Cognations of lexical characters.

This is the basic tabulation of lexical cognations that underlies the lexical charac-
ter coding of Ringe, Warnow, and Taylor 2002. The tables on the following pages give
the cognate judgments from which the coding of the characters for input to the algorithm
was derived, with notes on how the actual coding of each character was derived.
This document has been very lightly revised. The only cognations that have been
revised are those of 29 ‘die’ and 302 ‘arrow’ (since better information about the distribu-
tion of the OE synonyms shows that those characters are polymorphic in Late West
Saxon OE), and 135 ‘say’ (whose OE polymorphism had been omitted by accident). The
consequences of those changes have not been worked out.
Since we tried to extract as much information as possible about the diversification
of the family from these data, we have not always been satisfied with root-cognations; we
also code for distinctive derivations whenever there is any indication that that might re-
veal shared history. As a result, many characters were given alternative codings for input
to the algorithm (as illustrated in Ringe, Warnow, and Taylor 2002 using the example 69
‘hand’). In addition, we have kept in mind the requirements of cladistics using character
compatibility, e.g. suppressing unique states (which are compatible with any tree).
Finally, since our algorithm could not handle polymorphic characters, we em-
ployed two strategies to reduce single polymorphic characters to sets of monomorphic
characters. If in a polymorphic character the languages exhibiting state x are a proper
subset of those exhibiting state y, the character can be duplicated and two codings adop-
ted: a “narrower” coding in which languages exhibiting both states x and y are coded
separately from those exhibiting only y, and a “broader” coding in which all languages
exhibiting state y (with or without x) are assigned the same state. We call this strategy
“split coding”. In cases where such a subset relation does not obtain this strategy is not
feasible; however, in a limited number of cases an alternative, which we call “conflated
split coding”, can be used. Suppose that language A exhibits state x, language B exhibits
states x and y, and language C exhibits only y; but whereas languages A and B share
states of many characters exclusively, thus forming an obvious subgroup, neither shares a
state of any character exclusively with C, which is thus necessarily outside that subgroup.
In such a case we can duplicate the character, coding x and x/y together and y separately
in one duplicate (thus replicating the known subgroup) and all three together in the other
duplicate without introducing spurious incompatibilities into the data. However, such
cases are disappointingly few.

September 2012
 2

Lexical characters.

The reader is reminded that the coding of lexical characters presented here allows for
alternatives in actual working code. The notes to each character indicate which alterna-
tives we have adopted and why.

1 all (pl.)
[two characters]
Hitt. 1 Av. 5a Luv. 8 Goth. 6a
Arm. 2 OCS 5b Lyc. 8 ON 6a
Gk. 3 Lith. 5b TA 3 OHG 6a
Alb. 4 OE 6a OPer. 5a Welsh 9
TB 3 OI 6b OPru. 5b Osc. 10
Ved. 5a Lat. 7 Latv. 5b Umb. 11
3 *pā́ntes 6 *ol-
5 *wi- 6a (*olnoy >) PGmc. *allai
5a (*wíḱ-wo- >) PIIr. *víćva- 6b PCelt. *olyoy
5b (*wi-so- >) PBS *visa- 8 PLuv. *pūno-
We have coded this character both by root-etymology and by derivational morphology,
on the hypothesis that there is a direct historical connection between states 5a and 5b and
likewise between states 6a and 6b.
See Hübschmann 1897:416 on the Armenian word (< *sm ̥ -) and Stang 1966:97, 238 on
the Balto-Slavic forms.

2 and
Hitt. 1 Av. 6 Luv. 1 Goth. 14
Arm. 2 OCS 7 Lyc. 1 ON 15
Gk. 3 Lith. 7 TA 11 OHG 8
Alb. 4 OE 8 OPer. 6 Welsh 6
TB 5 OI 9 OPru. 12 Osc. 16
Ved. 6 Lat. 10 Latv. 13 [loan] Umb. 10
1 PAnat. *Ho 8 *h2entí ‘in front’
6 *kwe 10 *éti ‘in addition’
7 *r̥ (particle)
On the Anatolian forms see Melchert 1992:46 with fn. 13 and references.
The second component of Welsh a ~ ag is almost certainly PIE *kwe; we are grateful to

September 2012
 3

Joseph Eska for helpful discussion of that form.

3 animal
Hitt. 1 Av. 7 Luv. 10 Goth. 8
Arm. 2 OCS 3b Lyc. 11 ON 8
Gk. 3a Lith. 3c TA 5 OHG 8
Alb. 4 OE 8 OPer. 12 Welsh 14 [loan]
TB 5 OI 9a OPru. 13 Osc. 15
Ved. 6 Lat. 9b Latv. 3d Umb. 16
3a–d derivs. of *gwih3w- ‘live’ 8 PGmc. *deuzą
5 PToch. *luwo 9a–b derivs. of *h2enh1- ‘breathe’
Since all the derivatives of ‘live’ and ‘breathe’ are different in detail (not even the closely
related Lithuanian and Latvian agree!), parallel development is overwhelmingly likely;
we therefore do not code this character by root-etymology.
Note that Arm. anasown is literally ‘non-speaker’; not to 9.

4 ashes [polymorphic, but not effectively so]

[two characters]
Hitt. 1 Av. 7 Luv. 11 Goth. 1x
Arm. 2 OCS 8a Lyc. 12 ON 1x
Gk. 3 Lith. 8b TA 5/6 OHG 1x
Alb. 4 OE 1x OPer. 13 Welsh 9
TB 5/6 OI 9 OPru. 8b Osc. 14
Ved. 1 Lat. 10 Latv. 8b Umb. 15
1 *h2eHs- 8 *pel-
1x PGmc. *askōn- ~ *azgōn- 8a (reduplicated?)
5 PToch. *tawrəә 8b PBalt. *pelenaĩ
6 PToch. *t(u)wëyë 9 PCelt. *lowtis (*lewh3- ‘wash’)
Since states 5 and 6 always cooccur, we have coded them as a single state; the polymor-
phism of this character thus becomes ineffective.
We have coded this character both by root-etymology and by derivation, since it is rea-
sonable to hypothesize that state 1x replaced state 1 directly and that there is a real his-
torical connection between states 8a and 8b.

September 2012
 4

5 at
Hitt. 1 Av. 7 Luv. 11 Goth. 9
Arm. 2 OCS 8 Lyc. 12 ON 9
Gk. 3 Lith. 8 TA 5 OHG 9
Alb. 4 OE 9 OPer. 13 Welsh 15
TB 5 OI 10 OPru. 8 Osc. 9
Ved. 6 Lat. 9 Latv. 14 Umb. 9
5 PToch. *-në 8 *prey ‘before’ 9 *ad
On the Latvian form see Endzelīns 1923:524-6.

6 back [polymorphic]
Hitt. 1 Av. 6 Luv. 13 Goth. 18
Arm. 2 OCS 7 Lyc. 14 ON 9
Gk. 3 Lith. 8 TA 15 OHG 10
Alb. 4 OE 9/10 OPer. 16 Welsh 19
TB 5 OI 11 OPru. 17 Osc. 20
Ved. 6 Lat. 12 Latv. 8 Umb. 21
6 PIIr. *pr̥šthá- 9 PNWGmc. *baką
8 PEBalt. *nugVrā or *mugVrā 10 PWGmc. *hrugi, *hrugjgj-
The polymorphism is confined to Northwest Germanic and is leaf-connected.

7 bad
Hitt. 1 Av. 6 Luv. 1 Goth. 9
Arm. 2 OCS 7 Lyc. 12 ON 17
Gk. 3 Lith. 8 TA 13 OHG 9
Alb. 4 OE 9 OPer. 14 Welsh 18
TB 1 OI 10 OPru. 15 Osc. 11
Ved. 5 Lat. 11 Latv. 16 Umb. 19
1 *édwōl 11 PItal. *mal-
9 PGmc. *ubilaz (< *h2upélos, cf. Hitt. huwappas ‘evil’ < *h2wápos, Watkins
1969:30)
On the Tocharian B word see Adams 1999 s.v. yolo with references.

September 2012
 5

8 bark
Hitt. 1 Av. 7 Luv. 13 Goth. 19
Arm. 2 [loan] OCS 8 Lyc. 14 ON 20
Gk. 3 Lith. 9 TA 15 OHG 10
Alb. 4 OE 10 OPer. 16 Welsh 21
TB 5 OI 11 OPru. 17 Osc. 22
Ved. 6 Lat. 12 Latv. 18 Umb. 23
10 PWGmc. *rindā

9 because [polymorphic, but not effectively so]

Hitt. 1a Av. 4 Luv. 1ay Goth. 3+b
Arm. 1a OCS 2x Lyc. 6 ON 3+c
Gk. 1+2 Lith. 1bx/3x TA 1+3/1az OHG 3+d+1by
Alb. 1ax OE 3+a OPer. 7 Welsh 9
TB 1+3/3y OI 5 OPru. 8 Osc. 10
Ved. 4 Lat. 1b Latv. 3x+1b Umb. 11
w w
1 derivs. of interrogative/indefinite *k i- ~ *k o-
1a *kwíd, unextended
1ax, 1ay, 1az compounded with unique elements
1b *kwód, unextended
1bx compounded with unique element
(see also below)
2 derivs. of relative *Hyo-
1+2 (phrase →) compound of *Hyód and indef. *kwid
2x compounded with unique elements
3 derivs. of demonstrative *to-
1+3 compound of *kwíd and (an innovative form of) *to-
3x PEB *tādēl
3x+1b phrase
3y PToch. *məәktë
3+a–c phrases including caseforms of *to-
3+d+1by phrase including an adverb derived from 1b
4 PIIr. *źhí
The meanings of the protoforms argue massive parallel development; clearly it would be
extremely inadvisable to code this character by root-etymology. In our working code we
have assigned separate states to all the languages except (1) Hitt. and Arm.; (2) TB and

September 2012
 6

TA (state 1+3); (3) Ved. and Av.; (4) Lith. and Latv. Note that under such a (very con-
servative) coding the polymorphisms are effectively nullified.
On the Armenian form see Clackson 1994:56, 210 fn. 98 with references.
Though the relative stem is no longer distinguishable from the pronoun *i- ~ *e- in Balto-
Slavic, it seems reasonable to assign OCS je-lĭ-ma, i-žde to the relative stem etymologic-
ally because of their meanings.
We accept the derivation of Goth. unte < *und þē, pace Feist 1939 s.v.

10 belly
[two characters]
Hitt. 1 Av. 6a Luv. 12 Goth. 9
Arm. 2 OCS 7 Lyc. 13 ON 9
Gk. 3 Lith. 8 TA 5 OHG 9
Alb. 4 OE 9 OPer. 14 Welsh 15
TB 5 OI 10 OPru. 6b Osc. 16
Ved. 6a Lat. 11 Latv. 6b Umb. 17
6 *udero- / *wēdero- s
5 PToch. *kat o
6a PIIr. *udáram 9 PGmc. *wambō
6b PBalt. *vēderas
We have coded this character both by root-etymology and by derivation, since it is rea-
sonable to hypothesize a direct historical connection between states 6a and 6b.

11 big
[two characters]
Hitt. 1 Av. 2 Luv. 7 Goth. 2x
Arm. 2 OCS 4 Lyc. 8 ON 2x
Gk. 2 Lith. 5 TA 9 OHG 2x
Alb. 2 OE 2x OPer. 10 Welsh 6
TB 3 OI 6 OPru. 11 Osc. 13
Ved. 2 Lat. 2 Latv. 12 Umb. 13
2 *meǵh2- 6 *meh2-ro-
2x PGmc. *mikilaz
13 These Osco-Umbrian comparatives and superlatives are difficult to judge.
They might reflect *meh2-is-, but analogical remodelling of *mag-is- on
full-grade *mag-yos- (cf. Lat. maior, neut. maius) is also possible. We
have therefore assigned them a separate state.

September 2012
 7

Note that the Gmc. comparatives and superlatives, which reflect *meh2-is- and would
therefore be coded 6, have been omitted.
We have coded this character both by root-etymology and by derivation, on the hypothe-
sis that state 2x replaced state 2 directly (plausible but not certain, cf. the preceding com-
ment).

12 bird
Hitt. 1 Av. 6 Luv. 11 Goth. 9
Arm. 2 OCS 7 Lyc. 12 ON 9
Gk. 3 Lith. 8 TA 5 OHG 9
Alb. 4 OE 9 OPer. 13 Welsh 10
TB 5 OI 10 OPru. 14 Osc. 15
Ved. 6 Lat. 6 Latv. 7 Umb. 6
5 phrase ‘flying animal’ 9 PGmc. *fuglaz
6 *h2éwi- ~ *h2wéy- 10 (*pet-no- >) PCelt. *etnos
7 PBS *put-
In Welsh the inherited stem has become the plural, and a singular has been backformed to
it.

13 bite
Hitt. 1 Av. 3 Luv. 10 Goth. 7
Arm. 2 OCS 6 Lyc. 11 ON 7
Gk. 3 Lith. 6 TA 12 OHG 7
Alb. 4 OE 7 OPer. 13 Welsh 15
TB 5 OI 8 OPru. 14 Osc. 16
Ved. 3 Lat. 9 Latv. 6 Umb. 17
3 *denḱ- h
7 PGmc. *bītidi (< *b eyd- ‘split’)
6 PBS *kand-

14 black [with parallel development]

Hitt. 1 Av. 7 Luv. 1 Goth. 9
Arm. 2 [loan] OCS 6 Lyc. 12 ON 9
Gk. 3 Lith. 8 TA 5 OHG 9
Alb. 4 OE 9 OPer. 13 Welsh 10
TB 5 OI 10 OPru. 6 Osc. 14
Ved. 6 Lat. 11 Latv. 3 Umb. 11

September 2012
 8

1 PAnat. *dn̥gwi- 9 PGmc. *swartaz

3 *melh2-n- 10 PCelt. *dubus
w
5 PToch. *ërk ënt- 11 PItal. *ātros
6 *kr̥snós
On the Anatolian forms see Starke 1987:263, fn. 74.
The Armenian form is an Iranian loan, pace Hübschmann 1897:489; see Olsen 1999:906
with references.
The parallel development of this character is confined to the satem group; it probably
involves semantic shift between color terms of similar meaning.

15 blood [with parallel development]

[two characters]
Hitt. 1 Av. 4 Luv. 1 Goth. 6
Arm. 1 OCS 5a Lyc. 9 ON 6
Gk. 2 Lith. 5b TA 1 OHG 6
Alb. 3 OE 6 OPer. 10 Welsh 11
TB 1 OI 7 OPru. 5b Osc. 12
Ved. 1 Lat. 8 Latv. 1 Umb. 13
1 *ésh2r̥ 5 derivs. of *krewh2- ‘gore’
6 PGmc. *blōþą ~ *blōda- 5a *kruh2-i-
5b PBalt. *krauja-
The parallel development is confined to Balto-Slavic.
We have coded this character both by root-etymology and by derivation, on the hypothe-
sis that there is a close historical connection between states 5a and 5b.

16 blow [polymorphic]
Hitt. 1 Av. 6 Luv. 10 Goth. 6
Arm. 2 OCS 6 Lyc. 11 ON 14
Gk. 3 Lith. 7 TA 5 OHG 6/14
Alb. 4 OE 6 OPer. 12 Welsh 15
TB 5 OI 8 OPru. 13 Osc. 16
Ved. 6 Lat. 9 Latv. 7 Umb. 17
5 PToch. *piya- 7 PEBalt. *putja
6 *h2wḗh1ti 14 PNWGmc. *blāsidi
The polymorphism is confined to Northwest Germanic and is leaf-connected.

September 2012
 9

17 bone [polymorphic]
[two monomorphic characters by conflated split coding]
Hitt. 1 Av. 1 Luv. 1 Goth. 10
Arm. 1 OCS 4 Lyc. 8 ON 6
Gk. 1 Lith. 5 TA 3 OHG 6
Alb. 2 OE 6 OPer. 9 Welsh 1
TB 1/3 (?) OI 7 OPru. 5 Osc. 11
Ved. 1 Lat. 1 Latv. 5 Umb. 12
1 *h2óst ~ *h2ést- 5 PBalt. *kaulan
3 PToch. *ayəә 6 PNWGmc. *bainą
The polymorphism is confined to Tocharian and is leaf-connected. If the (pre-) PToch.
singular should instead be reconstructed as *ay (the apparent *-əә being the result of
analogical changes), and if *ay can reflect *ast or *ōst by regular sound change (Katz
1997:76-7) , then 3 = 1 and the polymorphism disappears; but in that case the TA word
must be a loan from TB (Katz, loc. cit. with references) and should therefore be assigned
a separate state. In the scenario preferred here, *ayəә > TA nom./obl. sg. *e, but āyä- in
the inflected forms, by regular sound change, and from the latter āy is reintroduced into
the nom./obl. sg.
We have reduced this character to two monomorphic characters by conflated split coding,
with 3 ∪ 1/3 coded against 1 in one character and all three coded together in the other.

18 breast
[two characters]
Hitt. 1 Av. 2 Luv. 1 Goth. 7a
Arm. 2 OCS 5 Lyc. 10 ON 7a
Gk. 3 Lith. 6 TA 2 OHG 7a
Alb. 4 OE 7a OPer. 11 Welsh 7b
TB 2 OI 8 OPru. 12 Osc. 13
Ved. 2 Lat. 9 Latv. 6 Umb. 14
1 PAnat. *dēdan 7 derivs. of *bhrews-
2 *pstén- 7a PGmc. *breust- ~ *brust-
6 PEBalt. *krūtis 7b PCelt. *brusn-
We have coded this character both by root-etymology and by derivation, since it is rea-
sonable to hypothesize a direct historical connection between states 7a and 7b. (Note,
however, that the OIr. cognate means ‘belly’.)

September 2012
 10

19 breathe
Hitt. 1 Av. 5 Luv. 9 Goth. 5
Arm. 2 OCS 6a Lyc. 10 ON 5
Gk. 3 Lith. 7 TA 11 OHG 14
Alb. 4 OE 5 OPer. 12 Welsh 5
TB 5 OI 5 OPru. 13 Osc. 15
Ved. 5 Lat. 8 Latv. 6b Umb. 16
5 *h2énh1ti and derivs. 6a, 6b derivs. of *dhwes-
The Balto-Slavic forms are only distantly related: the OCS verb exhibits a process of
vowel-lengthening (*dūš- ← *duš-) by which verbs are derived from other verbs, while
the Latvian verb is transparently derived from the noun dvasa ‘breath’. Coding by root-
etymology is therefore inadvisable.
The derivatives of PIE *h2enh1- ‘breathe’ are also very diverse, but there is a reasonable
likelihood that they replaced the inherited verb without intermediaries formed to other
roots; thus we have assigned them all the same state.
This character is consequently uninformative, as there is only one shared state.

20 burn
Hitt. 1 Av. 4 Luv. 9 Goth. 6
Arm. 2 OCS 5 Lyc. 10 ON 6
Gk. 3 Lith. 4 TA 4 OHG 6
Alb. 4 OE 6 OPer. 11 Welsh 7
TB 4 OI 7 OPru. 12 Osc. 13
Ved. 4 Lat. 8 Latv. 4 Umb. 14
h wh
4 *d ég etor (with various developments of passive voice morphology)
6 PGmc. *brinnidi 7 PCelt. *losk-
The most widespread form was originally a passive, apparently supplanting an older
intransitive verb *war- (state 1).

September 2012
 11

21 child
Hitt. 1 Av. 7 Luv. 13 Goth. 10
Arm. 2 OCS 8 Lyc. 14 ON 10
Gk. 3 Lith. 9 TA 5 OHG 18
Alb. 4 OE 10 OPer. 15 Welsh 19
TB 5 OI 11 OPru. 16 Osc. 20
Ved. 6 Lat. 12 Latv. 17 [loan] Umb. 21
5 PToch. *śawmo ‘person’ and deriv. (dimin. in TB)
10 PGmc. *barną
It cannot be demonstrated that Gk. πά(ϝ)ις and Lat. puer are cognate.

22 claw
Hitt. 1 Av. 6 Luv. 8 Goth. 13
Arm. 2 OCS 3 Lyc. 9 ON 7
Gk. 3 Lith. 3 TA 10 OHG 7
Alb. 4 OE 7 OPer. 11 Welsh 14
TB 5 OI 3 OPru. 12 Osc. 15
Ved. 3 Lat. 3 Latv. 3 Umb. 16
h
3 *h3nog -(u-) with deformations and derivs.
7 PNWGmc. *klawu

September 2012
 12

23 cloud, 57 fog, AND 146 sky [polymorphic set; see notes for coding]
146 sky 23 cloud 57 fog
Hitt. 1 2 3
Arm. 4 5 6a
Gk. 7 1 6b
Alb. 8 9 6c
TB 10 11 12
Ved. 13 1/6a 6d
Av. 13/14 5/6a/15 16
OCS 1 17 6b
Lith. 18 1x 6b
OE 19 20 6e
OI 1 1y 21
Lat. 22 15 1z
Luv. 1 23 24
Lyc. 25 26 27
TA 28 [loan] 11 29
OPer. 14 30 31
OPru. 18 32 33
Latv. 1x 1xx 6b
Goth. 19 34 35
ON 19 36 37
OHG 19 20 1z
Welsh 38 39 [loan] 40 [loan]
Osc. 41 42 43
Umb. 44 45 46
h
1 *néb os ‘cloud’ and derivs. 11 PToch. *təәrkəәr ‘cloud’
1x PEBalt. *debesis 13 *dyḗws ‘sky’
(1xx cpd. of 1x) 14 *h2éḱmō ‘(sharp) stone’
1y PCelt. *neblos (?) 15 *snewdh-
1z *nebheleh2 18 PBalt. *dangus ‘sky’
5 *h3emb- 19 PGmc. *hemunaz, *himinaz ‘sky’
h
6 derivs. of *(h3)meyg - 20 PWGmc. *wolkn ‘cloud’
6a *(h3)moyghós
6b *(h3)mighleh2
6c–e other

September 2012
 13

The Irish and Luvian deformations of *nébhos are poorly understood; since they do not
seem to be significantly shared, we have coded them separately.
The polymorphism of 146 ‘sky’ is confined to Iranian and is leaf-connected. For that
character we employ conflated split coding, with 7 ∪ 6/7 coded against 6 in one character
and all three coded together in the other; 146 ‘sky’ then becomes convex (though state 1
‘cloud’ is posited—wrongly—as the ancestral state).
Both in 146 ‘sky’ and in 23 ‘cloud’ we have adopted both codings for state 1 and its
substates; thus ‘cloud’ is also coded as two characters.
For 57 ‘fog’ we have used both codings of state 6 and its substates; but it seems clear that
in the alternative in which state 6 is coded as a unit parallel development must be recog-
nized.
Otherwise the polymorphism and parallel development of this set is extensive and com-
plex; recognition of the set is not particularly helpful in reconstructing what happened,
nor in constraining the tree.

