, and in Goth. <du:bo:>. Following
Kuhn's viewpoint in the paper cited, we may expect NWB loanwords to WGmc to be
occasionally borrowed further into NGmc, but seldom or never into Gothic. We
must thus regard *du:bo:n- 'dove' as Common Germanic. The inherited */u:/
likely represents the zero-grade */uh2/, and the IE root involved is *dHeuh2bH-.
As a rough diagnostic, then, if WGmc */u:/ agrees with Goth. /u:/, we have Gmc.
*/u:/ and no basis to suppose a NWB loanword. But if WGmc */u:/ corresponds to
Goth. /iu/ reflecting */eu/, we can at least suspect a loanword, or the
influence of one, in WGmc. For example OE <scu:fan> st. II 'to shove'
corresponds to OFrs <sku:va> and ON <sku:fa>, but Goth. <af-skiuban> (likewise
OHG <scioban> and the OE variant <sce:ofan>) requires */eu/. Here (by the
processes outlined above) IE *skeubH- became *skeub- in both Gmc. and NWB, and
the NWB present stem was borrowed into Western Common Gmc. as *skoub-, later
monophthongized to *sku:b-, which competed with the inherited Gmc. present stem
*skeub- and even spread to North Gmc., but there was no such competition in East
Gmc."
The IE root *dHeubH- gives the same NWB reflexes that *dHeub- would, and given the other evidence, the former is to be preferred. However, the theory outlined just above is inadequate to explain Go. _biugan_ and OHG _biogan_ 'to bend', with OHG _biugo_ 'sinus', which require */eu/ beside unshifted */g/. The facts call for an extension of Kuhn's theory, with the NWB language, provisionally labeled as "West Belgic", underlying the Low WGmc area, and a sister substrate, "East Belgic", underlying High WGmc and EGmc. Common Belgic had the same consonantism as Illyrian, but while East Belgic maintained the ablaut series */eu/, */au/, */u/ in common with Gmc., West Belgic/NWB shifted */eu/ to */ou/, which subsequently became */u:/ in Low WGmc borrowings.
Metal rings were used as currency, illustrated by OE _be:aggifa_ 'ring-giver, lord'. As a working hypothesis, I presume that this custom was adopted by the Germans from the Belgae along with the word *bauga- 'ring', from IE *bHóugo- 'bent object'. Another bent object, the bow, has zero-grade, Gmc. *bugo:n- (OE _boga_, etc.); presumably it originally referred to a Belgic bow, which replaced the technically inferior native Germanic type. But the Belgic abstract *buga(:) apparently meant not 'bending, flexion' but 'buying, purchase' due to its association with the currency *bauga-, and this yielded Gmc. *bugjanaN (Go. _bugjan_, etc.) 'to buy'.
DGK