24 cold [polymorphic]
Hitt. 1 Av. 7/8 Luv. 12 Goth. 10
Arm. 2 OCS 9 Lyc. 13 ON 10
Gk. 3 Lith. 8 TA 5 OHG 10
Alb. 4 OE 10 OPer. 14 Welsh 7x
TB 5 OI 7x OPru. 8 Osc. 15
Ved. 6 Lat. 11 Latv. 7/8 Umb. 16
w
5 PToch. *k əәrośce 8 *ḱolHtos
7 derivs. of *ow- 10 PGmc. *kaldaz (< *gol-)
7x PCelt. *ougros
Since the Celtic words and the apparently related words in satem languages are actually
formed to roots of different shape (the correct analysis of the Proto-Celtic form is
*oug-ro-s), we have coded them separately.
The polymorphism is confined to the satem group and is leaf-connected.
On the difficulties besetting the Tocharian forms see Hilmarsson 1996:182-3, 194-5.
Note that the Germanic forms are not related to set 8.

September 2012
 14

25 come, 185 walk, AND 343 go [polymorphic set; see notes for coding]
25 come 343 go 185 walk
Hitt. 1s 1t 1u
Arm. 2 3 4
Gk. 1v 1 5
Alb. 6 [loan] 7 8
TB 9 1 1
Ved. 1w/9x 1/9 9
Av. 1w/9x 1/9 9
OCS 1x 1 1
Lith. 1y 1 1
OE 9 10/11/12 10/12
OI 13x 13 13
Lat. 9 1 14
Luv. 1s 1 15
Lyc. 16 17 18
TA 9 1 1
OPer. 1w/9x 1 19
OPru. 1z 1 20
Latv. 21 1 1/13y
Goth. 9 12/22 12
ON 9 11/12/22 12
OHG 9 10/11/12 10/12
Welsh 23 24 25
Osc. 9y 1 26
Umb. 9 1 14
1 *h1éyti ‘go’ 10 PWGmc. *gai- ‘go’
1s–z cpds. and (1u) deriv. 11 PGmc. *faridi ‘go, travel’
1s PAnat. *awẹ̄ti ‘come’ 12 PGmc. *gangidi ‘go’
1w PIIr. *ā́ aiti ‘come’ 13 *steygh- ‘step’
9 *gwem- ‘step’ (pres. *gwm ̥ sḱéti ‘walk’) 13x–y cpd. and deriv.
9x PIIr. *ā́ gasćati ‘come (walking)’ 14 PItal. *amf-al- ‘walk’
9y cpd. with *ḱe- ‘hither’ 22 PGmc. *līþidi ‘go, travel’
Since the compounds of the basic verbs appear to be independent formations, they are
coded separately. Note further that 13y reflects a denominative formed from a noun
which is derived from *steygh-; thus it too is coded separately.

September 2012
 15

The character 25 ‘come’ by itself is actually convex on every likely tree, but the meaning
reconstructable for state 9 argues strongly that there has been extensive parallel develop-
ment. We have adopted both codings for state 9 and its substates; in the narrower coding
the polymorphism of the character is ineffective: since states1w and 9x always cooccur,
they are coded as a single state.
In the character 343 ‘go’, the Indo-Iranian polymorphism is local and leaf-connected; so
is the Germanic polymorphism, in spite of its greater complexity.
In the character 185 ‘walk’, the Latvian polymorphism is not effective; the effective poly-
morphism is confined to West Germanic and is leaf-connected.

26 count [polymorphic]
[two monomorphic characters, in part by conflated split coding]
Hitt. 1 Av. 7 Luv. 12 Goth. 17
Arm. 2 OCS 8a Lyc. 13 ON 10
Gk. 3 Lith. 8b TA 14 OHG 10
Alb. 4 OE 9/10 OPer. 15 Welsh 9
TB 5 OI 9 OPru. 16 Osc. 18
Ved. 6 Lat. 11 Latv. 8b Umb. 19
8 diversified Balto-Slavic set 9 *rīm-
8a pre-Sl. *keit- 10 PNWGmc. *taliþi (*talja-)
8b PEBalt. *skait-
The ablaut difference of the Balto-Slavic forms suggests that the East Baltic verb was
originally denominative; its *s-, however, is puzzling. Since it is reasonable to suppose
that a denominative replaced the basic verb directly, we have employed both codings.
The polymorphism is confined to Northwest Germanic and is leaf-connected. We em-
ploy conflated split coding to construct two monomorphic characters, with 10 ∪ 9/10
coded against 9 in one character and all three coded together in the other.

September 2012
 16

27 cut [polymorphic]
Hitt. 1 Av. 5b Luv. 1/5a Goth. 13
Arm. 2 OCS 7/8 Lyc. 10 ON 5c/9/13
Gk. 3 Lith. 5b/8 TA 5ax/6 OHG 9
Alb. 4 OE 5c/9 OPer. 5b Welsh 14
TB 5ax/6 OI 7 OPru. 11 Osc. 15
Ved. 5b Lat. 7 Latv. 12 Umb. 7
1 PAnat. *kwérti 6 PToch. *latkəәna(ṣəә)
5 extensions of *ker- 7 *sek-
5a *kers- 8 PBS *rēžjeti
5ax PToch. *kəәrst-n-a(-ṣəә) 9 PGmc. *snīþidi
5b *kert- 13 PGmc. *maitidi
5c PGmc. *skiridi (*skera-)
We have coded states 5a–c and 5ax separately, but the result is still a complex instance of
polymorphism, as follows:
the Anatolian polymorphism (1/5a) is ineffective, as state 5a is unique;
the Tocharian polymorphism (5ax/6) can be coded as a single state, since its com-
ponent states always cooccur;
the Northwest Germanic polymorphism (5c/9/13) is local and leaf-connected;
but states 5b, 7, and 8 must all be assigned to Proto-Balto-Slavic, and states 5b and 7 may
need to be assigned to various nodes higher in the tree.
The most probable explanation for this pattern is that we have been unable to specify the
meaning of this character narrowly enough, and that a substantial group of near-syno-
nyms contributes states to it in a large number of languages (both attested and reconstruc-
ted).

28 day (= 24 hrs.)
[two characters]
Hitt. 1w Av. 4 Luv. 7 Goth. 5
Arm. 2 OCS 1y Lyc. 8 ON 5
Gk. 2 Lith. 1y TA 3 OHG 5
Alb. 1x OE 5 OPer. 9 Welsh 10 [loan]
TB 3 OI 6 OPru. 1y Osc. 1z
Ved. 1 Lat. 1 Latv. 1y Umb. 1z

September 2012
 17

1 *dyḗws and derivs. 2 *ā́mr̥

1w PAnat. *díwots 3 PToch. *kawnəә
1x *dit- 5 PGmc. *dagaz
1y *deyn- ~ *din-
1z POU *dyēklo-
We have employed both alternative codings, on the reasonable hypothesis that states 1w,
etc. replaced state 1 directly.

28a day (vs. night)

[two characters]
Hitt. 1w Av. 4 Luv. 7 Goth. 5
Arm. 1 OCS 1y Lyc. 8 ON 5
Gk. 2 Lith. 1y TA 3 OHG 5
Alb. 1x OE 5 OPer. 9 Welsh 1
TB 3 OI 6 OPru. 10 Osc. 11
Ved. 4 Lat. 1 Latv. 1y Umb. 12
NOTE that not all the states of this character represent the same cognate sets as the states
of the preceding; state 4, for example, does not.
1 *dyḗws and derivs. 3 PToch. *kawnəә
1w PAnat. *díwots 4 PIIr. *áźhar ~ *áźhan-
1x *dit- 5 PGmc. *dagaz
1y *deyn- ~ *din-
We have employed both alternative codings, on the reasonable hypothesis that states 1w,
etc. replaced state 1 directly.
It would have been possible to code this pair of characters as a polymorphic set, but no-
thing would be gained by such a strategy, since (a) the shared states are not distributed in
significantly different patterns in the two component characters, and (b) both characters
are monomorphic in any case.

September 2012
 18

29 die [polymorphic]
[two characters originally; shared state 11 will necessitate recoding]
Hitt. 1 Av. 2 Luv. 8 Goth. 10x/11
Arm. 2 OCS 2 Lyc. 8 ON 10
Gk. 3 Lith. 2x TA 8 OHG 6/10
Alb. 4 OE 6/11 OPer. 2 Welsh 2
TB 5 OI 7 OPru. 9 Osc. 12
Ved. 2 Lat. 2 Latv. 2x Umb. 13
2 *mer- (pres. *mr̥yétor) 10 PGmc. *dawiþi (*dauja-)
2x PEB pres. *miršta 10x denom. *dauþ-na-
6 PWGmc. *stirbidi (*sterba-) 11 PGmc. *swiltidi (*swelta-)
8 *wel-
The polymorphisms are confined to Germanic and are leaf-connected.
We originally employed both potential codings (without state 11), since direct replace-
ment of state 2 with 2x (and of 10 with unique 10x) is probable.
The Old Prussian word is not cognate with set 8, but with Lith. liáutis ‘to cease’, lavónas
‘corpse’, etc.

30 dig
Hitt. 1 Av. 6 Luv. 11 Goth. 9
Arm. 2 OCS 7 Lyc. 12 ON 9
Gk. 3 Lith. 8 TA 5 OHG 9
Alb. 4 OE 9 OPer. 6 Welsh 10
TB 5 OI 10 OPru. 13 Osc. 15
Ved. 6 Lat. 1 Latv. 14 Umb. 16
1 *bhodh2- ~ *bhedh2- 9 PGmc. *grabidi
5 PToch. *rəәpa- 10 PCelt. *klād- ~ *klad-
6 PIIr. *kánti

31 dirty
Hitt. 1 Av. 7 Luv. 13 Goth. 19
Arm. 2 OCS 8 Lyc. 14 ON 19
Gk. 3 Lith. 9 TA 15 [loan] OHG 19
Alb. 4 OE 10 OPer. 16 Welsh 20
TB 5 OI 11 OPru. 17 Osc. 21
Ved. 6 Lat. 12 Latv. 18 Umb. 22

September 2012
 19

19 PGmc. *unhrainiz ‘unclean’

Note that the Tocharian B and Old Irish forms are unrelated; see Adams 1999 s.v. sal.

32 dog
Hitt. 1 Av. 1 Luv. 1 Goth. 1x
Arm. 1 OCS 3 Lyc. 5 ON 1x
Gk. 1 Lith. 1 TA 1 OHG 1x
Alb. 2 [loan] OE 1x OPer. 6 Welsh 1
TB 1 OI 1 OPru. 1 Osc. 7
Ved. 1 Lat. 4 Latv. 1 Umb. 8
1 *ḱwō 1x PGmc. *hundaz
The Latin word is not related; cf. Welsh ceneu ‘puppy’, pl. canawon.
On the evidence for the Hittite word see Melchert 1989.
We have coded states 1 and 1x separately, simply because if they are coded together there
will be only one shared state and the character will be uninformative.

33 drink
Hitt. 1 Av. 4 Luv. 1 Goth. 6
Arm. 2 OCS 3 Lyc. 7 ON 6
Gk. 3 Lith. 5 TA 1 OHG 6
Alb. 3 OE 6 OPer. 8 Welsh 3
TB 1 OI 3 OPru. 3 Osc. 3
Ved. 3 Lat. 3 Latv. 5 Umb. 9
1 *ḗh2gwhti 5 PEBalt. *gerja
3 *peh3- ~ *pī- (pres. *píbeti)6 PGmc. *drinkidi
On the Anatolian and Tocharian forms see Kim 2000; though considerable analogical
remodelling must be posited to explain the shape of the Tocharian verb, the two do
appear to be related.

September 2012
 20

34 dry [polymorphic]
[two characters]
Hitt. 1 Av. 6x Luv. 10 Goth. 8b
Arm. 2 OCS 6 Lyc. 11 ON 8b
Gk. 3 Lith. 6 TA 5 OHG 7/8b
Alb. 4 OE 7 OPer. 6x Welsh 12 [loan]
TB 5 OI 8a OPru. 6 Osc. 13
Ved. 6x Lat. 9 Latv. 6 Umb. 14
5 PToch. *asarë 7 PWGmc. *drūg- ~ *drug-
6 *sawsos 8 derivs. of *ters- ‘be dry’
6x PIIr. *suškas 8a pre-Celt. *tērs-
8b PGmc. *þursu- ~ *þurzu-
The polymorphism is confined to West Germanic and is leaf-connected.
Given that the Germanic and Celtic states are derivatives of a verb root, it is most un-
likely that there is any direct connection between them; we have therefore coded them
separately. On the other hand, it is reasonable to suppose that state 6x replaced state 6
directly; therefore our coding employs the “superstate” including 6 and 6x, as well as
those two states.

35 dull
Hitt. 1 Av. 7 Luv. 13 Goth. 19
Arm. 2 OCS 8 Lyc. 14 ON 20
Gk. 3 Lith. 9 TA 15 OHG 20
Alb. 4 OE 10 OPer. 16 Welsh 21
TB 5 OI 11 OPru. 17 Osc. 22
Ved. 6 Lat. 12 Latv. 18 Umb. 23
20 PNWGmc. *slewaz

36 dust [polymorphic, but not effectively so]

Hitt. 1 Av. 8 Luv. 14 Goth. 19
Arm. 2 OCS 9 Lyc. 15 ON 11b
Gk. 3 Lith. 10 TA 5/6 OHG 19
Alb. 4 OE 11a OPer. 16 Welsh 20
TB 5/6 OI 12 OPru. 17 Osc. 21
Ved. 7 Lat. 13 Latv. 18 Umb. 22

September 2012
 21

5 PToch. *twëyë 11a–b NWGmc. *dunst-, *dust- (see below)

6 PToch. *tawrəә 19 PGmc. *stubjuz
The Tocharian polymorphism is ineffective: since states 5 and 6 always cooccur, they
can be coded as a single state.
It seems prudent to code the Old English and Old Norse states separately, for the follow-
ing reasons. The long vowel of OE dūst is guaranteed by Middle English spellings (the
short u of northern dusst (in the Orrmulum) and the modern word can reflect borrowing
from Norse); thus the OE word should reflect PWGmc. *dunst. ON dust reflects a pre-
form without *-n-, and since the *-n- must be part of the root, the two words exhibit dif-
ferent root-shapes. Both shapes reappear elsewhere in West Germanic (in a range of
meanings including ‘fine meal, dust,’ and the like); moreover, OHG dunst apparently
reflects a form with *þ-, so that further interaction between roots must be reckoned with.
See now Lloyd et al. 1998 s.v. dunst.

37 ear
[two characters]
Hitt. 1 Av. 5 Luv. 1 Goth. 2x
Arm. 2 OCS 2 Lyc. 6 ON 2x
Gk. 2 Lith. 2 TA 3x OHG 2x
Alb. 2 OE 2x OPer. 5 Welsh 3y
TB 3x OI 3y OPru. 2 Osc. 7
Ved. 4 Lat. 2 Latv. 2 Umb. 8
1 *stómn̥ ~ *stm
̥ én- (‘ear’ ?) 3 derivs. of *ḱlew- ‘hear’
2 *h2éwsos 3x PToch. *klëwtso
2x PGmc. *ausōn- ~ *auzōn- 3y PCelt. *klowstā
5 PIr. *gaušah
We employ both codings for superstate 2, but since the two derivatives of ‘hear’ are
clearly independent we have coded them separately.

September 2012
 22

38 earth
[two characters]
Hitt. 1 Av. 1 Luv. 1 Goth. 4
Arm. 2 OCS 1x Lyc. 7 ON 4
Gk. 3 Lith. 1x TA 1 OHG 4
Alb. 1 OE 4 OPer. 8 Welsh 9
TB 1 OI 5 OPru. 1x Osc. 6
Ved. 1 Lat. 6 Latv. 1x Umb. 10
1 *d éǵ ōm, *ǵ m-, loc. *dhǵhém
h h h 4 PGmc. *erþō
1x PBS *žemjā 6 PItal. *tersā
We employ both codings for superstate 1.

39 eat [polymorphic]
Hitt. 1 Av. 4 Luv. 1 Goth. 1
Arm. 1 OCS 1 Lyc. 7 ON 1
Gk. 1 Lith. 5 TA 3 OHG 1
Alb. 2 OE 1 OPer. 8 Welsh 9
TB 3 OI 6 OPru. 1/5 Osc. 1
Ved. 1 Lat. 1 Latv. 1 Umb. 10
1 *h1ḗdsti 5 PBalt. *valgā
3 PToch. *śuwa(ṣəә)
The polymorphism is confined to Baltic and is leaf-connected.

40 egg
Hitt. 1 Av. 7 Luv. 10 Goth. 16
Arm. 2 OCS 3 Lyc. 11 ON 3
Gk. 3 Lith. 8 TA 12 OHG 3
Alb. 4 OE 3 OPer. 13 Welsh 3
TB 5 OI 9 OPru. 14 Osc. 17
Ved. 6 Lat. 3 Latv. 15 Umb. 18
3 *(h2)ōwióm
On the Welsh and Irish forms cf. Lewis and Pedersen 1961:14, 29.

September 2012
 23

41 eye
[two characters]
Hitt. 1a Av. 4 Luv. 1a Goth. 2x
Arm. 2 OCS 2 Lyc. 1a ON 2x
Gk. 2 Lith. 2 TA 2 OHG 2x
Alb. 3 OE 2x OPer. 4 Welsh 5
TB 2 OI 1b OPru. 2 Osc. 6
Ved. 2 Lat. 2 Latv. 2 Umb. 7
w
1 derivs. of *sek - ‘see’ 2 *h2ók w

1a PAnat. *sógwo- 2x PGmc. *augōn-

w
1b PCelt. *sok lis 4 PIr. *čašma
It is most unlikely that there is any direct connection between states 1a and 1b; coding by
root-etymology is therefore inadvisable. On the other hand, it is very likely that state 2x
replaced state 2 directly (since the Germanic word for ‘eye’ has clearly been deformed by
lexical analogy with ‘ear’); we have therefore employed both codings in that case.

September 2012
 24

42 fall, 53 float, 54 flow, 56 fly, AND 166 swim [polymorphic set; see notes for coding]
42 fall 56 fly 54 flow 53 float 166 swim
Hitt. 1 2 3 4 5
Arm. 6 7 8 9 9
Gk. 2 2 10 9 11
Alb. 12 13 14 9 15 [loan]
TB 16 9 17 9 11
Ved. 18 2 10/19 9 9
Av. 2 2/9 19 20 21
OCS 18 22 23 9 9
Lith. 24 25 23 9x/9y 9x
OE 26 9x 9z 9y 27
OI 28 9 29 11 11
Lat. 30 31 32 32 11
Luv. 33 34 3 35 36
Lyc. 37 38 39 40 41
TA 16 9 42 43 44
OPer. 45 46 47 48 49
OPru. 50 51 52 53 54
Latv. 24 55 23/9y 9y 56
Goth. 57 58 59 60 61
ON 26 9x 9y/59 9y 27
OHG 26 9x 9y/59 9y 27
Welsh 62 2 63 11 11
Osc. 64 65 66 67 68
Umb. 69 70 71 72 73
2 *pet(h2)- ‘fly’ 18 *ped- ‘fall’
3 *érsti ‘flow’ 19 PIIr. *ghžárati ‘flow’
9 *pléweti ‘float’ and derivs. 23 *tékwti ‘run’
9x *plew-k- 24 PEBalt. *kr__t- (?) ‘fall’
9y *plew-d- 26 PNWGmc. *fallidi ‘fall’
9z *plōw- (*pleh3-?) 27 PNWGmc. *swimmidi ‘swim’
10 *sréweti ‘flow’ 32 Latin flu- ‘flow’
11 *snéh2ti ‘bathe’ and derivs. 59 PGmc. *rinnidi ‘run’
16 PToch. *klowyotəәr ‘fall’
Though state 32 recurs, it is confined to a single language (Latin).

September 2012
 25

The East Baltic words assigned to state 24 are apparently connected, though the root syl-
lable has been distorted in one language or the other, so that it cannot be fully reconstruc-
ted. The Old Prussian root, however, is krū- (the -t is the infinitive ending), which ap-
pears to be different.
On the preform of state 9z see Rix et al. 2001 s.v. *pleh3-.
In 53 ‘float’ we have employed both codings for state 9 and its substates; since 9x is
unique, both the resulting characters are effectively monomorphic. In 54 ‘flow’ and 56
‘fly’ we have coded the substates separately, since those characters are polymorphic in
any case. In 166 ‘swim’ we have coded states 9 and 9x together, since the latter is
unique.
Even the members of this set which are not polymorphic exhibit extensive parallel devel-
opment. On the other hand, analysis of that phenomenon as borrowing of states between
different characters of the set gives an unusually “clean” and unproblematic picture of the
characters’ development.

43 far
[two characters]
Hitt. 1 Av. 1x Luv. 8 Goth. 2c
Arm. 2a OCS 5 Lyc. 9 ON 2c
Gk. 2b Lith. 6 TA 4 OHG 2c
Alb. 3 [loan] OE 2c OPer. 1x Welsh 11
TB 4 OI 7 OPru. 10 Osc. 12
Ved. 1x Lat. 2d Latv. 6 Umb. 13
1 *dweh2- 4 PToch. *law, *lawkë
1x *duh2ró- 6 PEBalt. *tāl-
2 derivs. of *per-
2a *per-s- (*per-n- ?) 2c PGmc. *ferr-
2b deriv. of *próti 2d (other deriv. of *pró)
States 2a–d are only distantly related. Attic Greek πόρρω also ‘further’, reflects πόρσω
< πρόσω (attested in other dialects), a derivative of πρός ‘towards’ < *próti, whose con-
nection with the other forms is remote. Armenian ṙ should reflect *rs (or *rn?; see Olsen
1999:435), yet in Germanic we find neither *rs nor *rz (nor *rn) but a puzzling *rr. We
have therefore coded all these states separately.
On the other hand, it is reasonable to suppose that state 1x replaced state 1 directly; for
that pair of states we have therefore used both codings.

September 2012
 26

44 fat
Hitt. 1 Av. 6a Luv. 11 Goth. 14
Arm. 2 [loan] OCS 7 Lyc. 12 ON 8
Gk. 3 Lith. 7 TA 5 OHG 6b
Alb. 4 [loan] OE 8 OPer. 13 Welsh 15
TB 5 OI 9 [loan] OPru. 7 Osc. 16
Ved. 6a Lat. 10 Latv. 7 Umb. 10
5 PToch. *ṣəәlypyë 7 PBS *taukas
6 derivs. of *peyH- ‘fat’ 8 PNWGmc. *smerwą
6a PIIr. *pī́vas 10 PItal. *adep-
6b PWGmc. *faitid-
On the Celtic words see Pedersen 1909:22-4.
States 6a, 6b are very distantly related; we have therefore coded them separately.

45 father [with parallel development]

Hitt. 1 Av. 2 Luv. 4 Goth. 1
Arm. 2 OCS 1 Lyc. 4 ON 2
Gk. 2 Lith. 3 TA 2 OHG 2
Alb. 1 OE 2 OPer. 2 Welsh 5
TB 2 OI 2 OPru. 3 Osc. 2
Ved. 2 Lat. 2 Latv. 3 Umb. 2
1 *átta ‘dad’ 3 PBalt. *tēvas
2 *ph2tḗr 4 PAnat. *dáda ‘dad’

46 fear [polymorphic]
[two monomorphic characters, in part by conflated split coding]
Hitt. 1 Av. 5a Luv. 8 Goth. 4b/6
Arm. 2 OCS 5b Lyc. 9 ON 11
Gk. 2 Lith. 5c TA 4a OHG 4b
Alb. 3 OE 4b OPer. 10 Welsh 12
TB 4a OI 6 OPru. 5c Osc. 13
Ved. 5a Lat. 7 Latv. 5d Umb. 14

September 2012
 27

2 *dwey- 5 satem *bheyH-

4 *prek- 5a PIIr. perf. *bhebhóy(H)e
4a PToch. *praska- ~ *pəәrska- 5c PBalt. *bijā
←< pres. *pr̥(k)-sḱé/ó- 5b, 5d (other derivs.)
4b derivs. of PGmc. *furhtaz 6 perf. *h2eh2óghe ‘be upset’
‘fearful’ < adj. *pr̥któs
There is a high probability that the connection between states 4a and 4b is indirect; we
have therefore coded them separately. On the other hand, it is reasonable to hypothesize
a direct historical connection between states 5a–d; we have therefore employed both
alternative codings for those states.
The polymorphism is confined to Germanic and is leaf-connected. We have employed
conflated split coding to recode this character as two monomorphic characters, with 4b ∪
4b/6 coded against 6 in one character and all three coded together in the other. In the
latter character we have coded superstate 5 as a unit.

47 feather
[two characters]
Hitt. 1 Av. 3z Luv. 8 Goth. 14
Arm. 2 OCS 3 Lyc. 9 ON 6a
Gk. 3x Lith. 5 TA 10 OHG 6a
Alb. 4 [loan] OE 6a OPer. 11 Welsh 15 [loan]
TB 3y OI 7 OPru. 12 Osc. 16
Ved. 3z Lat. 6b Latv. 13 Umb. 17
3 *peróm 6 derivs. of *pet- ‘fly’
3x → *pteróm by lexical analogy 6a PNWGmc. *feþru (as if)
with *pet- ~ *pt- ‘fly’ < *petreh2
3y remodelled as u-stem 6b *petneh2
3z *pernóm
6a and 6b appear to be independent derivations; we have therefore coded them separately.
On the other hand, we have employed both codings for superstate 3, since direct histori-
cal connection between states 3, 3x, etc. appears probable.

September 2012
 28

48 few
[two characters]
Hitt. 1 Av. 7 Luv. 12 Goth. 10a
Arm. 2 OCS 8 Lyc. 13 ON 10a
Gk. 3 Lith. 9 TA 14 OHG 10a
Alb. 4 [loan] OE 10a OPer. 7 Welsh 16
TB 5 OI 11 OPru. 15 Osc. 17
Ved. 6 Lat. 10b Latv. 9 Umb. 18
7 PIr. *kamna- 10a PGmc. *fawai < *paw-o-
9 PEBalt. *mažai 10b *paw-ko-
The -h- of OHG dat. pl. fōh- is probably only graphic (pace Braune and Reiffenstein
2004: 225).
We have adopted both alternative codings for states 10a, 10b, since it is reasonable to
hypothesize a direct connection between them.

49 fight [polymorphic]
[two characters by conflated split coding; both still polymorphic]
Hitt. 1 Av. 6/7 Luv. 13 Goth. 10
Arm. 2 [loan] OCS 7/8 Lyc. 14 ON 8/10
Gk. 3 Lith. 9 TA 15 OHG 11
Alb. 4 OE 10/11 OPer. 16 Welsh 18
TB 5 OI 10 OPru. 17 Osc. 19
Ved. 6 Lat. 12 Latv. 9 Umb. 20
6 PIIr. *yúdhyati 9 derivs. of *kaw- ‘chop’
7 derivs. of *per- ‘beat’ 10 *wikéti
8 *bhorH- ‘beat’ 11 PWGmc. *fihtidi (*fehta-)
The present stems of state 8 do not match perfectly; see Rix et al. 2001 s.v. *bherH- for
discussion.
There are multiple polymorphisms, some extensive, but all are leaf-connected. It is pos-
sible that state 7 should be split, but that will not reduce the polymorphism by much. We
have constructed two characters by conflated split coding, with 6 ∪ 6/7 coded against 7
in one character and all three coded together in the other; but that does not eliminate all
the polymorphism.

September 2012
 29

50 fire [with parallel development]

Hitt. 1 Av. 4 Luv. 1 Goth. 1
Arm. 1 OCS 3 Lyc. 6 ON 7
Gk. 1 Lith. 3 TA 1 OHG 1
Alb. 2 OE 1 OPer. 4 Welsh 5
TB 1 OI 5 OPru. 1 Osc. 8
Ved. 3 Lat. 3 Latv. 3 Umb. 1
1 *péh2wr̥ 4 PIr. *ātr-
3 *Vgnís ‘fire-god’ 5 PCelt. *tenet-

51 fish
[two characters]
Hitt. 1 Av. 5 Luv. 8 Goth. 7a
Arm. 2 OCS 6 Lyc. 9 ON 7a
Gk. 2 Lith. 2 TA 10 OHG 7a
Alb. 3 [loan] OE 7a OPer. 11 Welsh 12 [loan]
TB 4 OI 7b OPru. 2 Osc. 13
Ved. 5 Lat. 7c Latv. 2 Umb. 14
2 *dhǵhúHs 7a PGmc. *fiskaz < *piskos
5 PIIr. *mátsyas 7b *peyskos
7c *piskis
We have employed both alternative codings for states 7a–c, since a direct historical
connection between them seems likely.

52 five
Hitt. 1 Av. 2 Luv. 3 Goth. 2x
Arm. 2 OCS 2 Lyc. 4 ON 2x
Gk. 2 Lith. 2 TA 2 OHG 2x
Alb. 2 OE 2x OPer. 5 Welsh 2
TB 2 OI 2 OPru. 2 Osc. 2
Ved. 2 Lat. 2 Latv. 2 Umb. 2
2 *pénkwe 2x PGmc. *fimf with unexpected *-f
We have coded states 2 and 2x separately, simply because if they are coded together there
will be only one shared state and the character will be uninformative.

September 2012
 30

53 float
Forms a polymorphic set with 42 fall (q.v.), etc.
54 flow
Forms a polymorphic set with 42 fall (q.v.), etc.

55 flower
Hitt. 1 Av. 7 Luv. 11 Goth. 10a
Arm. 2 OCS 8 Lyc. 12 ON 10a
Gk. 3 Lith. 9 TA 5 OHG 10a
Alb. 4 OE 10a OPer. 13 Welsh 10b
TB 5 OI 10b OPru. 14 Osc. 10c
Ved. 6 Lat. 10c Latv. 15 Umb. 16
5 PToch. *pyapyo 10a PGmc. *blōm- (remodelled in OE?)
10 derivs. of *bhloH- 10b PCelt. *blāto-
10c PItal. *flōs
Since states 10a–c can clearly be independent derivatives of a verb ‘bloom’, we have
coded them separately.

56 fly (vb.)
Forms a polymorphic set with 42 fall (q.v.), etc.
57 fog
Forms a polymorphic set with 23 cloud (q.v.) and 146 sky.

58 foot
Hitt. 1 Av. 1 Luv. 1 Goth. 1
Arm. 1 OCS 3 Lyc. 1 ON 1
Gk. 1 Lith. 4 TA 1 OHG 1
Alb. 2 OE 1 OPer. 1 Welsh 5
TB 1 OI 5 OPru. 3 Osc. 1
Ved. 1 Lat. 1 Latv. 4 Umb. 1
1 *pṓds ~ *pód- ~ *ped- 4 PEBalt. *kājā
3 PBS *nagā 5 PCelt. *traget-

September 2012
 31

59 four
[two characters]
Hitt. 1 Av. 2 Luv. 1 Goth. 2x
Arm. 2 OCS 2 Lyc. 3 ON 2x
Gk. 2 Lith. 2 TA 2 OHG 2x
Alb. 2 OE 2x OPer. 4 Welsh 2
TB 2 OI 2 OPru. 2 Osc. 2
Ved. 2 Lat. 2 Latv. 2 Umb. 2
1 PAnat. *mǣu- w w w
2 *k etwóres, fem. *k étesres, neut. *k etwṓr
2x PGmc. *fedwōr with unexpected *f-
We use both codings, since a direct development of 2 to 2x is unarguable.

60 freeze
Hitt. 1 Av. 7 Luv. 13 Goth. 10
Arm. 2 OCS 8 Lyc. 14 ON 10
Gk. 3 Lith. 9 TA 15 OHG 10
Alb. 4 OE 10 OPer. 16 Welsh 10
TB 5 OI 11 OPru. 17 Osc. 18
Ved. 6 Lat. 12 Latv. 9 Umb. 19
9 PEBalt. *šal- 10 *prews- ‘burn’

61 fruit
Hitt. 1 Av. 7 Luv. 13 Goth. 19
Arm. 2 OCS 8 [loan] Lyc. 14 ON 20
Gk. 3 Lith. 9 TA 15 [loan] OHG 10
Alb. 4 [loan] OE 10 OPer. 16 Welsh 21 [loan]
TB 5 OI 11 OPru. 17 Osc. 22
Ved. 6 Lat. 12 Latv. 18 Umb. 23
10 PWGmc. *obat

September 2012
 32

62 full
[two characters]
Hitt. 1 Av. 2d Luv. 4 Goth. 2d
Arm. 2a OCS 2d Lyc. 5 ON 2d
Gk. 2b Lith. 2d TA 3 OHG 2d
Alb. 2c OE 2d OPer. 6 Welsh 2d
TB 3 OI 2d OPru. 2d Osc. 7
Ved. 2d Lat. 2dx Latv. 2d Umb. 2dx
2a–d derivs. of *pleh1- ‘fill’
2d *pl̥h1nós
2dx *plēnos ←< *pl̥h1nós by analogy with vb. *plē- < *pleh1-
3 PToch. *wyitë
On the Tocharian words see Adams 1999 s.v. ite with references.
We have coded states of superstate 2a–d separately, since they appear to be independent
derivatives of the basic verb; however, since 2dx clearly developed from 2d, we employ
both codings for superstate 2d.

63 give
[two characters]
Hitt. 1x Av. 2b Luv. 1x Goth. 4
Arm. 2a OCS 2bx Lyc. 1x ON 4
Gk. 2b Lith. 2bx TA 1 OHG 4
Alb. 3 OE 4 OPer. 2b Welsh 2c
TB 1 OI 5 OPru. 2bx Osc. 2b
Ved. 2b Lat. 2b Latv. 2bx Umb. 2b
1 *ay-
1x PAnat. *p-ay-
2 *deh3-
2a, 2c original pres. unclear
2b pres. *dédeh3ti and developments of same
2bx PBS pres. *dōd- (apparently ← *ded4- < *dédeh3-, but how?)
4 PGmc. *gibidi (*geba-)
The reduplicating syllable *de- was replaced by the productive *di- in Greek and Italic
(an unremarkable parallel development); in Osco-Umbrian the stem was thematized, but
in Latin the reduplication was lost by sound change in compounds and the dereduplicated
form was then generalized to the simplex (see e.g. Sommer 1948:538-9).

September 2012
 33

In Celtic and Iranian this verb was confused with *dheh1- ‘put’ because the two had be-
come very similar by sound change.
We have employed both alternative codings.

64 good
Hitt. 1 Av. 6 Luv. 6 Goth. 9
Arm. 2 OCS 7 Lyc. 12 ON 9
Gk. 3 Lith. 8 TA 5 OHG 9
Alb. 4 OE 9 OPer. 6 Welsh 14
TB 5 OI 10 OPru. 13 Osc. 15
Ved. 6 Lat. 11 Latv. 13 Umb. 16
5 PToch. *krënt- 9 PGmc. *gōdaz
6 *wḗsus ~ *wésu- 13 PBalt. *labas

65 grass
Hitt. 1 Av. 7 Luv. 13 Goth. 17
Arm. 2 OCS 8 Lyc. 14 ON 10
Gk. 3 Lith. 9 TA 5 OHG 10
Alb. 4 OE 10 OPer. 15 Welsh 1
TB 5 OI 11 OPru. 16 Osc. 18
Ved. 6 Lat. 12 Latv. 9 Umb. 19
1 *wel- 9 PEBalt. *žālē
5 PToch. *atiya- 10 PGmc. *grasą ‘plant’

66 green
[two characters]
Hitt. 1 Av. 6 Luv. 1 Goth. 13
Arm. 2 OCS 3 Lyc. 10 ON 7
Gk. 3 Lith. 3x TA 11 [loan] OHG 7
Alb. 4 [loan] OE 7 OPer. 12 Welsh 14
TB 5 OI 8 OPru. 3x Osc. 15
Ved. 3 Lat. 9 Latv. 3x Umb. 16
1 PAnat. *HaHl̥went- 3 derivs. of *ǵhelh3-
7 PNWGmc. *grōniz 3x PBalt. *žaljas
The Welsh word might conceivably belong under 3 (as “3y”), but the phonology is diffi-
cult.

September 2012
 34

We employ both alternative codings, since a direct connection between state 3x and the
other derivatives of *ǵhelh3- is likely.

67 guts
Hitt. 1 Av. 5 Luv. 10 Goth. 16
Arm. 2 OCS 6 Lyc. 11 ON 8
Gk. 2 Lith. 7 TA 12 OHG 8
Alb. 3 OE 8 OPer. 13 Welsh 17
TB 4 OI 9 OPru. 14 Osc. 18
Ved. 2 Lat. 2 Latv. 15 Umb. 19
2 derivs. of *en-tero- ‘inside’ 8 PGmc. *þarmaz (< *tórmos ‘hole’,
root *terh1- ‘bore’)

68 hair [polymorphic]
Hitt. 1 Av. 7a Luv. 11 Goth. 17
Arm. 2 OCS 7a Lyc. 12 ON 9/17
Gk. 3 Lith. 8 TA 13 OHG 9
Alb. 4 OE 9 OPer. 14 Welsh 7b
TB 5 OI 7b OPru. 15 Osc. 18
Ved. 6 Lat. 10 Latv. 16 Umb. 19
7 derivs. of *wel- 9 PNWGmc. *hārą
7a satem *wolḱos 17 PGmc. *skuftą
7b PCelt. *woltos
The relation between states 7a and 7b is remote; we have coded them separately.
The polymorphism is confined to Northwest Germanic and is leaf-connected.
Note that there is no connection between the Tocharian B and Latvian words, nor be-
tween Armenian her and the superficially similar Germanic word.

69 hand
[two characters]
Hitt. 1 Av. 1x Luv. 1 Goth. 3
Arm. 1 OCS 2 Lyc. 1 ON 3
Gk. 1 Lith. 2 TA 1 OHG 3
Alb. 1 OE 3 OPer. 1x Welsh 4
TB 1 OI 4 OPru. 2 Osc. 5
Ved. 1x Lat. 5 Latv. 2 Umb. 5

September 2012
 35

1 *ǵhésōr 3 PGmc. *handuz

1x PIIr. *źhástas < 4 PCelt. *lāmā (< *pĺ̥h2meh2
*ǵhéstos (remodelled, ‘palm’)
but how?) 5 PItal. *man-
2 PBS *rankā
We employ both codings, since it is likely that state 1x replaced state 1 directly.

70 he, 168 that, 170 they, AND 174 this [polymorphic set]
he that they this
Hitt. 1 7 1 6
Arm. 2 2 2 6
Gk. 3 9 3 4
Alb. 4 4 4 14
TB 4 4 4 4
Ved. 4 4 4 1
Av. 4 4 4 1
OCS 1/5 1/4/5 1/5 6/12
Lith. 1 4 1 6
OE 6 4 6 4
OI 1 4 1 4
Lat. 1 10 1 15
Luv. 1/7 7 1/7 6
Lyc. 1/7 11 1/7 7
TA 4 4 4 4
OPer. 4 12 13 1/4
OPru. 4+5 4 4+5 6
Latv. 8 4 8 6
Goth. 1 4 1 4
ON 6 4 4 4
OHG 1 4 4 4
Welsh 4 4 4 4
Osc. 1 10 1 1
Umb. 1 10 1 1
1 *éy, fem. *íh2, neut. *íd, adj. *ós etc. and derivs.
4 *só, fem. *séh2, neut. *tód ‘that’ and derivs.
5 *eno- ~ *ono- ‘that’

September 2012
 36

6 *ḱéy, adj. *ḱós, etc. ‘this’ and derivs.

7 PAnat. *obós ‘that’
10 PItal. *ollo- ‘that’
12 *awo-
The following states, though appearing more than once in the table, are confined to par-
ticular languages:
2 Arm. forms with the deictic particle -n-
3 Gk. αὐτό-
8 Latv. *vinja-
We have cautiously coded Gk. ἐκεῖνος ‘that’ separately from state 6.
Finally, note that Lat. hic ‘this’ (state 15) is coded according to its stem, not according to
the particle -c(e) (which is etymologically connected with state 6).
On the Old Prussian state 4+5 see Stang 1966:235-6.

71 head
Hitt. 1 Av. 6 Luv. 11 Goth. 8
Arm. 2 OCS 7 Lyc. 12 ON 10
Gk. 3 Lith. 7 TA 13 OHG 8
Alb. 4 OE 8 OPer. 14 Welsh 9
TB 5 OI 9 OPru. 7 Osc. 15
Ved. 6 Lat. 10 Latv. 7 Umb. 16
6 PIIr. *ćŕ̥Has < *ḱr̥h2- 9 PCelt. *kwennas
7 PBS *galvā 10 *kaput
8 PGmc. *haubidą
There is probably some relation between states 8 and 10, though its exact nature is ob-
scure; see e.g. Feist 1939 s.v. haubiþ with references. If 8 and 10 are coded together, the
nonconvexity of this character on every likely tree is ameliorated.

72 hear [with parallel development]

[two characters]
Hitt. 1 Av. 2 Luv. 1 Goth. 3
Arm. 2 OCS 2y Lyc. 7 ON 3
Gk. 3 Lith. 5 TA 2x OHG 3
Alb. 4 OE 3 OPer. 8 Welsh 2
TB 2x OI 2 OPru. 5 Osc. 9
Ved. 2 Lat. 6 Latv. 5 Umb. 10

September 2012
 37

1 PAnat. deriv. of *stómn̥ 3 *h2ḱ-h2ows-ié-ti ‘be sharp-eared’

2 *ḱlew- ‘hear’ (pres. *ḱl̥néwti) 5 PBalt. *girdi ‘hear’
y
2x PToch. *kl ews-
2y other s-extension (orig. desiderative ‘listen’?)
We employ both potential codings, since the shapes of the PToch. and OCS verbs are
innovative, but a direct connection with the other forms of superstate 2 is likely. The
root-ablaut of the OCS verb betrays a complex prehistory, not necessarily connected in
any way with the Tocharian verb.
The meaning reconstructable for state 3 strongly argues parallel development.
The consistent k- of the Old Prussian form is puzzling.

73 heart
[two characters]
Hitt. 1 Av. 1x Luv. 1 Goth. 1y
Arm. 1 OCS 1 Lyc. 4 ON 1y
Gk. 1 Lith. 1 TA 3 OHG 1y
Alb. 2 OE 1y OPer. 5 Welsh 6
TB 3 OI 1 OPru. 1 Osc. 7
Ved. 1x Lat. 1 Latv. 1 Umb. 8
1 *ḱḗr, *ḱr̥d- and derivs.
1x init. cons. replaced by unexpected PIIr. *źh-
1y PGmc. *hertōn-
3 PToch. *arəәñce
We employ both alternative codings.

74 heavy [polymorphic]
[two monomorphic characters by conflated split coding;
parellel development still present]
Hitt. 1 Av. 3 Luv. 1 Goth. 3
Arm. 2 OCS 5 Lyc. 9 ON 7
Gk. 3 Lith. 6 TA 3x OHG 1/7
Alb. 4 OE 1/7 OPer. 10 Welsh 8
TB 3x OI 8 OPru. 11 Osc. 13
Ved. 3 Lat. 3 Latv. 12 Umb. 14

September 2012
 38

1 *sworu- ~ *swēru- 7 PNWGmc. *hafīgaz, *hafugaz

3 *gwréh2u- ~ *gwr̥h2éw- and derivs. 8 PCelt. *trummos
w
3x deriv. of PToch. *k raməәr
‘burden’ < *gwróh2mr̥
Since the connection of state 3x with state 3 is patently indirect, we have coded them sep-
arately.
We have reduced this to two monomorphic characters using conflated split coding, with 7
∪ 1/7 coded against 1 in one character and all three coded together in the other.
Although the overt polymorphism is confined to West Germanic, both state 1 and state 3
must be posited for large numbers of internal nodes in any likely tree.

75 here AND 169 there [polymorphic set]

Hitt. 1, 2 Av. 7, 8 Luv. 1, 11 Goth. 1, 6
Arm. 1, 3 OCS 1, 6 Lyc. 2, 12 ON 1, 6
Gk. 4, 5 Lith. 6, 6 TA 6, 6 OHG 1, 6
Alb. 6, 6 OE 1, 6 OPer. 7, 13 Welsh 14, 15
TB 6, 6 OI 6, 9 OPru. 6, 6 Osc. 16, 7
Ved. 7, 6 Lat. 10, 7 Latv. 6, 6 Umb. 17, 7
1 derivs. of *ḱe- ~ *ḱi- ~ *ḱo- ‘this’ 6 derivs. of *só- ~ *tó- ‘that’
2 derivs. of PAnat. *obós ‘that’ 7 *idhé(y)
We have cautiously coded Gk. ἐκεῖ ‘there’ separately from state 1.
For the (often surprising) etymologies of the other forms the standard comparative gram-
mars should be consulted
Finally, note that Lat. hīc ‘here’ (state 10) is coded according to its stem, not according to
the particle -c(e) (which is etymologically connected with state 1).

76 hit [polymorphic]
[two monomorphic characters by conflated split coding]
Hitt. 1 Av. 6 Luv. 12 Goth. 9
Arm. 2 OCS 7 Lyc. 12 ON 9/11
Gk. 3 Lith. 8 TA 5 OHG 9
Alb. 4 OE 9 OPer. 6 Welsh 15
TB 5 OI 10 OPru. 13 Osc. 16
Ved. 6 Lat. 11 Latv. 14 Umb. 17

September 2012
 39

5 PToch. *kəәrnəәsk- 11 *bher-

6 *gwhénti 12 PLuv. *tūbīdi
9 PG *slahidi
Note that there is no connection between the Proto-Luvian form and Gk. τύπτει; see
Melchert 1994:242.
The polymorphism is confined to Germanic and is leaf-connected. We have constructed
two monomorphic characters by conflated split coding, with 9 ∪ 9/11 coded against 11 in
one character and all three coded together in the other.

77 hold AND 350 have [polymorphic set]

Hitt. 1, 1 Av. 8, 10 Luv. 20, 21 Goth. 16, 16/28
Arm. 2, 2 OCS 11, 12 Lyc. 22, 23 ON 15, 16/28
Gk. 3, 3 Lith. 13, 14 TA 24, 25 OHG 16, 16
Alb. 4, 5 OE 15, 16 OPer. 8, 26 Welsh 29, 30
TB 6, 7 OI 17, 18 OPru. 13, 14 Osc. 31, 17x
Ved. 8, 9 Lat. 19, 17x Latv. 14, 27 Umb. 19, 17x
h
8 PIIr. *d āráyati ‘hold’ h
17 *g ab- ‘take, hold’
13 PBalt. *laikā ‘hold’ 17x PItal. *habēt ‘have’
14 PBalt. *turi ‘have (?)’ 19 PItal. *tenēt ‘hold’
15 PGmc. *haldidi ‘keep’ 28 PGmc. *aih, *aig- ‘have’
16 PGmc. *habaiþi ‘have, hold’
The following states, though appearing more than once in the table, are confined to par-
ticular languages:
1 Hitt. harzi, hark-
2 Arm. owni
3 Gk. ἔχει
We accept the hypothesis that states 17 and 17x are etymologically connected; their dis-
tribution can be taken as a (weak) validation of the Italo-Celtic subgroup.
The internal polymorphism of 350 ‘have’ is confined to Germanic and is leaf-connected.
The clear and unarguable parallel development, involving states 15 and 16, is likewise
confined to Germanic. Note that it has continued beyond the period of the languages
investigated here, so that halten is now the usual word for ‘hold’ in High German (though
haltan usually meant ‘keep, preserve, pasture’ in OHG).
The development of the Baltic states is difficult to judge: the absence of a word for
‘have’ in Latvian looks like an archaism (see below), but Lithuanian and the less closely
related Old Prussian support the reconstructions given above.

September 2012
 40

350 ‘have’ is perhaps the only character for which absence of a lexeme might reflect
shared history, since it seems reasonably clear that there was no such lexeme in PIE (the
usual expression for ‘x has y’ being literally ‘y is to x’). We might therefore assign state
7 in the meaning ‘have’ not only to TB, but also to Vedic, Avestan, Old Irish (since OIr.
táithi means ‘it is to him’), TA, and Welsh—all well-attested languages for which the
absence of ‘have’ is plausibly ancestral. However, it is difficult to decide whether to
include poorly attested archaic languages (Luvian, Lycian, Old Persian) in this coding,
since the absence of ‘have’ in our records of those languages could be an accident; the
Latvian situation is also problematic (see above). We have therefore cautiously assigned
unique states for 350 ‘have’ to all the languages which lack such a lexeme.
Note that there is no cognation between states 16 and 17 (!!). Though the suffixes of 16
and 17x are probably ultimately related, the root of PGmc. *habaiþi is cognate with that
of Lat. capit ‘takes’, not with that of habet ‘has’. This is an excellent example of conver-
gent development in sound and meaning, which is very common. We have not included
Albanian ka ‘has’ in a superstate with state 16 because its shape poses problems severe
enough to make its root-etymology doubtful (see Demiraj 1997:212 for brief discussion).

78 horn [with parallel development]

[two characters]
Hitt. 1 Av. 1 Luv. 1 Goth. 1x
Arm. 2 OCS 4 Lyc. 6 ON 1x
Gk. 1 Lith. 4 TA 2 OHG 1x
Alb. 3 OE 1x OPer. 7 Welsh 8 [loan]
TB 2 OI 5 OPru. 4 Osc. 9
Ved. 1 Lat. 1 Latv. 4 Umb. 10
1 derivs. of *ḱer- ‘head-bone’ 2 *ghrḗwr̥
1x PGmc. *hurną 4 PBS *ragas
We have employed both codings, since the connection between state 1x and the other
derivatives of *ḱer- is likely to have been direct. It is conceivable that there is a direct
connection between the Germanic, Latin, and Luvian forms with *-n-, and between the
Greek and Indo-Iranian s-stems; but since that is not demonstrable, we have cautiously
coded all but the Germanic derivative with the superstate.
On the complex history of the family of words represented by state 1 see Nussbaum
1986.

September 2012
 41

79 how
Hitt. 1 Av. 2 Luv. 4 Goth. 2
Arm. 2 OCS 2 Lyc. 5 ON 2
Gk. 2 Lith. 2 TA 6 OHG 2
Alb. 2 OE 2 OPer. 7 Welsh 2
TB 3 OI 2 OPru. 2 Osc. 8
Ved. 2 Lat. 2 Latv. 2 Umb. 9
2 derivs. of the interrogative stem *kwe- ~ *kwi- ~ *kwo-
See Olsen 1999:806 with references on the first part of the Armenian compound.
It is not clear to us that Tocharian B mäkte contains any reflex of the interrogative stem
(pace Adams 1999 s.v.), though the parallel of mäksu ‘which?’, transparently a com-
pound of su ‘that’, makes it clear that it does contain a reflex of the demonstrative.

80 hunt
Hitt. 1 Av. 7 Luv. 13 Goth. 17
Arm. 2 OCS 8 Lyc. 14 ON 18
Gk. 3 Lith. 9 TA 15 OHG 18
Alb. 4 OE 10 OPer. 16 Welsh 19
TB 5 OI 11 OPru. 9 Osc. 20
Ved. 6 Lat. 12 Latv. 9 Umb. 21
9 derivs. of PBalt. *medjan ‘woods’ 18 PNWGmc. *waiþ-

September 2012
 42

81 husband, 99 man, AND 117 person [polymorphic set]

81 husband 99 man 117 person
Hitt. 1 1 2
Arm. 3 3 4a
Gk. 3 3 5
Alb. 6 6 3
TB 7 8 9
Ved. 7 3 10
Av. 7 3/11 4b
OCS 10 10 12
Lith. 11 11 13a
OE 11 11 10
OI 11 11 13b
Lat. 14 11 13a
Luv. 15 15 15
Lyc. 15 16 17
TA 7 8 18
OPer. 19 3 4b
OPru. 11 11 13a
Latv. 11 11 20 [loan]
Goth. 21 11/13a 10
ON 11 10 10
OHG 13a+10 10 10
Welsh 11 11 13b
Osc. 22 23 13a
Umb. 24 25 13a
3 *h2nḗr ‘man’ 10 *manu- ~ *manw- (‘person’ ?)
4 derivs. of *mer- ‘die’ 11 *wih1rós ‘young man’ (cf. TA wir ‘young’)
4a *mr̥tós ‘mortal’ 13 derivs. of *dhéǵh4m ‘earth’
4b *mr̥tyós ‘mortal’ 13a n-stem ‘earthling’
7 *pótis ‘master, husband’ 13b yo-stem ‘earthling’
8 PToch. *ënkwë ‘man’ 15 PLuv. *zīdis ‘man’
(← ‘person’, < *n̥ḱwós ‘mortal’)
The following states, though appearing more than once in the table, are confined to par-
ticular languages:
1 Hitt. pisēnas ‘man’ 6 Alb. burrë ‘man’

September 2012
 43

The reconstructable meanings reveal extensive parallel development. States 4a–b, 13a–b
have all been coded separately.

82 I AND 82a me [polymorphic set]

Hitt. 1, 2 Av. 1, 2 Luv. 2, 2 Goth. 1, 2y
Arm. 1, 2 OCS 1, 2x Lyc. 2, 2 ON 1, 2y
Gk. 1, 2 Lith. 1, 2x TA 4, 4 OHG 1, 2y
Alb. 3, 2 OE 1, 2 [dat.] OPer. 1, 2 Welsh 2, 2
TB 4, 4 OI 2, 2 OPru. 1, 2x Osc. 1, 5
Ved. 1, 2 Lat. 1, 2 Latv. 1, 2x Umb. 6, 2
1 nom. *éǵh2 ‘I’ 4 Toch. forms with n-, ñ-
2 acc. *mé (*emé ?; cf. Melchert 1994:74-5), enclitic *me ‘me’
2x extended stem *me-n-
2y emphatic *mége
The accusative form current in the late West Saxon dialect of Old English is the inherited
dative; other dialects of Old English preserve the inherited Germanic accusative 2y.
We have coded the substates of state 2 separately. However, we have not coded the ap-
pearance of the emphatic particle *-ém ~ *-óm, which exhibits no recognizable pattern.
Note that the final -g of the Hittite accusative cannot reflect the emphatic particle *-ge
(which would have become “-gi”).
On the Albanian nominative see Demiraj 1997:400 with references. We have cautiously
coded the Tocharian forms separately; see Jasanoff 1989 for an explanation which is
workable but still somewhat speculative.

83 ice
Hitt. 1 Av. 7 Luv. 10 Goth. 14
Arm. 2 OCS 8 Lyc. 11 ON 7
Gk. 3 Lith. 8 TA 12 OHG 7
Alb. 4 OE 7 OPer. 13 Welsh 1
TB 5 OI 1 OPru. 8 Osc. 15
Ved. 6 Lat. 9 Latv. 8 Umb. 16
1 *yeg- 7 *eys- 8 PBS *ledus
On the Avestan form see Bartholomae 1979 s.v. aēxa-.

September 2012
 44

84 if [with parallel development]

Hitt. 1 Av. 6 Luv. 8 Goth. 6y
Arm. 2 OCS 7 Lyc. 10 ON 11
Gk. 3 Lith. 6x TA 5 OHG 11
Alb. 4 OE 6y OPer. 6 Welsh 12
TB 5 OI 8 OPru. 6z (?) Osc. 9
Ved. 6 Lat. 9 Latv. 6 Umb. 9
5 derivs. of interrogative *kwi- 8 *mā-
6 relative nt. sg. *Hyód(±i) and derivs. 9 PItal. *swai
6y PGmc. *jab- 11 PGmc. *ib-, deriv. of
6x, 6z other derivs. demonstrative *i-
We have cautiously coded states 6 and 6x–z separately, since there is a reasonable like-
lihood that they did not replace one another directly.
Note that the Welsh word is etymologically a form of the verb ‘be’, not a derivative of
the interrogative (Evans 1964:242-3).
Since the demonstrative *i- and the relative *Hyó- have been largely confused in Balto-
Slavic, there is some question whether the Baltic forms (especially the Old Prussian
form) should be coded 6 or 11. Otherwise the parallel development seems to be confined
to Germanic and to involve interchange between the parallel stems *jab- and *ib-.

85 in
[two characters]
Hitt. 1x Av. 4 Luv. 1x Goth. 1
Arm. 1 OCS 1 Lyc. 1x ON 1
Gk. 1 Lith. 1 TA 2 OHG 1
Alb. 1 OE 1 OPer. 5 Welsh 1
TB 2 OI 1 OPru. 1 Osc. 1
Ved. 3 Lat. 1 Latv. 6 Umb. 1
1 *en 1x *éndom 2 PToch. *-në
We employ both codings, since it is reasonable to hypothesize a direct historical con-
nection between states 1 and 1x.

September 2012
 45

86 kill [polymorphic]
Hitt. 1 Av. 1 Luv. 11 Goth. 8/14
Arm. 2 OCS 6 Lyc. 12 ON 14
Gk. 3 Lith. 7 TA 5 OHG 8/14
Alb. 4 OE 8 OPer. 1 Welsh 15
TB 5 OI 9 OPru. 13 Osc. 16
Ved. 1 Lat. 10 Latv. 13 Umb. 17
1 *gwhénti 13 PBalt. *galin-
5 PToch. *kawṣəә(ṣəә) 14 PGmc. *daudīþi (deriv. of *daudaz
8 PGmc. *slahidi ‘dead’)
The polymorphism is confined to Germanic and is leaf-connected.

87 knee
[two characters]
Hitt. 1 Av. 1 Luv. 3 Goth. 1x
Arm. 1 OCS 2a Lyc. 4 ON 1x
Gk. 1 Lith. 2b TA 1 OHG 1x
Alb. 1 OE 1x OPer. 5 Welsh 1y
TB 1 OI 1y OPru. 6 Osc. 7
Ved. 1 Lat. 1 Latv. 2b Umb. 8
1 *ǵónu ~ (loc.) *ǵénu ~ *ǵnéw- and derivs.
1x PGmc. *knewą
1y PCelt. *glūn- < *gnūn-
2 derivs. of *kwel- ‘turn’
2a pre-Slav. *kolen-
2b PEBalt. *keljas
We accept the usual derivation of the Albanian form from *glun- < *gnun- (cf. dialectal
gjũ, glu, Demiraj 1997:190) and the usual judgment that this and the similar Celtic devel-
opment were independent.
For superstate 1 we employ both codings, since direct replacement of 1 by 1x and 1y
seems very likely. States 2a and 2b have been coded separately, however, because there
seems to be no close connection between them.

September 2012
 46

88 know [polymorphic]
[two characters by conflated split coding; both still polymorphic]
Hitt. 1 Av. 2/5 Luv. 7 Goth. 2
Arm. 2 OCS 2/5 Lyc. 8 ON 2
Gk. 2 Lith. 5 TA 4/5 OHG 2
Alb. 3 OE 2 OPer. 5 Welsh 2
TB 4 OI 2 OPru. 2 Osc. 9
Ved. 2 Lat. 6 Latv. 5 Umb. 10
2 perf. *wóyde and derivs. 5 *ǵneh3- ‘recognize, know (by sight)’
4 PToch. *kəәrsna(ṣəә)
The polymorphism is extensive but leaf-connected. We employ conflated split coding,
with 4 ∪ 4/5 coded against 5 in one character and all three coded together in the other,
but that does not eliminate all the polymorphism.

89 lake [polymorphic]
Hitt. 1 Av. 6 Luv. 11 Goth. 8+9
Arm. 2 OCS 7 Lyc. 12 ON 14
Gk. 3 Lith. 7 TA 3 OHG 9
Alb. 4 OE 8/9 OPer. 13 Welsh 15
TB 3 OI 10 OPru. 7 Osc. 16
Ved. 5 Lat. 10 Latv. 7 Umb. 17
3 *léymon- ~ *limn-´ (loc. *limén) 9 PGmc. *saiwiz
7 PBS *ežeran 10 PIC *lóku ~ *l̥kéw-
8 *móri ‘sea’
The polymorphism is confined to Germanic and is leaf-connected.

90 laugh
Hitt. 1 Av. 6 Luv. 12 Goth. 9
Arm. 2 OCS 7 Lyc. 13 ON 9
Gk. 2 Lith. 8 TA 4 OHG 9
Alb. 3 OE 9 OPer. 14 Welsh 16
TB 4 OI 10 OPru. 15 Osc. 17
Ved. 5 Lat. 11 Latv. 7 Umb. 18
2 *ǵelh2- 7 derivs. of *smey- ‘smile’
4 PToch. *kër- 9 PGmc. *hlahidi (*hlahja-)

September 2012
 47

91 leaf [polymorphic]
Hitt. 1 Av. 7 Luv. 12 Goth. 10
Arm. 2 OCS 8 Lyc. 13 ON 10+16
Gk. 3 Lith. 9 TA 5 OHG 10/16
Alb. 4 OE 10 OPer. 14 Welsh 11
TB 5 OI 11 OPru. 15 Osc. 17
Ved. 6 Lat. 3 Latv. 9 Umb. 18
3 *bhólyom (?) 10 PGmc. *laubaz
y
5 PToch. *p əәlta 11 PCelt. *dolin-
9 PEBalt. *lap- 16 PNWGmc. *bladą
According to the judgment we have preferred, Gk. φύλλον is cognate with Lat. folium
but has undergone lexical analogy with φυτόν ‘plant’. However, it is also possible that
the Greek word is isolated, and that Lat. folium reflects *dhol- and is thus cognate with
the Celtic words.
The polymorphism is confined to Germanic and is leaf-connected.

92 left(-hand)
Hitt. 1 Av. 6 Luv. 11 Goth. 16
Arm. 2 OCS 6 Lyc. 12 ON 8
Gk. 3 Lith. 7 TA 5 OHG 8
Alb. 4 OE 8 OPer. 13 Welsh 17
TB 5 OI 9 OPru. 14 Osc. 18
Ved. 6 Lat. 10 Latv. 15 Umb. 18
5 PToch, *ś(uw)āl(i)y- (?) 8 PNWGmc. *winistraz
6 *sewyós 18 POU *nertro- (orig. ‘lower’)
On the Tocharian forms see Pinault 2002:248-61.

93 leg
Hitt. 1 Av. 7 Luv. 12 Goth. 17
Arm. 2 OCS 8 Lyc. 13 ON 18
Gk. 3 Lith. 9 TA 14 OHG 19
Alb. 4 OE 10 OPer. 15 Welsh 11
TB 5 OI 11 OPru. 16 Osc. 20
Ved. 6 Lat. 2 Latv. 9 Umb. 21
2 *ḱrūs- 11 PCelt. *koksā (cf. Lat. coxa ‘hip’, etc.)
9 PEBalt. *kājā

September 2012
 48

94 lie [with parallel development]

Hitt. 1 Av. 1 Luv. 1 Goth. 4
Arm. 2 OCS 4 Lyc. 1 ON 4
Gk. 1 Lith. 5 TA 4 OHG 4
Alb. 3 OE 4 OPer. 7 Welsh 9
TB 4 OI 4 OPru. 8 Osc. 10
Ved. 1 Lat. 6 Latv. 5 Umb. 11
1 *ḱéyor 5 PEBalt. *guli
h
4 *lég yetor ‘lie down [eventive]’
The parallel development consists in extending the meaning of eventive *legh- to include
also the stative meaning (originally indicated by *ḱey-).

95 live
[two characters]
Hitt. 1 Av. 2b Luv. 5 Goth. 4
Arm. 2a OCS 2b Lyc. 6 ON 4
Gk. 2b Lith. 2c TA 2b OHG 4
Alb. 3 OE 4 OPer. 2b Welsh 2d
TB 2b OI 2d OPru. 2b Osc. 7
Ved. 2b Lat. 2b Latv. 2e Umb. 8
w
2 *g eyh3- and derivs. 4 PGmc. *libaiþi
2a inherited pres. unclear
2b pres. *gwíh3weti
2c–e phrases including, and late denominatives of, adj. *gwih3wós
We employ both codings, since a direct connection between the states of superstate 2 is
very likely.

96 liver
Hitt. 1a Av. 2 Luv. 9 Goth. 13
Arm. 1b OCS 5 Lyc. 10 ON 7
Gk. 2 Lith. 6 TA 11 OHG 7
Alb. 3 OE 7 OPer. 12 Welsh 8
TB 4 OI 8 OPru. 2 Osc. 14
Ved. 2 Lat. 2 Latv. 2 Umb. 15
1a, 1b *lis- (divergent suffixes) 7 PNWGmc. *libaru
2 *Hyḗkwr̥ ~ *Hyékwn- 8 PCelt. *ow-

September 2012
 49

On *lis- ‘liver’ see Schindler 1966. We have coded states 1a and 1b separately because
the historical relation between them is unclear: possibly the Armenian word is a loan
from Anatolian; in any case its suffix shows that it has been “contaminated” by lexical
analogy (probably with state 2).
The reconstruction of a complete protoform for the Celtic words remains problematic; see
Pedersen 1909:61, 313.

97 long
[two characters]
Hitt. 1 Av. 6 Luv. 9 Goth. 7
Arm. 2 OCS 6 Lyc. 10 ON 7
Gk. 3 Lith. 6x TA 5 OHG 7
Alb. 4 OE 7 OPer. 6 Welsh 12
TB 5 OI 8 OPru. 6x Osc. 13
Ved. 6 Lat. 7 Latv. 11 Umb. 14
5 PToch. *pəәrkrë (< *bhr̥ǵhrós ‘tall’) 7 *longhos
6 *dl̥h1ghós
6x PBalt. *ilgas (with unexpected loss of *d-)
On the Hittite word see Melchert 1994:67; there is some connection with state 6, but the
root-shape is clearly different.
We reject the traditional connection of Persian dræng with the western words (state 7)
and thus do not reconstruct *dl- for the latter; it seems much more likely that the shape of
the Persian word is an independent development (from a cognate of set 6).
We employ both codings, since direct development of state 6x from 6 seems almost
certain.

98 louse
Hitt. 1 Av. 7 Luv. 13 Goth. 18
Arm. 2 OCS 8 Lyc. 14 ON 10
Gk. 3 Lith. 9 TA 15 OHG 10
Alb. 4 OE 10 OPer. 16 Welsh 10
TB 5 OI 11 OPru. 17 Osc. 19
Ved. 6 Lat. 12 Latv. 9 Umb. 20
9 PEBalt. *ut- 10 *lúHs

September 2012
 50

99 man
Forms a polymorphic set with 81 husband (q.v.) and 117 person.

100 many [polymorphic]

Hitt. 1 Av. 3 Luv. 9 Goth. 3/6
Arm. 2 [loan] OCS 6 Lyc. 10 ON 6
Gk. 3 Lith. 7 TA 1 OHG 3/6
Alb. 4 OE 3/6 OPer. 3 Welsh 12
TB 1 OI 3 OPru. 11 Osc. 13
Ved. 5 Lat. 8 Latv. 7 Umb. 14
1 *meǵh2- ‘big’ 6 *mVnog o- h

3 *pélh1u- ~ *pl̥h1éw- 7 PEBalt. *daug-

h
Arm. bazowmk can only be an Iranian loan; if it were cognate with the Vedic word, its
first syllable would end in a nasal.
The polymorphism involves the last common ancestor of Balto-Slavic and Germanic but
is leaf-connected (in Germanic).

101 meat
Hitt. 1 Av. 4 Luv. 8 Goth. 2
Arm. 2 OCS 2 Lyc. 9 ON 13
Gk. 3 Lith. 2 TA 10 OHG 5
Alb. 2 OE 5 OPer. 11 Welsh 14
TB 2 OI 6 OPru. 2 Osc. 15
Ved. 2 Lat. 7 Latv. 12 Umb. 16
2 *mḗms ~ *méms- 5 PWGmc. *flaiski
On the preform of state 2 see Ringe 1996:70-1.

102 moon [polymorphic, with parallel development]

[two characters]
Hitt. 1 Av. 5 Luv. 1 Goth. 5x
Arm. 2a OCS 2b/5 Lyc. 1 ON 7
Gk. 3 Lith. 5 TA 5 OHG 5x
Alb. 4 OE 5x OPer. 5 Welsh 2c
TB 5 OI 6 OPru. 5 Osc. 8
Ved. 5 Lat. 2b Latv. 5 Umb. 9

September 2012
 51

1 PAnat. *ormos 5 *mḗh1n̥s and derivs.

2a–c derivs. of *lewk- ‘light’ 5x PGmc. *mēnan-
2b *lowksneh2 ‘luminary’
Though the overt polymorphism is confined to Slavic, both the distribution of states 2a–c
and the meaning of their root argue parallel development; note especially that an exact
cognate of state 2b appears in Old Prussian in the meaning ‘star’, which virtually forces
the inference that specialization to ‘moon’ occurred independently in Latin and Slavic.
We have coded states 2a–c separately; but for superstate 5 we have employed both cod-
ings, since there is no reason not to suggest direct development of the inherited word into
an n-stem in Germanic.

103 mother
Hitt. 1 Av. 2 Luv. 1 Goth. 4
Arm. 2 OCS 2 Lyc. 1 ON 2
Gk. 2 Lith. 2 TA 2 OHG 2
Alb. 3 OE 2 OPer. 2 Welsh 5
TB 2 OI 2 OPru. 2 Osc. 2
Ved. 2 Lat. 2 Latv. 2 Umb. 2
1 PAnat. *annos 2 *meh2tḗr
The Albanian and Welsh forms are “nursery words” and should not be assigned the same
state.

104 mountain
[two characters]
Hitt. 1 Av. 6a Luv. 11 Goth. 15
Arm. 2 OCS 6b Lyc. 12 ON 16
Gk. 3 Lith. 7 TA 5 OHG 8
Alb. 4 OE 8 OPer. 13 Welsh 10b
TB 5 OI 9 OPru. 14 Osc. 17
Ved. 6a Lat. 10a Latv. 7 Umb. 18
y
5 PToch. *ṣw əәlë 7 PEBalt. *kālnas
6a PIIr. *gr̥ís 8 PWGmc. *berg
6b PSlav. *gora 10a, 10b derivs. of *mon-
We have employed both codings, since it is reasonable to hypothesize a direct connection
between states 6a and 6b, and between 10a and 10b.

September 2012
 52

105 mouth
Hitt. 1 Av. 1 Luv. 1 Goth. 8
Arm. 2 OCS 6 Lyc. 10 ON 8
Gk. 3 Lith. 7 TA 5 OHG 8
Alb. 4 OE 8 OPer. 11 Welsh 9
TB 5 OI 9 OPru. 6 Osc. 13
Ved. 1 Lat. 1 Latv. 12 Umb. 14
1 *h1éh3s ~ *h1h3és- 8 PGmc. *munþaz
5 PToch. *kọyn 9 PCelt. *genus (←< *ǵḗnus ‘jaw’)
6 PBS *austo- (< satem *austh2o- or *ousth2o- ‘lip’)
On the Hittite cognate see Melchert 1994:115-6. We reject the connection between states
1 and 6 that has traditionally been posited.
On the Tocharian forms see now Hilmarsson 1996:171-2.
Though Arm. beran and Lith. burnà could share a root, its original meaning was clearly
not ‘mouth’, and the derivations of the two words are wholly different (cf. Olsen 1999:
297); they have therefore been assigned unique states.

106 name
[two characters]
Hitt. 1x Av. 1 Luv. 1y Goth. 1
Arm. 1 OCS 1 Lyc. 1y ON 1
Gk. 1 Lith. 2 TA 1 OHG 1
Alb. 1 OE 1 OPer. 1 Welsh 1
TB 1 OI 1 OPru. 1 Osc. 1
Ved. 1 Lat. 1 Latv. 2 Umb. 1
1 *h1nḗh3mn̥ ~ *h1néh3mn- 2 PEBalt. *vārdas
1x, 1y altered by dissimilation
We employ both codings of superstate 1.

September 2012
 53

107 narrow [polymorphic]

[two characters]
Hitt. 1 Av. 6 Luv. 9 Goth. 2/14
Arm. 2 OCS 2 Lyc. 10 ON 2/14
Gk. 3 Lith. 7 TA 11 OHG 2
Alb. 4 [loan] OE 2 OPer. 12 Welsh 2x/8
TB 5 OI 2x/8 OPru. 13 Osc. 15
Ved. 2 Lat. 2 Latv. 7 Umb. 16
h
2 *h2énǵ us and derivs. 8 PCelt. *koilos
2x PCelt. *komangus 14 derivs. of PGmc. *þrinhaną ‘squeeze’
7 PEBalt. *sjauras
The two instances of polymorphism are confined to Germanic and Celtic respectively;
both are leaf-connected.
We employ both codings, since it is reasonable to suppose that the Celtic (intensive?)
compound directly replaced the simplex adjective. Under the narrower coding we have
coded 2x/8 as a single state, since states 2x and 8 always occur together.

108 near
Hitt. 1 Av. 7 Luv. 13 Goth. 10
Arm. 2 OCS 8 Lyc. 14 ON 10
Gk. 3 Lith. 9 TA 15 OHG 10
Alb. 4 OE 10 OPer. 7 Welsh 18 [loan?]
TB 5 OI 11 OPru. 16 Osc. 19
Ved. 6 Lat. 12 Latv. 17 Umb. 19
7 PIr. *asnai 19 POU (superlative) *nessīmo-
10 PGmc. *nēhwa-
On the Celtic words see Pedersen 1909:161.

109 neck [polymorphic]

[two monomorphic characters by conflated split coding]
Hitt. 1 Av. 6/7 Luv. 12 Goth. 10
Arm. 2 OCS 8 Lyc. 13 ON 10
Gk. 3 Lith. 9 TA 14 OHG 10
Alb. 4 OE 10 OPer. 15 Welsh 7
TB 5 OI 11 OPru. 16 Osc. 17
Ved. 6 Lat. 10 Latv. 9 Umb. 18

September 2012
 54

6 PIIr. *grīvā 9 PEBalt. *kaklas

7 *mon- 10 *kólsos
The polymorphism is confined to Indo-Iranian and is leaf-connected. We employ con-
flated split coding to produce two monomorphic characters, 6 ∪ 6/7 coded against 7 in
one character and all three coded together in the other.

110 new [polymorphic]

Hitt. 1 Av. 1 Luv. 1 Goth. 1y
Arm. 1x OCS 1 Lyc. 4 ON 1y
Gk. 2 Lith. 1y TA 1 OHG 1y
Alb. 3 OE 1y OPer. 5 Welsh 1y
TB 1 OI 1y OPru. 1 Osc. 7
Ved. 1/1y Lat. 1 Latv. 6 Umb. 8
1 *néwos 1x (deriv. with r-suffix) 1y *néwios
We have coded states 1, 1x, and 1y separately, simply because if superstate 1 is coded as
a unit it is the only nonunique state and the character becomes uninformative.
The polymorphism of states 1 and 1y, leaf-connected at Vedic, is extraordinarily exten-
sive, involving Proto-Indo-Iranian, Proto-Baltic, Proto-Balto-Slavic, Proto-Italo-Celtic,
and all internal nodes that dominate them up to and including Proto-Nuclear IE. We sug-
gest that that is a viable hypothesis: early Vedic is attested from about 1500 B.C.E., its
last common ancestor with Italo-Celtic is most unlikely to predate 3500 B.C.E., and a
case of polymorphism between two nearly identical states over two millennia is not
unreasonable (though the necessity of extending it down through Proto-Baltic, probably
much later than 1500 B.C.E., is unsettling).

111 night [polymorphic, with parallel development]

Hitt. 1 Av. 1 Luv. 6 Goth. 3
Arm. 2 OCS 3 Lyc. 7 ON 3
Gk. 3 Lith. 3 TA 4 OHG 3
Alb. 3 OE 3 OPer. 1 Welsh 3
TB 4 OI 5 OPru. 3 Osc. 8
Ved. 1/3 Lat. 3 Latv. 3 Umb. 9
w
1 *k sep- 4 PToch. *wyəәṣẹye
3 *nókwts ~ *nékwt- (‘evening’?)
Like the preceding character, this appears to be a case of very extensive polymorphism
leaf-connected at Vedic, a language attested very early. But in this case there is evidence

September 2012
 55

that parallel development has occurred: the Hittite cognate of state 3, nekuz (mēhur),
means ‘evening (time)’, and Tocharian offers related adverbs meaning ‘in the evening’ or
‘at night’ (TB nekcīye, TA nakcu, etc.; see Adams 1999 s.v. nekcīye with references). It
is therefore plausible to suggest that state 3 began to replace state 1 in Nuclear IE and
eventually did so, largely independently, in every daughter except Indo-Iranian.

112 nose
Hitt. 1 Av. 6 Luv. 8 Goth. 11
Arm. 2 OCS 6 Lyc. 9 ON 6
Gk. 3 Lith. 6 TA 5 OHG 6
Alb. 4 OE 6 OPer. 6 Welsh 12
TB 5 OI 7 OPru. 6 Osc. 13
Ved. 6 Lat. 6 Latv. 10 Umb. 14
5 PToch. *mëlë- 6 *nās- ~ *nas-

113 not
Hitt. 1 Av. 1 Luv. 1 Goth. 1
Arm. 2 OCS 1 Lyc. 1 ON 4
Gk. 2 Lith. 1 TA 3 OHG 1
Alb. 1 OE 1 OPer. 1 Welsh 1
TB 3 OI 1 OPru. 1 Osc. 1
Ved. 1 Lat. 1 Latv. 1 Umb. 1
1 *né and extensions 3 PToch. *ma
2 *h2óyu ‘life’
On the Greek and Armenian forms see Cowgill 1960.

114 old [polymorphic, with parallel development]

Hitt. 1 Av. 2 Luv. 8 Goth. 12a
Arm. 2 OCS 6 Lyc. 9 ON 12b
Gk. 3 Lith. 2 TA 5 OHG 7/12a
Alb. 4 [loan] OE 7 OPer. 10 Welsh 2
TB 5 OI 2 OPru. 11 Osc. 13
Ved. 2 Lat. 6 Latv. 6 Umb. 14

September 2012
 56

2 *sénos 12a PGmc. *firnijaz

5 PToch. *k(əә)tsaitstsë 12b PGmc. or pre-ON *furnaz
6 *wet-us- ‘year-old’
7 PWGmc. *ald (< *‘grown up’, at first applied only to living things)
The overt polymorphism is confined to West Germanic; however, the distribution of
states 2 and 6 is incompatible with any likely tree—note that even Lithuanian and Latvian
are separated—and the fact that the latter is derived from ‘year’ strongly argues parallel
development.
We code states 12a, 12b together, since the latter is unique and unmediated replacement
is very likely.

115 one
Hitt. 1a Av. 1c Luv. 5 Goth. 1d
Arm. 2 OCS 3 Lyc. 6 ON 1d
Gk. 2 Lith. 4 TA 2 OHG 1d
Alb. 2 OE 1d OPer. 1c Welsh 1d
TB 2 OI 1d OPru. 1d Osc. 7
Ved. 1b Lat. 1d Latv. 4 Umb. 8
1 derivatives of *oy- ‘single’ 2 *sem-
1a *oyos 4 PEBalt. *vienas
1b *óykos
1c *óywos
1d *óynos
There are fairly strong indications that the original meaning of state 1 was not the nu-
meral ‘one’; for instance, the Greek cognate of 1c, οἶος, means ‘alone’, while that of 1d,
οἴνη, means ‘one-spot (on dice)’, and the Latin adverb ‘once’ is semel—arguably
“stranded” derivationally when the numeral from which it was derived was replaced by
*oinos > ūnus. We have therefore coded states 1a–d separately. On the Hittite form see
Eichner 1992:34, 42-4.
We have not coded the Slavic and East Baltic forms as substates of 1 because they do not
fit the set by regular sound correspondences; we believe that the origin of those forms
remains very unclear.

September 2012
 57

116 other
Hitt. 1 Av. 4a Luv. 7 Goth. 4b
Arm. 2 OCS 5 Lyc. 8 ON 4b
Gk. 2 Lith. 6 TA 2 OHG 4b
Alb. 3 OE 4b OPer. 4a Welsh 2
TB 2 OI 2 OPru. 6 Osc. 2
Ved. 4a Lat. 2 Latv. 6 Umb. 9
2 *ályos 4b PGmc. *anþeraz (< *án-teros)
4a PIIr. *Vnyás 6 PBalt. *kitas
We have coded states 4a, 4b separately because it is not clear that there is any direct
connection between them—especially in view of the fact that there may be some root-
connection with state 2 (conceivably *ál-yo-s : *án-tero-s, with an archaic consonant
alternation).

117 person
Forms a polymorphic set with 81 husband (q.v.) and 99 man.

118 pierce
Hitt. 1 Av. 7 Luv. 12 Goth. 15
Arm. 2 OCS 8 Lyc. 13 ON 16
Gk. 3 Lith. 9 TA 5 OHG 17
Alb. 4 OE 10 OPer. 14 Welsh 18
TB 5 OI 11 OPru. 8 Osc. 19
Ved. 6 Lat. 8 Latv. 9 Umb. 20
5 PToch. *tsopəә(ṣəә) 9 PEBalt. *durja
8 *bhodhh2- ‘dig’

119 play
Hitt. 1 Av. 7 Luv. 13 Goth. 19
Arm. 2 OCS 8 Lyc. 14 ON 20
Gk. 3 Lith. 9 TA 15 OHG 21
Alb. 4 OE 10 OPer. 16 Welsh 22
TB 5 OI 11 OPru. 17 Osc. 23
Ved. 6 Lat. 12 Latv. 18 [loan] Umb. 24
(No cognates.)

September 2012
 58

120 pull [polymorphic]

Hitt. 1 Av. 6 Luv. 12 Goth. 17
Arm. 2 [loan] OCS 7 Lyc. 13 ON 9
Gk. 3 Lith. 8 TA 14 OHG 10/17
Alb. 4 OE 9/10 OPer. 15 Welsh 18 [loan]
TB 3 OI 11 OPru. 16 Osc. 19
Ved. 5 Lat. 9 Latv. 7 Umb. 20
3 *selk- 10 *dewk- ‘lead’
7 *h2welk- 17 PGmc. *þinsidi
9 *dhragh-
It is possible that the Albanian form reflects some sort of conflation of states 3 and 7 (cf.
Demiraj 1997:48), but it cannot reflect either one directly.
The (first) Arm. term is an Iranian loan; cf. Hübschmann 1899:47.
The Germanic polymorphism is local, but a triple set 10/11/17 (not actually attested)
must be posited for Proto-West Germanic, and a double set 10/17 (not actually attested)
for the other internal nodes of the Germanic subtree.

121 push
Hitt. 1 Av. 6 Luv. 12 Goth. 9
Arm. 2 OCS 7 Lyc. 13 ON 9
Gk. 3 Lith. 8 TA 5 OHG 9
Alb. 4 OE 9 OPer. 14 Welsh 16
TB 5 OI 10 OPru. 15 Osc. 17
Ved. 5 Lat. 11 Latv. 8 Umb. 18
5 *newd- 9 PGmc. *skeubidi
8 PEBalt. *stumja
On the unexpected *ū in the ON and OE cognates of set 9 see now Ringe and Taylor,
forthcoming.

September 2012
 59

122 rain
[two characters]
Hitt. 1 Av. 5 Luv. 11 Goth. 8
Arm. 2 OCS 6 Lyc. 12 ON 8
Gk. 3 Lith. 7 TA 3x OHG 8
Alb. 3 OE 8 OPer. 13 Welsh 15
TB 3x OI 9 OPru. 14 Osc. 16
Ved. 4 Lat. 10 Latv. 7 Umb. 17
3 derivs. of *suh2- (‘pour’?) 7 PEBalt. *lietus
3x PToch. *suwës- 8 PGmc. *regną
Since in both Tocharian languages the verb ‘rain’ reflects states 3 and 3x, a direct link
between them is demonstrable; accordingly we employ both alternative codings for this
character, the related noun.

123 red [polymorphic]

Hitt. 1 Av. 3b Luv. 6 Goth. 3d
Arm. 2 [loan] OCS 5 Lyc. 7 ON 3d
Gk. 3a Lith. 3c TA 3a OHG 3d
Alb. 4 OE 3d OPer. 8 Welsh 3d
TB 3a OI 3d OPru. 9 Osc. 11
Ved. 3b Lat. 3a Latv. 10 Umb. 3a/3d
3 derivs. of *h1rewdh- ‘be red’
3a *h1rudhrós 3c PEBalt. or pre-Lith. *raudānas
3b PIIr. *ráudhitas 3d *h1rowdhos
States 5 and 9 are apparently independent derivatives of words for ‘worm’.
The polymorphism is leaf-connected (at Umbrian) but must be posited not only for the
internal nodes of Italic but also for Proto-Italo-Celtic; in addition, state 3d is shared with
Germanic. The latter phenomenon could reflect early lexical borrowing; but since all the
states of superstate 3 are derivatives of a Caland root, parallel development cannot be
excluded.
We have coded states 3a–d separately, simply because if superstate 3 is coded as a unit it
is the only nonunique state and the character becomes uninformative.

September 2012
 60

124 right(-hand)
Hitt. 1 Av. 3c Luv. 6 Goth. 3f
Arm. 2 OCS 3c Lyc. 7 ON 12
Gk. 3a Lith. 3c TA 8 OHG 3f
Alb. 3b OE 5 OPer. 9 Welsh 3g
TB 4 OI 3d OPru. 10 Osc. 3e
Ved. 3c Lat. 3e Latv. 11 Umb. 3e
3 derivs. of *deḱs-; the suffixes are:
3a *-io- 3c *-ino- 3e *-(i)tero- 3g *-ewo-
3b *-to- 3d *-o- (?) 3f *-won-
On the Albanian form see Demiraj 1997:137-8; on the Tocharian forms see Schmidt
1994:281, 1996:276, and especially Pinault 2002:248-61.
We have coded states 3a–g separately, simply because if superstate 3 is coded as a unit it
is the only nonunique state and the character becomes uninformative.

125 right [polymorphic, but not effectively so]

Hitt. 1a Av. 1c Luv. 10 Goth. 6b
Arm. 2 OCS 7 Lyc. 11 ON 6b
Gk. 3 Lith. 8 TA 12 OHG 6b
Alb. 4 OE 6b OPer. 13 Welsh 9
TB 5 OI 9 OPru. 14 Osc. 16
Ved. 1b/6a Lat. 6b Latv. 15 Umb. 6b
1a – 1c derivs. of *ar- ‘fit’
6 derivs. of *h3reǵ- ‘put in a straight line’
6a *h3réǵ-u- ~ *h3r̥ǵ-éw- ‘straight’
6b verbal adj. *h3reǵtós (with remodelled ablaut)
9 PCelt. *kowīros ‘absolutely true’
We have coded states 1a–c and states 6a, 6b separately, because the reconstructable
meanings virtually guarantee that parallel development is involved; given such a coding,
the polymorphism is effectively eliminated.
Note that OPers. rāsta is not part of set 6; its root is PIIr. *rādh- ‘bring to a goal success-
fully’.

September 2012
 61

126 river
Hitt. 1 Av. 7 Luv. 1 Goth. 10
Arm. 2 OCS 8 Lyc. 12 ON 10
Gk. 3 Lith. 9 TA 13 OHG 10
Alb. 4 OE 10 OPer. 7 Welsh 1
TB 5 OI 1 OPru. 9 Osc. 14
Ved. 6 Lat. 11 Latv. 9 Umb. 15
1 *h2ébō ~ *h2ébon- 9 PBalt. *upē
7 derivs. of *srew- ‘flow’ 10 PGmc. *ahwō (= Lat. aqua ‘water’)
Note that Gothic flodus and OHG fluz differ in root-ablaut, suffix, and derivational rela-
tionships within their respective languages (even though they may reflect the same PIE
root); we have therefore coded them separately.

127 road [polymorphic]

Hitt. 1 Av. 6/7 Luv. 12 Goth. 9
Arm. 2 [loan] OCS 6 Lyc. 13 ON 9
Gk. 3 Lith. 8 TA 5 OHG 9
Alb. 4 OE 9 OPer. 6 Welsh 14 [loan]
TB 5 OI 10 OPru. 6 Osc. 11
Ved. 6/7 Lat. 11 Latv. 8 Umb. 11
5 PToch. *yəәtariye h
7 PIIr. *ád vā 9 PGmc. *wegaz
6 *póntoh2s ‘path’ 8 PEBalt. *keljas 11 PItal. *wiā
Both Armenian words appear to be loans (cf. Olsen 1999:892-3, 933).
The polymorphism is confined to Indo-Iranian and is leaf-connected.

128 root
Hitt. 1 Av. 7 Luv. 12 Goth. 10
Arm. 2 OCS 8 Lyc. 13 ON 10
Gk. 3 Lith. 9 TA 14 OHG 10
Alb. 4 OE 10 OPer. 15 Welsh 10
TB 5 OI 11 OPru. 9 Osc. 16
Ved. 6 Lat. 10 Latv. 9 Umb. 17
9 PBalt. *šaknis 10 *wreh2d- ~ *wr̥h2d-

September 2012
 62

129 rope [polymorphic]

Hitt. 1a Av. 7 Luv. 13 Goth. 17
Arm. 2 OCS 8 Lyc. 14 ON 1c/9/10
Gk. 3 Lith. 8 TA 15 OHG 1b/10
Alb. 4 OE 1b/9/10 OPer. 16 Welsh 18
TB 5 OI 11 OPru. 8 Osc. 19
Ved. 6 Lat. 12 Latv. 8 Umb. 20
1a–c derivs. of *sh2ey- ‘tie’ 9 PNWGmc. *raipa-
1b PWGmc. *sail 10 PNWGmc. *strangiz
8 PBS *wirwi-
On the Armenian words see Clackson 1994:228 fn. 175.
The three states of superstate 1 are clearly independent derivatives, and we have therefore
coded them separately.
The Germanic polymorphism is local and leaf-connected.

130 rotten
Hitt. 1 Av. 6b Luv. 9 Goth. 6d
Arm. 2 OCS 7 Lyc. 10 ON 6g
Gk. 3 Lith. 6c TA 11 OHG 6d
Alb. 4 OE 6d OPer. 12 Welsh 14 [loan]
TB 5 OI 8 OPru. 13 Osc. 15
Ved. 6a Lat. 6e Latv. 6f Umb. 16
6a–g derivs. of *pū- 6d PGmc. *fūlaz
We coded the substates of 6 separately, but the character is still uninformative.

131 round
Hitt. 1 Av. 7 Luv. 13 Goth. 18
Arm. 2 OCS 8 Lyc. 14 ON 10
Gk. 3 Lith. 9 TA 15 OHG 10
Alb. 4 OE 10 OPer. 16 Welsh 11
TB 5 OI 11 OPru. 17 Osc. 19
Ved. 6 Lat. 12 Latv. 9 Umb. 20
9 PEBalt. *apvalus 11 PCelt. *krundis
10 NWGmc. *sin-wal- (various extensions)
The East Baltic and Northwest Germanic compounds are clearly related, but as the com-
mon element is a verb root meaning ‘roll’, they can be completely independent forma-

September 2012
 63

tions.

132 rub
Hitt. 1 Av. 7 Luv. 12 Goth. 18
Arm. 2 OCS 8 Lyc. 13 ON 18
Gk. 3 Lith. 8 TA 14 OHG 9
Alb. 4 OE 9 OPer. 15 Welsh 19 [loan]
TB 5 OI 10 OPru. 16 Osc. 20
Ved. 6 Lat. 11 Latv. 17 Umb. 21
8 derivs. of *ter- 18 PGmc. *bnūidi
9 PWGmc. *gnīdidi

133 salt
Hitt. 1 Av. 5 Luv. 7 Goth. 2y
Arm. 2 OCS 2 Lyc. 8 ON 2y
Gk. 2 Lith. 6 TA 2x OHG 2y
Alb. 3 OE 2y OPer. 9 Welsh 2z
TB 2x OI 2z OPru. 2 Osc. 10
Ved. 4 Lat. 2 Latv. 2 Umb. 2
2 *sal- 2y PGmc. *saltą
y
2x PToch. *sal ẹye 2z PCelt. *saleno-
We coded the substates of 2 separately, since otherwise the character is uninformative.
(However, it is reasonable to supposed that 2x–z replaced unextended 2 directly.)

134 sand
Hitt. 1 Av. 6 Luv. 11 Goth. 15
Arm. 2 OCS 7 Lyc. 12 ON 2
Gk. 2 Lith. 8 TA 4 OHG 2
Alb. 3 OE 2 OPer. 13 Welsh 16
TB 4 OI 9 OPru. 14 Osc. 17
Ved. 5 Lat. 10 Latv. 8 Umb. 18
2 *sámh2dhos (see notes) 8 PEBalt. *smiltis
4 PToch. *warVñc-
We adopt the hypothesis that Gk. ἄµµος is the product of lexical analogy between
ἄµαθος, which is clearly related to Gmc. *sandaz, and an unrelated word ψάµµος (see
Frisk 1960, 1970 s.vv.); further, that Arm. awaz likewise belongs to this cognate set

September 2012
 64

(though -r rather than -z should probably be expected as a reflex of *dh; see also Olsen
1999:24).
The vowels of the second syllables of the Tocharian words do not match; since both
words are poorly attested, it is difficult to know what to make of that.

135 say [polymorphic]

Hitt. 1 Av. 6 Luv. 11 Goth. 15
Arm. 2 OCS 7 Lyc. 12 ON 8/15
Gk. 3 Lith. 1/8 TA 13 OHG 8/15
Alb. 4 OE 8/15 OPer. 4 Welsh 16
TB 5x OI 9 OPru. 14 Osc. 10
Ved. 5/6 Lat. 10 Latv. 8 Umb. 10
1 *ter- 6 *mléwHti ~ *mluH-
4 *ḱḗh1sti ~ *ḱéh1s- 8 *sekw-
5 *wekw- 10 PItal. *deyket (*deyḱ- ‘show’)
5x denom. of *wokw- ‘voice’ 15 PGmc. *kwiþidi (*kweþa-)
The polymorphism of this character is complex and extensive; especially striking are the
distributions of states 1 (Hittite and Lithuanian) and 4 (Albanian and Old Persian). It
seems very likely that PIE and many of its daughters had several verbs meaning ‘say’.
We have coded states 5 and 5x separately, since the latter represents a derivative of a
derivative of the basic verb.

136 scratch
Hitt. 1 Av. 7 Luv. 13 Goth. 19
Arm. 2 OCS 8 Lyc. 14 ON 12
Gk. 3 Lith. 9 TA 15 OHG 20
Alb. 4 OE 10 OPer. 16 Welsh 21
TB 5 OI 11 OPru. 17 Osc. 22
Ved. 6 Lat. 12 Latv. 18 Umb. 23
12 *skabh-

September 2012
 65

137 sea [polymorphic]

Hitt. 1 Av. 7 Luv. 11 Goth. 8
Arm. 2 OCS 8 Lyc. 12 ON 10
Gk. 3 Lith. 9 TA 13 [loan] OHG 8/10
Alb. 4 OE 10 OPer. 7 Welsh 8
TB 5 [loan] OI 8 OPru. 9 Osc. 14
Ved. 6 Lat. 8 Latv. 9 Umb. 15
7 PIran. *zrayah 9 PBalt. *jūrā
8 *móri ~ *mréy- 10 PGmc. *saiwiz
The polymorphism is confined to Germanic and is leaf-connected.

138 see
Hitt. 1 Av. 7 Luv. 12 Goth. 10
Arm. 2 OCS 8 Lyc. 13 ON 10
Gk. 3 Lith. 9 TA 5 OHG 10
Alb. 4 OE 10 OPer. 7 Welsh 15
TB 5 OI 11 OPru. 8 Osc. 16
Ved. 6 Lat. 8 Latv. 14 Umb. 8
5 PToch. *ləәka(ṣəә) 8 *wid-éh1-, stative of *weyd- ‘catch sight of’
7 PIran. *vainati 10 PGmc. *sihwidi (*sehwa-)
On the Albanian form see Demiraj 1997:57.

139 seed [ineffectively polymorphic; probably with parallel development]

Hitt. 1 Av. 8 Luv. 1 Goth. 12
Arm. 2 OCS 9a Lyc. 10 ON 12
Gk. 3 Lith. 9b TA 5/6 OHG 9a
Alb. 4 OE 9c OPer. 11 Welsh 9e
TB 5/6 OI 9d OPru. 9a Osc. 13
Ved. 7 Lat. 9a Latv. 9b Umb. 14
1 PAnat. *warwalan 9 derivs. of *seh1- ‘sow’
5 PToch. *sarm¡ 9a *séh1mn̥ ~ *sh1mén-
6 PToch. *śəәktalye 9b PEBalt. *sēklā
12 PGmc. *fraiwą 9c – e (other derivs.)
No relation between the Armenian, Greek, and Tocharian words can be demonstrated.
The Tocharian polymorphism is local and leaf-connected. Since states 5 and 6 always
occur together, we have coded 5/6 as a single state.

September 2012
 66

The distribution of states 9a – e poses interesting problems. Because these derivatives of

‘sow’ appear to be independent, we have coded them separately; but the character is still
incompatible with any plausible tree for Germanic.

140 sew
Hitt. 1 Av. 7 Luv. 9 Goth. 6
Arm. 2 OCS 6 Lyc. 10 ON 6x
Gk. 3 Lith. 6 TA 11 OHG 14
Alb. 4 OE 6 OPer. 12 Welsh 15
TB 5 OI 8 OPru. 13 Osc. 16
Ved. 6 Lat. 6 Latv. 6 Umb. 17
6 *siHw- 6x *saumīþi, deriv. of PGmc. *saumaz, itself a deriv. of 6
We have coded states 6 and 6x separately. (The character is uninformative no matter
which choice is made.)

141 sharp [polymorphic]

Hitt. 1 Av. 5b Luv. 9 Goth. 6
Arm. 2 OCS 3c Lyc. 10 ON 6/7
Gk. 3a Lith. 3c TA 11 OHG 6/7
Alb. 4 OE 6/7 OPer. 5b Welsh 13
TB 3b OI 8 OPru. 12 Osc. 14
Ved. 5a Lat. 3d Latv. 3e Umb. 15
3a–e derivs. of *h2eḱ- 6 PGmc. *hwassaz
3c PBS *aštrus 7 PNWGmc. *skarpaz
5a–b derivs. of *tig-
5b *tigrós
We have coded the various derivatives of 3 and 5 separately.
The polymorphism is confined to Northwest Germanic and is leaf-connected.

142 short
Hitt. 1 Av. 3 Luv. 1 Goth. 15
Arm. 2 OCS 7 Lyc. 11 ON 16
Gk. 3 Lith. 8 TA 12 OHG 16
Alb. 4 [loan] OE 9 OPer. 13 Welsh 10
TB 5 OI 10 OPru. 14 Osc. 17
Ved. 6 Lat. 3 Latv. 14 Umb. 18

September 2012
 67

1 PAnat. *mannenkwo- 14 PBalt. *insas

3 *mréǵhus ~ *mr̥ǵhéw- 16 PNWGmc. *skamm-
10 PCelt. *birros

143 sing
Hitt. 1 Av. 7 Luv. 11 Goth. 9
Arm. 2 OCS 5 Lyc. 12 ON 9
Gk. 3 Lith. 8 TA 5 OHG 9
Alb. 4 [loan] OE 9 OPer. 13 Welsh 10
TB 5 OI 10 OPru. 14 Osc. 16
Ved. 6 Lat. 10 Latv. 15 Umb. 10
5 *peyH- 10 PItCelt. *kaneti
9 PGmc. *singwidi

144 sit [polymorphic, with parallel development]

Hitt. 1 Av. 1 Luv. 1 Goth. 2b
Arm. 2a OCS 2a Lyc. 5 ON 2b
Gk. 1 Lith. 2a TA 4 OHG 2b
Alb. 3 OE 2b OPer. 6 Welsh 2e
TB 1/4 OI 2c OPru. 2d Osc. 7
Ved. 1 Lat. 2a Latv. 2a Umb. 2a
1 *ḗsor 2 *sed- ‘sit down’ (eventive)
y
4 PToch. *ṣəәm əә(ṣəә) 2a derived stative *sed-éh1-
2b pres. *sed-ye/o-
2c – e (other)
The parallel development in this character consists in adapting forms of the eventive root
*sed-, and of the Tocharian root that apparently replaced it, to replace inherited stative
*ēs-. We have accordingly coded the states of 2 separately; but parallel development (of
state 2a) must still be posited.

September 2012
 68

145 skin [polymorphic]

Hitt. 1 Av. 7 Luv. 12 Goth. 15
Arm. 2 OCS 8 Lyc. 13 ON 10
Gk. 3 Lith. 9 TA 5 OHG 10/15
Alb. 4 OE 10 OPer. 14 Welsh 16
TB 5 OI 11 OPru. 10 Osc. 17
Ved. 6 Lat. 10 Latv. 9 Umb. 18
5 PToch. *yëtsë 10 derivs. of *kewH-
9 PEBalt. *ādā 15 PGmc. *fellą (cf. Lat. pellis ‘hide’)
It is not clear whether the Avestan word belongs with state 10; see Bartholomae 1979
s.v., Buck 1949:201.
The polymorphism is confined to Germanic and is leaf-connected.

146 sky
Forms a polymorphic set with 23 cloud (q.v.) and 57 fog.

147 sleep [with parallel development]

Hitt. 1 Av. 6 Luv. 11 Goth. 8
Arm. 2 OCS 6 Lyc. 12 ON 6
Gk. 3 Lith. 7 TA 5 OHG 8
Alb. 4 OE 8 OPer. 13 Welsh 6x
TB 5 OI 9 OPru. 7 Osc. 15
Ved. 1 Lat. 10 Latv. 14 Umb. 16
1 *sésti 7 PBalt. *meig-
5 PToch. *kləәns- 8 PGmc. *slēpidi
6 *swep- ‘fall asleep’
6x deriv. of noun *swépnos ‘sleep’
We have coded states 6 and 6x separately.
The parallel development consists in the replacement of stative *ses- by eventive *swep-.

September 2012
 69

148 small [polymorphic]

Hitt. 1 Av. 7 Luv. 14 Goth. 10
Arm. 2 OCS 8 Lyc. 15 ON 10
Gk. 3 Lith. 9 TA 5 OHG 10/11
Alb. 4 OE 10/11 OPer. 16 Welsh 12
TB 5 OI 12 OPru. 5 Osc. 17
Ved. 6 Lat. 13 Latv. 9 Umb. 18
5 *lik- (?) 11 PGmc. *smalaz
9 PEBalt. *mažas 12 PCelt. *biggos
10 PGmc. *lītilaz (with unexplained u-vocalism in the WGmc. forms)
It is not certain that the Tocharian and Old Prussian forms are related; for an alternative
possibility see Adams 1999 s.v. lykaśke.
The polymorphism is confined to West Germanic and is leaf-connected.

149 smell
Hitt. 1 Av. 6 Luv. 11 Goth. 15
Arm. 2 OCS 7 Lyc. 12 ON 16
Gk. 2 Lith. 8 TA 4 OHG 9
Alb. 3 OE 9 OPer. 13 Welsh 17
TB 4 OI 10 OPru. 14 Osc. 18
Ved. 5 Lat. 2 Latv. 2 Umb. 19
2 *(H)od- 9 PWGmc. *stinkwidi
4 PToch. *wəәrṣṣəә(ṣəә)

150 smoke [polymorphic, but not effectively so]

Hitt. 1a Av. 6 Luv. 10 Goth. 14
Arm. 2 OCS 1b Lyc. 11 ON 7
Gk. 3 Lith. 1b TA 12 OHG 7
Alb. 4 OE 7/8a OPer. 13 Welsh 8b
TB 5 OI 9 OPru. 1b Osc. 15
Ved. 1b Lat. 1b Latv. 1b Umb. 16
1 derivs. of *dhuh2- 7 PGmc. *reuk- ~ *rauk-
h
1a *d uh2wey- 8a–b derivs. of *smewK- (see below)
1b *dhuh2mós
Though there is clearly some connection between the Old English and Welsh forms, the
root-final consonants do not match (the OE form reflecting PIE *-g- or *-ǵ-, the Welsh

September 2012
 70

form PIE *-k- or *-ḱ-); we have therefore coded them separately, and that makes the OE
polymorphism ineffective. We have also coded the derivatives of 1 separately, since they
appear to be independent.

151 smooth
Hitt. 1 Av. 7 Luv. 11 Goth. 16
Arm. 2 OCS 8 Lyc. 12 ON 16
Gk. 3 Lith. 8 TA 13 OHG 16
Alb. 4 OE 9 OPer. 14 Welsh 17
TB 5 OI 10 OPru. 15 Osc. 18
Ved. 6 Lat. 3 Latv. 8 Umb. 19
3 *ley-w- 16 PGmc. *slehtaz
8 Balto-Slavic *glād- ~ *glud- (details very unclear).

152 snake [polymorphic; with parallel development?]

[two characters by split and conflated split coding; polymorphism or
parallel development still present]
Hitt. 1 Av. 3 Luv. 11 Goth. 8/9
Arm. 2 OCS 6 Lyc. 12 ON 8/9
Gk. 3 Lith. 7 TA 5 OHG 8/9
Alb. 4 OE 8/9 OPer. 13 Welsh 9
TB 2/5 OI 9 OPru. 10 Osc. 15
Ved. 3 Lat. 10 Latv. 14 Umb. 16
2 *awǵ- 8 PGmc. *wurmiz ‘worm’
wh
3 *h3ég is 9 *neh1tri- (vel sim.)
5 PToch. *arṣaklo 10 *angw(h)is
The polymorphisms are confined to Tocharian and Germanic and are leaf-connected. We
employ split coding for the Germanic polymorphism 8/9 and the Celtic state 9, and con-
flated split coding for Tocharian and Armenian states, with the conflation 2/5 ∪ 5.
The distribution of states 3 and 10 suggests either some further polymorphism at internal
nodes or parallel development—it is difficult to determine which.
Note that Welsh sarff is a loan; on Albanian gjarpër see Demiraj 1997:183-4.

September 2012
 71

153 snow AND 415 winter [polymorphic set; see notes for coding]
Hitt. 1, 2 Av. 5, 2 Luv. 8, 9 Goth. 6a, 7
Arm. 2, 2 OCS 6a, 2 Lyc. 10, 11 ON 6a, 7
Gk. 2, 2 Lith. 6a, 2 TA 12, 2 OHG 6a, 7
Alb. 3, 2 OE 6a, 7 OPer. 13, 14 Welsh 15, 2
TB 4, 2 OI 6b, 2 OPru. 6a, 2 Osc. 16, 17
Ved. 2, 2 Lat. 6c, 2 Latv. 6a, 2 Umb. 18, 19
2 *ǵhéyōm ‘winter’ and derivs. 6 derivs. of verb *sneygwh- ‘snow’
7 PGmc. *wintruz ‘winter’ 6a *snoygwhos
6b, c (other)
We have adopted both codings for superstate 6; the result is that 153 ‘snow’ is two char-
acters, both of which exhibit parallel development of state 2.
The semantic expansion of state 2 (which can actually be documented in Greek) is a
southern areal phenomenon. In those languages that exhibit it there is secondary differ-
entiation of the words for ‘snow’ and ‘winter’, typically by derivational suffixation of the
latter; but since the suffixes and suffix complexes are unique (pointing to independent
development), we have not coded the forms separately.
Note that 415 ‘winter’ is an unexceptional monomorphic character which can be used as
input to the algorithm.

154 some
Hitt. 1a Av. 5 Luv. 7 Goth. 4b
Arm. 1b OCS 1d Lyc. 8 ON 4b
Gk. 2 Lith. 1e TA 4a OHG 4b
Alb. 3 OE 4b OPer. 1c Welsh 11
TB 4a OI 6x OPru. 9 Osc. 12
Ved. 1c Lat. 6+1 Latv. 10 Umb. 13
1a–e derivs. of indefinite *kwi- ~ *kwo-, including phrases
4 derivs. of *sem- ‘one’ 6 derivs. of *ályos ‘other’
4a PToch. *ṣemẹ 6x reduplicated
4b PGmc. *sumai 6+1 compound with 1
On the Armenian form see Olsen 1999:806 with references.
Since the reconstructable meanings of the states argue strongly argue parallel develop-
ment, we have coded all distinguishable states separately.

September 2012
 72

155 spit
Hitt. 1 Av. 3 Luv. 8 Goth. 3
Arm. 2 OCS 3 Lyc. 9 ON 3
Gk. 3 Lith. 3 TA 10 OHG 3
Alb. 4 OE 3 OPer. 11 Welsh 13
TB 5 OI 7 OPru. 12 Osc. 14
Ved. 6 Lat. 3 Latv. 3 Umb. 15
3 *spyewH- and derivatives
We have not coded the derivatives of the basic verb separately because they appear to be
unique.
The shape of the Vedic word does not fit set 3; we suggest that it is onomatopoeic.

156 split [polymorphic]

[two characters, both polymorphic]
Hitt. 1 Av. 6 Luv. 10 Goth. 14
Arm. 1 OCS 7 Lyc. 11 ON 8
Gk. 2 Lith. 1x TA 4 OHG 8
Alb. 3 OE 8 OPer. 12 Welsh 1
TB 4 OI 9 OPru. 13 Osc. 15
Ved. 5 Lat. 2/5 Latv. 1x Umb. 16
1 *sḱelH- 4 PToch. *kawtna(ṣəә)
1x pres. *sḱélyeti 5 *bheyd- (pres. *bhinédsti)
2 *sḱeyd- (pres. *sḱinédsti) 8 PNWGmc. *kliubidi (*kleuba-)
We employ both codings for superstate 1.
We reluctantly accept the hypothesis that Gk. σχίζει belongs with Lat. scindit (and Ved.
chinátti ‘cuts off’); replacement of the inherited nasal-infixed present is normal in Greek,
but the aspiration of the second consonant is difficult to account for.
Though the overt polymorphism is confined to Latin, the distribution of states suggests
polymorphism or parallel development elsewhere in the tree. Our remarks on 27 ‘cut’
(see above) apply to this character as well.

September 2012
 73

157 squeeze
Hitt. 1 Av. 6 Luv. 12 Goth. 9
Arm. 2 OCS 7 Lyc. 13 ON 9
Gk. 3 Lith. 8 TA 14 OHG 18
Alb. 4 OE 9 OPer. 15 Welsh 10
TB 5 OI 10 OPru. 16 Osc. 19
Ved. 3 Lat. 11 Latv. 17 Umb. 20
3 *pi-sed- 10 PCelt. *wāsk- ~ *wask-
9 PGmc. *þrinh- and derivs.

158 stand
[two characters]
Hitt. 1 Av. 3 Luv. 3 Goth. 3y
Arm. 2 OCS 3 Lyc. 3 ON 3y
Gk. 3 Lith. 3x TA 5 OHG 3y
Alb. 4 OE 3y OPer. 3 Welsh 3
TB 5 OI 3 OPru. 3 Osc. 3z
Ved. 3 Lat. 3z Latv. 3x Umb. 3z
3 *steh2-, pres. *stísteh2ti, perf. *stestóh2e, and derivs.
3x PEBalt. *stāv-
3y PGmc. *standidi
3z PItal. *staēt
5 PToch. *kəәlyəәtəәr
The stative meaning ‘be in a standing position’ was originally expressed by the perfect
stem, but was later shifted to other stems in various branches; we have not coded for that
complex of morphological changes.
We have employed both codings for superstate 3, since direct replacement of the inher-
ited verb by the derivatives is very likely.
On the Italic forms see Cowgill 1973.

September 2012
 74

159 star
[two characters]
Hitt. 1 Av. 1 Luv. 5 Goth. 1y
Arm. 1 OCS 3 Lyc. 6 ON 1y
Gk. 1 Lith. 3 TA 1x OHG 1y
Alb. 2 OE 1y OPer. 7 Welsh 1
TB 1x OI 4 OPru. 8 Osc. 9
Ved. 1 Lat. 1 Latv. 3 Umb. 10
1 *h2stḗr and derivs. 3 PBS *žvaigždā (vel sim.; see Buck
1x PToch. *ścəәryẹ 1949:56)
1y PGmc. *sternan-
We have employed both codings for superstate 1.

160 stick
Hitt. 1 Av. 7 Luv. 12 Goth. 17
Arm. 2 OCS 8 Lyc. 13 ON 18
Gk. 3 Lith. 9 TA 14 OHG 19
Alb. 4 OE 10 OPer. 15 Welsh 20
TB 5 OI 11 OPru. 9 Osc. 21
Ved. 6 Lat. 3 Latv. 16 Umb. 22
3 *bak- 9 PBalt. *lazdā

161 stone
Hitt. 1 Av. 6 Luv. 10 Goth. 7
Arm. 2 OCS 6 Lyc. 11 ON 7
Gk. 3 Lith. 6 TA 12 OHG 7
Alb. 4 OE 7 OPer. 6 Welsh 14
TB 5 OI 8 OPru. 13 Osc. 15
Ved. 6 Lat. 9 Latv. 6 Umb. 9
6 *h2éḱmō 9 PItal. *lapid-
7 PGmc. *stainaz
The unexplained velar of the Balto-Slavic forms is part of a well-known larger phenome-
non and does not cast doubt on the coherence of set 6.

September 2012
 75

162 straight
Hitt. 1 Av. 6a Luv. 9 Goth. 6b
Arm. 2 OCS 7 Lyc. 10 ON 6b
Gk. 3 Lith. 8 TA 11 OHG 6b
Alb. 4 [loan] OE 6b OPer. 12 Welsh 14
TB 5 OI 6a OPru. 13 Osc. 15
Ved. 6a Lat. 6b Latv. 8 Umb. 16
6 derivs. of *h3reǵ- ‘put in a straight line’
6a *h3réǵ-u- ~ *h3r̥ǵ-éw- (cpd. in OIr.)
6b verbal adj. *h3reǵtós (with remodelled ablaut)
8 PEBalt. *tiesus
We have coded states 6a, 6b separately, since they are independent derivatives of a verb
root.

163 suck [polymorphic]

[two characters, one monomorphic]
Hitt. 1 Av. 5 Luv. 7 Goth. 12
Arm. 2 OCS 3b Lyc. 8 ON 3c
Gk. 2 Lith. 6 TA 9 OHG 3c
Alb. 3a OE 3c OPer. 10 Welsh 2/3d
TB 4 OI 2 OPru. 11 Osc. 13
Ved. 2 Lat. 3c Latv. 6 Umb. 14
h
2 *d eh1- 6 PEBalt. *žind-
3 *seuK- / *sūK-, where *K is some palatal stop
3c *sūǵh-
Old Irish súgid is probably a Latin loanword, to judge from its “weak” inflection (see
Thurneysen 1946:574, Quin et al. 1983 s.v. súigid).
We have employed both codings for superstate 3, since the reasons for the discrepancy in
the root-final stop are very unclear. In the narrower coding the character is monomor-
phic, since state 3d is unique.
The polymorphism is very extensive but leaf-connected (in Celtic).

September 2012
 76

164 sun [polymorphic]

[two characters, one monomorphic]
Hitt. 1 Av. 3 Luv. 4b Goth. 3/3y
Arm. 2 OCS 3 Lyc. 7 ON 3
Gk. 3 Lith. 3x TA 5 OHG 3y
Alb. 4a OE 3y OPer. 8 Welsh 3
TB 5 OI 6 OPru. 3x Osc. 9
Ved. 3 Lat. 3 Latv. 3x Umb. 10
3 *seh2wel- and derivs. 4a–b derivs. of *diw- ‘sky’
3x PBalt. *saulē 5 PToch. *kawnəә
3y PGmc. *sunnōn-
We have coded states 4a, 4b separately, since they are independent derivatives, but we
employ both codings for superstate 3, since direct replacement of the inherited state by
the derived states is highly likely; under the broader coding the polymorphism of Gothic
disappears.

165 swell
Hitt. 1 Av. 5 Luv. 11 Goth. 17
Arm. 2 OCS 6 Lyc. 12 ON 8
Gk. 2 Lith. 7 TA 13 OHG 8
Alb. 3 OE 8 OPer. 14 Welsh 18
TB 4 OI 9 OPru. 15 Osc. 19
Ved. 5 Lat. 10 Latv. 16 Umb. 20
2 *h3eyd- 8 PNWGmc. *swillidi (*swella-)
5 PIIr. *ćvay-

166 swim
Forms a polymorphic set with 42 fall (q.v.), etc.

167 tail
Hitt. 1 Av. 7 Luv. 12 Goth. 18
Arm. 2 OCS 8 Lyc. 13 ON 19
Gk. 3a Lith. 9 TA 14 OHG 10
Alb. 4 OE 10 OPer. 15 Welsh 20
TB 5 OI 3b OPru. 16 Osc. 21
Ved. 6 Lat. 11 Latv. 17 Umb. 22

September 2012
 77

3a–b derivs. of *órsos ‘arse’ 10 PGmc. *taglą ‘tail-hair (of horses)’

We have coded states 3a, 3b separately, since it is clear that they are very different inde-
pendent derivatives; the character is therefore uninformative, as it exhibits only one non-
unique state.

168 that
Forms a polymorphic set with 70 he (q.v.), etc.
169 there
Forms a polymorphic set with 75 here (q.v.).
170 they
Forms a polymorphic set with 70 he (q.v.), etc.

171 thick
Hitt. 1 Av. 6 Luv. 11 Goth. 17
Arm. 2 OCS 7 Lyc. 12 ON 9
Gk. 3 Lith. 8 TA 13 OHG 9
Alb. 4 OE 9 OPer. 14 Welsh 9
TB 5 OI 9 OPru. 15 Osc. 18
Ved. 3 Lat. 10 Latv. 16 Umb. 19
3 *bhénǵhus ~ *bhn̥ǵhéw- 9 *tégus
h
The relation between Vedic st ūrás and Lithuanian stóras is remote: the root-shapes do
not match, and a parallel semantic development *‘coagulated’ → *‘dense’ → ‘thick’
must probably be reconstructed. We have therefore coded them with unique states (the
unique Vedic state being automatically suppressed because Vedic also exhibits a shared
state for this character). Whether Armenian stowar is made to the same root is even more
uncertain; see Clackson 1994:43 for discussion.

172 thin
Hitt. 1 Av. 7 Luv. 9 Goth. 14
Arm. 2 OCS 6 Lyc. 10 ON 6
Gk. 3 Lith. 8 TA 11 OHG 6
Alb. 4 OE 6 OPer. 12 Welsh 6
TB 5 OI 6 OPru. 13 Osc. 15
Ved. 6 Lat. 6 Latv. 6 Umb. 16
6 *ténh2us ~ *tn̥h2éw- and derivs.

September 2012
 78

173 think [polymorphic]

Hitt. 1 Av. 6 Luv. 6 Goth. 9
Arm. 2 OCS 7 Lyc. 13 ON 9
Gk. 3 Lith. 8a TA 5 OHG 10
Alb. 4 [loan] OE 9/10 OPer. 6 Welsh 15
TB 5 OI 11 OPru. 14 Osc. 16
Ved. 6 Lat. 12 Latv. 8b Umb. 17
5 PToch. *pəәlskna(ṣəә) 9 PGmc. *hugiþi (*hugja-; pret. *hugdē)
6 *ményetor 10 PGmc. *þankīþi
8a–b derivs. of PBS *dūma- ~ *dōuma-
On the Luvian word see Melchert 1994:169, 275.
The East Baltic words are quite different derivatives of a Balto-Slavic stem meaning
originally ‘judgment’ or the like; we have therefore coded them separately.
The polymorphism is confined to West Germanic and is leaf-connected.

174 this
Forms a polymorphic set with 70 he (q.v.), etc.

175 thou and 175a thee [polymorphic set]

Hitt. 1, 2 Av. 1, 2 Luv. 1, 2 Goth. 1, 2y
Arm. 1, 2 OCS 1, 2[dat.] Lyc. 3, 4 ON 1, 2y
Gk. 1, 2 Lith. 1, 2x TA 1, 2[dat.] OHG 1, 2y
Alb. 1, 2 OE 1, 2[dat.] OPer. 1, 2 Welsh 2, 2
TB 1, 2[dat.] OI 1, 2 OPru. 1, 2x Osc. 1, 5
Ved. 1, 2 Lat. 1, 2 Latv. 1, 2z Umb. 6, 2
1 nom. *túh2 (on the Anatolian forms see Melchert 1994:84)
2 acc. *twé ~ *te, or dative (variously formed)
2x extended stem *te-n-
2y emphatic *tége
2z extended stem *te-w-
The accusative form current in the late West Saxon dialect of Old English is the inherited
dative; other dialects of Old English preserve the inherited Germanic accusative 2y.
We have coded the substates of 2 separately, but we have not coded the datives separate-
ly, since the use of a dative personal pronoun for the accusative is an easy parallel devel-
opment.
We also have not coded the appearance of the emphatic particle *-ém ~ *-óm, which

September 2012
 79

exhibits no recognizable pattern. Note that the final -g of the Hittite accusative cannot
reflect the emphatic particle *-ge (which would have become “-gi”).

176 three
Hitt. 1 Av. 1 Luv. 1 Goth. 1
Arm. 1 OCS 1 Lyc. 1 ON 1
Gk. 1 Lith. 1 TA 1 OHG 1
Alb. 1 OE 1 OPer. 1 Welsh 1
TB 1 OI 1 OPru. 1 Osc. 1
Ved. 1 Lat. 1 Latv. 1 Umb. 1
1 *tréyes, fem. *tisrés, neut. *tríh2

177 throw
Hitt. 1 Av. 1 Luv. 10 Goth. 7
Arm. 2 OCS 6 Lyc. 11 ON 7
Gk. 3 Lith. 6 TA 12 OHG 7
Alb. 4 OE 7 OPer. 1 Welsh 13
TB 5 OI 8 OPru. 6 Osc. 14
Ved. 1 Lat. 9 Latv. 6 Umb. 15
1 *sh1iéti (cf. Melchert 1994:154) 7 PGmc. *wirpidi (*werpa-)
6 PBS *meteti

178 tie [polymorphic]

Hitt. 1 Av. 1/6 Luv. 1 Goth. 6
Arm. 2 [loan] OCS 8 Lyc. 11 ON 6
Gk. 3 Lith. 9 TA 5 OHG 6
Alb. 4 OE 6 OPer. 6 Welsh 10
TB 5 OI 7/10 OPru. 9 Osc. 13
Ved. 1/3/6/7 Lat. 4 Latv. 1 Umb. 14
1 *sh2ey- 5 PToch. *kəәrk- 9 PBalt. *reiš- ~ *riš-
3 *deh1- h h
6 *b end - 10 PCelt. *reig- ~ *rig-
4 *liǵ- 7 *nedh- (or *nadh-)
The polymorphism of this character is extensive and complex. Our remarks on 27 ‘cut’
(see above) apply here as well.

September 2012
 80

179 tongue
[two characters]
Hitt. 1 Av. 2uu Luv. 1 Goth. 2x
Arm. 2s OCS 2v Lyc. 5 ON 2x
Gk. 3 Lith. 2w TA 2t OHG 2x
Alb. 4 OE 2x OPer. 2uu Welsh 2y
TB 2t OI 2y OPru. 2v Osc. 2z
Ved. 2u Lat. 2 Latv. 6 Umb. 7
1 PAnat. *lalos
2 *dn̥ǵhwéh2s ~ *dn̥ǵhuh2-, variously deformed (see Peters 1991):
2s, 2w deformed by lexical analogy with ‘lick’
2t stops metathesized (Ringe 1996:45-6)
2u PIIr. *źiźhvā́
2uu initial consonant altered further by dissimilation
2v *d- lost
2x extended as an n-stem
2y *d- replaced by t-
2z manner of articulation of the stops metathesized
We employ all three codings for superstate 2 (i.e., all the substates coded together; only
2u and 2uu together; and all coded separately).

180 tooth [with parallel development]

Hitt. 1 Av. 2 Luv. 4 Goth. 2
Arm. 2x OCS 3 Lyc. 5 ON 2
Gk. 2 Lith. 2 TA 3 OHG 2
Alb. 3 OE 2 OPer. 6 Welsh 2
TB 3 OI 2 OPru. 2 Osc. 2
Ved. 2 Lat. 2 Latv. 3 Umb. 7
2 *h1dónts ~ *h1dn̥t- h
3 *ǵómb os ‘row of teeth’
2x suffix altered (see Olsen 1999:505 with references)
We have coded state 2x together with state 2.
The parallel development in this character consists in the multiple replacement of state 2
with state 3; note that even the East Baltic subgroup is split.

September 2012
 81

181 tree
[two characters]
Hitt. 1 Av. 5 Luv. 9 Goth. 14a
Arm. 2 OCS 1 Lyc. 10 ON 1x
Gk. 1 Lith. 6 TA 3 OHG 14b
Alb. 1 OE 1x OPer. 11 Welsh 7
TB 3 OI 7 OPru. 12 Osc. 15
Ved. 4 Lat. 8 Latv. 13 Umb. 16
1 *dóru ~ *dréw- and derivs. 14 Gmc. *baXmaz
1x PGmc. *trewą 14a pre-Gothic *bagmaz
3 PToch. *staməә 14b PWGmc. *baumaz
7 PCelt. *kwrennos
State 14 is problematic (cf. also ON baðmr). If it was really a PGmc. word, then it
appears that there was widespread polymorphism in Germanic, though each of the
languages settled on one of the words as basic for ‘tree’.
We employ both codings for superstate 1 but code states 14a, 14b separately.

182 turn
Hitt. 1 Av. 7 Luv. 11 Goth. 9
Arm. 2 OCS 8 Lyc. 12 ON 13
Gk. 3 Lith. 6 TA 5 OHG 9
Alb. 4 OE 9 OPer. 6 Welsh 14
TB 5 OI 10 OPru. 6 Osc. 15
Ved. 6 Lat. 6 Latv. 6 Umb. 6
5 PToch. *spartwa- 9 PGmc. *wandīþi
6 *wert-

183 two
Hitt. 1 Av. 1 Luv. 2 Goth. 1
Arm. 1 OCS 1 Lyc. 1 ON 1
Gk. 1 Lith. 1 TA 1 OHG 1
Alb. 1 OE 1 OPer. 1 Welsh 1
TB 1 OI 1 OPru. 1 Osc. 3
Ved. 1 Lat. 1 Latv. 1 Umb. 1
1 *duóh1

September 2012
 82

184 vomit
Hitt. 1 Av. 3 Luv. 9 Goth. 13
Arm. 2 OCS 6 Lyc. 10 ON 7
Gk. 3 Lith. 3 TA 11 OHG 7
Alb. 4 OE 7 OPer. 12 Welsh 8
TB 5 OI 8 OPru. 3 Osc. 14
Ved. 3 Lat. 3 Latv. 3 Umb. 15
3 *wémh1ti 7 PGmc. *spīwidi 8 PCelt. *skei-

185 walk
Forms a polymorphic set with 25 come (q.v.) and 343 go.

186 warm [polymorphic]

Hitt. 1 Av. 2/6 Luv. 9 Goth. 8
Arm. 2 OCS 6 Lyc. 10 ON 8
Gk. 2 Lith. 7 TA 11 OHG 8
Alb. 3 OE 8 OPer. 12 Welsh 6
TB 4 OI 6 OPru. 13 Osc. 14
Ved. 5 Lat. 6 Latv. 7 Umb. 15
2 *gwhr̥mós (with ablaut variously remodelled)
6 derivs. of *tep- 8 PGmc. *warmaz
7 PEBalt. *šiltas
The polymorphism is extensive but leaf-connected (at Avestan). Alternatively, it is pos-
sible that parallel development has occurred, since both *tep- and *gwher- are verb roots.

187 wash [polymorphic]

Hitt. 1 Av. 3 Luv. 9 Goth. 7
Arm. 2 OCS 5 Lyc. 10 ON 7
Gk. 2/3 Lith. 6 TA 4 OHG 8
Alb. 2 OE 7/8 OPer. 11 Welsh 12
TB 4 OI 3 OPru. 5 Osc. 13
Ved. 3 Lat. 2 Latv. 6 Umb. 2
2 *lewh3- 5 PBS *mū- 8 PWGmc. *waskidi
3 *neygw- 6 PEBalt. *mazgā-
4 PToch. *lika- 7 PGmc. *þwahidi
The West Germanic polymorphism is local and leaf-connected; the polymorphism invol-

September 2012
 83

ving states 2 and 3 is extensive but still leaf-connected (at Greek).

188 water [polymorphic, with parallel development]

Hitt. 1 Av. 4 Luv. 3 Goth. 1
Arm. 2 OCS 1 Lyc. 6 ON 1
Gk. 1 Lith. 1 TA 1 OHG 1
Alb. 1 OE 1 OPer. 4 Welsh 7
TB 1 OI 1 OPru. 1 Osc. 5
Ved. 1/3 Lat. 5 Latv. 1 Umb. 1
1 *wódr̥ ~ *udén- and derivs. 4 *h2ep- ‘running water’
3 *wéh1r̥ ‘liquid’ (Watkins 1987:401-3) 5 *akweh2 ~ *ākweh2 ‘running
water’
We adopt the hypothesis that the Oscan word is cognate with the Latin rather than with
the Iranian forms.
The polymorphism is confined to the unusual suppletive paradigm of Vedic.
The reconstructable meanings of states 3 (see Watkins ad loc.), 4 (cf. Vedic ā́pas ‘the
(divine) Waters’), and 5 (cf. PGmc. *ahwō ‘river’) strongly argue parallel development.

189 we AND 189a us [polymorphic set]

Hitt. 1, 2 Av. 1, 2 Luv. 3, 2 Goth. 1, 2
Arm. 1x, 2 OCS 1y, 2 Lyc. 4, 5 ON 1, 2
Gk. 2, 2 Lith. 1y, 1y TA 1+2, 1+2 OHG 1, 2
Alb. 2, 2 OE 1, 2 OPer. 1, 6 Welsh 2, 2
TB 1+2, 1+2 OI 2, 2 OPru. 1y, 1y Osc. 7, 8
Ved. 1, 2 Lat. 2, 2 Latv. 1y, 1y Umb. 9, 10
1 nom. *wéy and subsequent developments
1x, 1y *w- replaced by *m-
2 acc. *n̥smé, enclitic *nos, and subsequent developments
The replacement of *w- by *m- can be a parallel development, at least in part (cf. also
Pali mayam ‘we’, unarguably a direct reflex of Vedic vayám); we have therefore coded
the Armenian and Balto-Slavic nominatives separately. An obvious model (still surviv-
ing in Old Church Slavonic) is the verb endings, in which the 1du. begins with -v- but the
1pl. with -m-.
The choice of stressed or enclitic forms for generalization in the oblique is clearly a re-
peatable change; therefore we have not distinguished them in coding.
We have coded the Tocharian forms, in which the initial consonant reflects the nomina-

September 2012
 84

tive and the rest of the form the oblique enclitics, separately.
On the Celtic and Armenian reflexes of state 2 see Katz 1998:96-105, 186-8.

190 wet [polymorphic]

Hitt. 1 Av. 7 Luv. 14 Goth. 19
Arm. 2 OCS 8 Lyc. 15 ON 10
Gk. 3 Lith. 9 TA 16 OHG 11/19
Alb. 4 OE 10/11 OPer. 17 Welsh 12
TB 5 OI 12 OPru. 18 Osc. 20
Ved. 6 Lat. 13 Latv. 9 Umb. 21
9 PEBalt. *šlapjas w
12 PCelt. *wlik us
10 PNWGmc. *wātiz 19 PGmc. *nataz
11 PWGmc. *fūht-
The polymorphism is confined to Germanic. A triple state 10/11/18 must be posited for
Proto-West Germanic, and an otherwise unattested double state 10/18 for Proto-North-
west Germanic.

191 what
Hitt. 1 Av. 1 Luv. 1 Goth. 1
Arm. 1 OCS 1 Lyc. 1 ON 1
Gk. 1 Lith. 1 TA 1 OHG 1
Alb. 1 OE 1 OPer. 2 Welsh 1
TB 1 OI 1 OPru. 1 Osc. 1
Ved. 1 Lat. 1 Latv. 1 Umb. 1
w w
1 *k íd and/or adj. *k ód (and subsequent developments)
Since the choice between the two stems is clearly a repeatable change, we have not
distinguished them in coding.

192 when [polymorphic]

[two characters, one monomorphic]
Hitt. 1a Av. 1e Luv. 1j Goth. 1g
Arm. 1b OCS 1f Lyc. 3 ON 1g
Gk. 1c Lith. 1e TA 1k/2 OHG 1g
Alb. 1d OE 1g OPer. 4 Welsh 1l
TB 2 OI 1h OPru. 1e Osc. 5
Ved. 1e Lat. 1i Latv. 1e Umb. 6

September 2012
 85

1a–l derivs. of *kwi- ~ *kwo- ~ *ku-

1e satem *kwod + vowel-initial particle
1g PGmc. *hwan(-)
2 PToch. *ënt- ~ *əәnt-
We employ both codings for superstate 1; in the narrower coding the unique substate 1k
is automatically suppressed, making the character monomorphic.

193 where [with parallel development]

Hitt. 1a Av. 1e Luv. 1i Goth. 1g
Arm. 1b OCS 1f Lyc. 1j ON 1g
Gk. 1c Lith. 1b TA 3 OHG 1g
Alb. 1d OE 1g OPer. 4 Welsh 1l
TB 2 OI 1h OPru. 1k Osc. 1f
Ved. 1e Lat. 1f Latv. 1b Umb. 1f
w w
1a–l derivs. of *k i- ~ *k o- ~ *ku-
1b *kur(-) 1f *kudhé(±i)
1e PIIr. *kútra 1g PGmc. *hwar
All derivatives have been coded separately, since otherwise the character is uninforma-
tive.
It seems clear that the choice between states 1b and 1f, or perhaps the formation of the
former, was a repeatable change.

194 white [polymorphic]

[two monomorphic characters by conflated split coding]
Hitt. 1 Av. 5a Luv. 9 Goth. 5b
Arm. 2 [loan] OCS 6a Lyc. 10 ON 5b
Gk. 3 Lith. 6b TA 1 OHG 5b
Alb. 4 OE 5b OPer. 11 Welsh 7
TB 1 OI 7 OPru. 12 Osc. 13
Ved. 1/5a Lat. 8 Latv. 6b Umb. 8
1 *h2erǵ- 6 derivs. of PBS *bel-
5 derivs. of *ḱwey- 6a pre-Slavic *bēlas
5a PIIr. *ćvaitas 6b P(E)Balt. *baltas
5b PGmc. *hwītaz 7 PCelt. *windos
8 PItal. *alfos < *albhós
We have coded states 5a, 5b and states 6a, 6b separately, since they are clearly indepen-

September 2012
 86

dent derivatives.
The polymorphism is confined to Indo-Iranian and is leaf-connected. We have reduced
this to two monomorphic characters by conflated split coding, with 5a ∪ 1/5a coded
against 1 in one character and all three coded together in the other.

195 who
Hitt. 1 Av. 1 Luv. 1 Goth. 1
Arm. 1 OCS 1 Lyc. 1 ON 1
Gk. 1 Lith. 1 TA 1 OHG 1
Alb. 1 OE 1 OPer. 1 Welsh 1
TB 1 OI 1 OPru. 1 Osc. 1
Ved. 1 Lat. 1 Latv. 1 Umb. 1
w w
1 *k éy and/or adj. *k ós (and subsequent developments)
On the Armenian form see Olsen 1999:806.
Since the choice between the two stems is clearly a repeatable change, we have not
distinguished them in coding.

196 wide [polymorphic]

[two characters, both polymorphic]
Hitt. 1a Av. 1b Luv. 8 Goth. 5
Arm. 1bx OCS 4 Lyc. 9 ON 5/6
Gk. 2 Lith. 1b TA 2x OHG 5/6
Alb. 3 OE 5/6 OPer. 10 Welsh 1bx
TB 2x OI 1bx OPru. 11 Osc. 12
Ved. 1b/2 Lat. 7 Latv. 1b Umb. 13
1a *pl̥h2- 5 PGmc. *braidaz
1b *pláth2us ~ *pl̥th2éw- and derivs. 6 PNWGmc. *wīdaz
1bx *pl̥th2n̥ós (cf. Olsen 1999:767)
2 *h1wérus ~ *h1uréw- and derivs.
2x PToch. extension with *-tstsë
We have coded states 1a and 1b separately, since they exhibit different root-shapes
(though some ultimate etymological connection is probable); but we have adopted both
codings for (super)state 1b, since it is at least a plausible hypothesis that the derivative in
*-nó- replaced the Caland adjective directly, and also for superstate 2.
The NWGmc. polymorphism is local and leaf-connected. The polymorphism between
states 1b and 2, as coded here, is extensive but leaf-connected (at Vedic); in fact, it can be

September 2012
 87

shown that it is even more extensive than our data suggest. For instance, though πλατύς
means ‘flat’ in Attic Greek (which is the basis of our coding), it still means ‘wide’ in
Homer, who also uses εὐρύς; thus the polymorphism extends at least to Greek as well.

197 wife
Hitt. 1 Av. 3 Luv. 6 Goth. 1
Arm. 1 OCS 1 Lyc. 7 ON 1
Gk. 1 Lith. 4 TA 1 OHG 1
Alb. 2 OE 1 OPer. 8 Welsh 10
TB 1 OI 1 OPru. 1 Osc. 11
Ved. 1 Lat. 5 Latv. 9 Umb. 12
1 *gwḗn ~ *gwénh2- ~ *gwnéh2- ‘woman’ and derivs.

198 wind
[two characters]
Hitt. 1a Av. 1c Luv. 5 Goth. 1b
Arm. 2 OCS 1d Lyc. 6 ON 1b
Gk. 2 Lith. 1e TA 1b OHG 1b
Alb. 3 OE 1b OPer. 7 Welsh 1b
TB 1b OI 4 OPru. 1d Osc. 8
Ved. 1c Lat. 1b Latv. 1e Umb. 9
1 derivs. of *h2weh1- ‘blow’
1a *h2wéh1-n̥t-s (vel sim., Melchert 1994:54 with references)
1b “post-laryngeal” *h2wēntós
1c PIIr. *váatas < *h2wéh1-n̥t-o-s
1d PBS *vētras
1e PEBalt. *vējas
2 *h2ónh1mos ‘breath’ (Olsen 1999:27; ablaut adjusted in Greek)
We have adopted both codings for superstate 1, since it is very likely that 1a is the inher-
ited word and the other derivatives of ‘blow’ replaced it without the mediation of deriva-
tives of any other root. The resulting impression that state 1c replaced 1b is an illusion so
far as the phonology is concerned—that is, the apparently uniform *-ēn- of state 1b re-
flects parallel development—but not necessarily in morphological terms.

September 2012
 88

199 wing [polymorphic, but not effectively so]

[two characters]
Hitt. 1/2a Av. 1x Luv. 8 Goth. 13
Arm. 3 OCS 6 Lyc. 9 ON 14
Gk. 1+2 Lith. 1y TA 10 OHG 2b
Alb. 4 OE 2b OPer. 11 Welsh 2e
TB 5 OI 2c OPru. 12 Osc. 15
Ved. 1x Lat. 7 Latv. 1y Umb. 16
1 derivs. of *per- 2a–d derivs. of *pet(h2)- ‘fly’
1x *pernóm 2b derivs. of PNWGmc. *feþru
1y PEBalt. *sparnas ‘feather’
We have adopted both codings for superstate 1, but have coded the states of 2 separately,
since they are clearly independent derivatives of a verb root. The unique state 2a is thus
automatically suppressed, eliminating the polymorphism. The Greek form shows con-
tamination of the two roots; we have treated it as a substate of 1 in coding.
We reject the connection of the Armenian word with state 2 posited by Olsen 1999:51-2.

200 wipe
Hitt. 1 Av. 7 Luv. 1 Goth. 17
Arm. 2 OCS 8 Lyc. 13 ON 18
Gk. 3 Lith. 9 TA 5 OHG 17
Alb. 4 OE 10 OPer. 14 Welsh 19
TB 5 OI 11 OPru. 15 Osc. 20
Ved. 6 Lat. 12 Latv. 16 Umb. 21
1 PAnat. *ómsei 17 PGmc. *swirbidi (*swerba-)
5 PToch. *lyiyask-

201 with [polymorphic]

Hitt. 1 Av. 6/7 Luv. 11 Goth. 9
Arm. 2 OCS 8 Lyc. 12 ON 9
Gk. 3 Lith. 8 TA 5 OHG 9
Alb. 4 OE 9 OPer. 7 Welsh 14
TB 5 OI 10 OPru. 8 Osc. 10
Ved. 6/7 Lat. 10 Latv. 13 Umb. 10

September 2012
 89

5 PToch. *śəәlë 8 PBS *sVn

6 PIIr. *smát 9 *medhi or *metí
7 PIIr. *sadhá 10 *kom
It is not clear that the Balto-Slavic words belong with the Greek form. The latter was
originally ξύν (attested in various dialects); the vowel of the PBS word is problematic (cf.
Stang 1966:32).
It is also not clear whether the Albanian word belongs with state 6 or state 9 (cf. Demiraj
1997:274-5, but also 55!); we have cautiously assigned it a unique state.
The polymorphism is confined to Indo-Iranian and is leaf-connected.

202 woman
Hitt. 1 Av. 1 Luv. 1 Goth. 1
Arm. 1 OCS 1 Lyc. 7 ON 1
Gk. 1 Lith. 4 TA 3 OHG 5
Alb. 2 OE 5 OPer. 8 Welsh 1
TB 3 OI 1 OPru. 1 Osc. 10
Ved. 1 Lat. 6 Latv. 9 Umb. 11
1 *gwḗn ~ *gwénh2- ~ *gwnéh2- and derivs. 5 PWGmc. *wīb
3 PToch. *kwlyiye

203 woods [polymorphic]

Hitt. 1 Av. 7 Luv. 13 Goth. 17
Arm. 2 OCS 8 Lyc. 14 ON 10
Gk. 3 Lith. 9 TA 5 OHG 11
Alb. 4 OE 10/11 OPer. 15 Welsh 10
TB 5 OI 10 OPru. 16 Osc. 18
Ved. 6 Lat. 12 Latv. 16 Umb. 19
5 PToch. *wəәrtto 11 PWGmc. *wald
h
10 *wid us 16 PBalt. *medjan
Note that Tocharian A kārāś is a loanword.
The polymorphism is confined to West Germanic and is leaf-connected.

September 2012
 90

204 worm [polymorphic]

Hitt. 1 Av. 6 Luv. 8 Goth. 13
Arm. 2 OCS 4 Lyc. 9 ON 7/13
Gk. 3 Lith. 4 TA 5 OHG 7
Alb. 4 OE 7 OPer. 10 Welsh 4
TB 5 OI 4 OPru. 11 Osc. 14
Ved. 4 Lat. 7 Latv. 12 Umb. 15
4 *kwr̥mis 7 *wr̥mis
y
5 PToch. *w eləә 13 PGmc. *maþ-
The polymorphism is confined to Germanic and is leaf-connected.

205 ye AND 205a you [polymorphic set]

Hitt. 1, 1 Av. 2, 1 Luv. 5, 1 Goth. 2, 1
Arm. 2, 3 OCS 1, 1 Lyc. 6, 7 ON 2, 1
Gk. 1, 1 Lith. 2, 2 TA 2+1, 2+1 OHG 2, 1
Alb. 4, 4 OE 2, 1 OPer. 8, 9 Welsh 1, 1
TB 2+1, 2+1 OI 1, 1 OPru. 2, 1 Osc. 1, 10
Ved. 2, 1 Lat. 1, 1 Latv. 2, 2 Umb. 11, 12
1 acc. *uswé, enclitic *wos, and subsequent developments
2 nom.*yū́ and subsequent developments
Our coding relies heavily on the enlightening discussion of these forms in Katz 1998.
We regard the Armenian oblique stem as still not satisfactorily explained and have there-
fore assigned it a unique state. On the difficulties surrounding the interpretation of Alba-
nian ju, which we have also coded separately, see Demiraj 1997:209 with references.

206 year [polymorphic]

[two characters, both polymorphic]
Hitt. 1 Av. 4 Luv. 1x Goth. 4x/8
Arm. 2 OCS 5 Lyc. 1x ON 4x
Gk. 1 Lith. 6 TA 3 OHG 4x
Alb. 1 OE 4x OPer. 9 Welsh 7
TB 3 OI 7 OPru. 6 Osc. 8
Ved. 1 Lat. 8 Latv. 10 Umb. 8

September 2012
 91

1 *wet- and derivs. 6 PBalt. *metan

1x PLuv. *utsis 7 PCelt. *bleidanī
y
3 PToch. *p əәkwəәl 8 *atnos
4 *yóh1r̥ ‘period of time’ (cf. Greek ὥρᾱ)
4x PGmc. *jērą
We have employed both codings for superstate 1, but have coded states 4 and 4x sepa-
rately, since they seem to be parallel developments of a word of originally different
meaning.
The polymorphism is confined to Germanic and is leaf-connected.

207 yellow
[two characters]
Hitt. 1 Av. 6x Luv. 1 Goth. 12
Arm. 2 OCS 6 Lyc. 9 ON 6
Gk. 3 Lith. 6y TA 10 OHG 6z
Alb. 4 [loan] OE 6z OPer. 11 Welsh 13
TB 5 OI 7 OPru. 6y Osc. 14
Ved. 6x Lat. 8 Latv. 6y Umb. 15
1 PAnat. *HaHl̥went-
6 derivs. of *ǵhel-
6x PIIr. *źhális
6y PBalt. *gelt-
6z PWGmc. *gelw
The unexplained velar of the Baltic forms is part of a well-known larger phenomenon and
does not cast doubt on the coherence of set 6.
We have adopted both codings for superstate 6.

301 arm
Hitt. 1 Av. 5 Luv. 11 Goth. 8
Arm. 2 [loan] OCS 6 Lyc. 12 ON 8
Gk. 3 Lith. 7 TA 5 OHG 8
Alb. 4 OE 8 OPer. 13 Welsh 14 [loan]
TB 5 OI 9 OPru. 8 Osc. 15
Ved. 5 Lat. 10 [loan] Latv. 7 Umb. 16
5 *bhāǵhus 8 *arHmos ~ *r̥Hmos ‘joint’
7 PBS *rankā ‘hand’

September 2012
 92

302 arrow [polymorphic, with parallel development]

Hitt. 1 Av. 5 Luv. 10 Goth. 7
Arm. 1 OCS 6 Lyc. 11 ON 7
Gk. 2 Lith. 6 TA 4 OHG 6
Alb. 3 [loan] OE 6/7 OPer. 12 Welsh 15 [loan]
TB 4 OI 8 [loan] OPru. 13 Osc. 16
Ved. 5 Lat. 9 Latv. 14 Umb. 17
1 *ned- ~ *nod- ‘reed’ 6 *streh1leh2
4 PToch. *pəәrë- (?) 7 PGmc. *arhwō
5 *ísus
The polymorphism is leaf-connected. Both the meaning of the preform for state 1 and the
distribution of state 6 (unless the latter reflects borrowing) suggest parallel development.

303 ask [polymorphic]

[two characters, both polymorphic]
Hitt. 1 Av. 2 Luv. 8 Goth. 2x
Arm. 2 OCS 2w Lyc. 9 ON 2y
Gk. 3 Lith. 5 TA 2v OHG 2z/6
Alb. 4 OE 2x/6 OPer. 2 Welsh 11
TB 2v OI 2 OPru. 10 Osc. 12
Ved. 2 Lat. 7 Latv. 2w Umb. 13
2 *preḱ-, pres. *pr̥sḱéti 6 PWGmc. *aiskōþi
2v PToch. pres. *prëks-
2w intensive *proḱéyeti
2x PGmc. pres. *frigniþi (*fregna-)
2y, 2z other derivs.
The original present is extended in Armenian and compounded in Old Irish; we have not
coded those forms separately.
We employ both codings for superstate 2.
The polymorphism is confined to West Germanic and is leaf-connected.

September 2012
 93

304 autumn
Hitt. 1 Av. 7 Luv. 12 Goth. 3
Arm. 2 OCS 3 Lyc. 13 ON 16
Gk. 3 Lith. 8 TA 14 OHG 9
Alb. 4 OE 9 OPer. 15 Welsh 17
TB 5 OI 10 OPru. 3 Osc. 18
Ved. 6 Lat. 11 Latv. 8 Umb. 19
3 *ósr̥, *esen- and derivs. 9 PWGmc. *har(u)bist
8 PEBalt. *ruden-

305 be awake
Hitt. 1 Av. 3 Luv. 9 Goth. 7a
Arm. 2 OCS 6 Lyc. 10 ON 7a
Gk. 3 Lith. 6 TA 11 OHG 7a
Alb. 4 OE 7a OPer. 12 Welsh 14
TB 5 OI 8 OPru. 6 Osc. 15
Ved. 3 Lat. 7b Latv. 13 Umb. 16
3 *h1geh1góre 7 derivs. of *weǵ- (*weg-?)
6 PBS *bud- < *bhudh- ‘wake up’ 7a PGmc. *wakaiþi
7b deriv. of Lat. vigil ‘awake’
On the Gothic form see Krause 1953:230.
We have coded states 7a, 7b separately.

306 ax [polymorphic, with parallel development]

Hitt. 1 Av. 7 Luv. 12 Goth. 3
Arm. 2 [loan] OCS 8/9 Lyc. 13 ON 3
Gk. 3/4 Lith. 10 TA 6 [loan] OHG 3/9
Alb. 5 OE 3 OPer. 14 Welsh 11a
TB 6 [loan] OI 11a OPru. 11b Osc. 15
Ved. 4 Lat. 8 Latv. 10 Umb. 16
3 *agwes- 9 derivs. of *bharsdheh2 ‘beard’
4 *peleḱus 10 PEBalt. *kirvjas
6 PToch. *peretəә 11 derivs. of *bheyH- ‘strike’
8 derivs. of *sek- ‘cut’ with 11a PCelt. *bielis (vel sim.)
unusual suffix *-ūr- 11b *bhiH-l-
We have coded states 11a, 11b separately, since they are derivationally distinct.

September 2012
 94

The shape of the Celtic suffix remains unclear, as the attested forms do not match. The
Greek reflex of state 3 is also anomalous (we expect -ψ-, not -ξ-).
TheTocharian word was borrowed from some northeast Iranian reflex of state 4 before
the Proto-Tocharian period; thus we assign both Tocharian languages the same state, but
a different state from 4.
The polymorphism of this character is very complex and extensive. The reconstructable
meanings of states 8, 9, and 11 strongly argue parallel development; further, the phono-
logical anomalies of some reflexes (see above) and the odd shapes of the preforms for
states 3 and 4 suggest undetected borrowing of words denoting an important trade and
culture item.

307 axle
Hitt. 1 Av. 6 Luv. 8 Goth. 12
Arm. 2 OCS 3 Lyc. 9 ON 3
Gk. 3 Lith. 3 TA 10 OHG 3
Alb. 4 OE 3 OPer. 11 Welsh 3
TB 5 OI 7 OPru. 3 Osc. 13
Ved. 3 Lat. 3 Latv. 3 Umb. 14
3 *h2eǵ-s-

308 be [polymorphic]
[two characters, both polymorphic]
Hitt. 1 Av. 1 Luv. 1 Goth. 1/5
Arm. 1 OCS 1 Lyc. 1 ON 1/5
Gk. 1 Lith. 1/3 TA 1/2/6 OHG 1/4/5
Alb. 1 OE 1/4/5 OPer. 1 Welsh 1
TB 1/2/6 OI 1/4 OPru. 1 Osc. 1/4
Ved. 1 Lat. 1/4 Latv. 1/3 Umb. 1/4
1 *h1ésti, opt. *h1siéh1d, etc. h
4 derivs. of *b uh2- ‘become’
2 PToch. nësəә(ṣəә) 5 PGmc. *wesaną < *h2wes- ‘stay overnight’
3 PEBalt. *īra 6 PToch. *məәsketəәr ‘be (in a place)’
In accordance with our decision regarding suppletive verbs, only forms of the present
stem(s) have been coded, except for Welsh; but this verb is suppletive even in the present
in many languages. In the Italic languages state 4 is represented in the present system by
Lat. foret, Osc. fusíd, etc.; in OIr. it is represented by the consuetudinal present and im-
perative. State 6 reflects a separate verb with partly overlapping functions.

September 2012
 95

We have employed split coding for the polymorphisms 1/2/6 and 1/3, which are confined
to Tocharian and East Baltic respectively.

309 bear
Hitt. 1 Av. 1 Luv. 7 Goth. 11
Arm. 1 OCS 3 Lyc. 8 ON 5
Gk. 1 Lith. 4 TA 9 OHG 5
Alb. 1 OE 5 OPer. 10 Welsh 1
TB 2 OI 6 OPru. 4 Osc. 12
Ved. 1 Lat. 1 Latv. 4 Umb. 13
1 *h2ŕ̥tḱos 5 PNWGmc. *beran-
4 PBalt. *lākijas

310 beard
Hitt. 1 Av. 4 Luv. 7 Goth. 11
Arm. 1 OCS 5 Lyc. 8 ON 12
Gk. 2 Lith. 5 TA 9 OHG 5
Alb. 1 OE 5 OPer. 10 Welsh 13 [loan]
TB 3 OI 6 OPru. 5 Osc. 14
Ved. 1 Lat. 5 Latv. 5 Umb. 15
1 *smáḱru (*sméḱru ?) 5 *b arsdheh2
h

311 bee
Hitt. 1 Av. 5 Luv. 8 Goth. 12
Arm. 2a OCS 6 Lyc. 9 ON 6
Gk. 2b Lith. 6x TA 10 [loan] OHG 6
Alb. 2c OE 6 OPer. 11 Welsh 13
TB 3 OI 6 OPru. 6x Osc. 14
Ved. 4 Lat. 7 Latv. 6x Umb. 15
2a–c derivs. of *mélit ‘honey’ 6 derivs. of *bhey-
6x PBalt. *bitē
We have coded states 2a–c separately, since they appear to reflect independent deriva-
tion. In consequence we have also coded 6 and 6x separately, since if they are coded
together the character is uninformative.

September 2012
 96

312 be born
Hitt. 1 Av. 2 Luv. 8 Goth. 7
Arm. 2 OCS 5 Lyc. 9 ON 7
Gk. 2 Lith. 6 TA 4 OHG 7
Alb. 3 OE 7 OPer. 10 Welsh 2
TB 4 OI 2 OPru. 6 Osc. 2
Ved. 2 Lat. 2 Latv. 6 Umb. 11
2 *ǵenh1-, pres. *ǵn̥h1yétor 6 PBalt. *gem-, pres. *gimsta
4 PToch. *təәmnəәstəәr 7 PGmc. *beradai ‘is carried’ and syntactic
replacements of the same
The original present of state 2 has been replaced in various daughters; since the replace-
ments are unique, we have not coded them separately.

313 bow
[two characters]
Hitt. 1 Av. 6x Luv. 11 Goth. 15
Arm. 2 OCS 7 Lyc. 12 ON 8
Gk. 3 Lith. 7 TA 13 [loan] OHG 8
Alb. 4 [loan] OE 8 OPer. 6x Welsh 16 [loan]
TB 5 OI 9 [loan] OPru. 14 Osc. 17
Ved. 6 Lat. 10 Latv. 7 Umb. 18
6 PIIr. *dhánvr̥ 7 PBS *lankas
6x PIr. *θanvr̥ (θ- ← *θanǰ- ‘draw’) 8 PNWGmc. *bogan-
On the initial consonant of the Iranian form cf. Mayrhofer 1992 s.v. dhánuṣ-. We have
adopted both codings for superstate 6, since it is virtually certain that 6x arose directly
from 6 by analogical replacement of the initial consonant.

314 branch [polymorphic]

Hitt. 1 Av. 8 Luv. 13 Goth. 2
Arm. 2 OCS 7b Lyc. 14 ON 18
Gk. 3 Lith. 6 TA 5 OHG 2/9/10
Alb. 4 OE 9/10 OPer. 15 Welsh 19
TB 5 OI 11 OPru. 16 Osc. 20
Ved. 6/7a Lat. 12 Latv. 17 Umb. 21

September 2012
 97

2 *h3ósdos 7a, 7b derivs of *weyh1- ‘plait’

5 PToch. *kërakəә 9 PWGmc. *telg-
6 *ḱokh2eh2 (*-ō-; orig. collective?) 10 PWGmc. *twig / *twīg
The Tocharian A forms appears to have been extended with a suffix *-ẹye.
Since states 9 and 10 always occur together, we have coded 9/10 as a single state. We
have coded states 7a, 7b separately because they appear to represent independent deriva-
tives.

315 break
[two characters]
Hitt. 1 Av. 6 Luv. 10 Goth. 9a
Arm. 2 OCS 7 Lyc. 11 ON 13
Gk. 3 Lith. 8 TA 5 OHG 9a
Alb. 4 OE 9a OPer. 12 Welsh 14
TB 5 OI 2 OPru. 7 Osc. 15
Ved. 2 Lat. 9b Latv. 8 Umb. 16
h
2 *b eg- h
9 *b rVg-
5 PToch. *kawtna(ṣəә) 9a PGmc. *brikidi (*breka-)
7 *lem- 9b *bhrag-
8 PEBalt. *laužja
We accept the usual view that Old Irish boingid belongs with state 2.
We have adopted both codings for superstate 9, since a direct connection between the two
substates is probable in spite of the puzzling difference in vowels.

316 bronze
Hitt. 1 Av. 7 Luv. 11 Goth. 7
Arm. 2 [loan] OCS 8 Lyc. 12 ON 7
Gk. 3 Lith. 9 [loan] TA 13 OHG 7
Alb. 4 [loan] OE 7 OPer. 14 Welsh 10
TB 5 OI 10 OPru. 15 Osc. 7
Ved. 6 Lat. 7 Latv. 16 [loan] Umb. 7
7 *áyos 10 PCelt. *omiom
It is likely that áyas means ‘bronze’ in several instances in the Rigveda; in that case
Vedic too would share state 7.

September 2012
 98

317 brother
[two characters]
Hitt. 1 Av. 2 Luv. 1 Goth. 2
Arm. 2 OCS 2 Lyc. 1 ON 2
Gk. 3 Lith. 2x TA 2 OHG 2
Alb. 4 OE 2 OPer. 2 Welsh 2
TB 2 OI 2 OPru. 2 Osc. 2
Ved. 2 Lat. 2 Latv. 2x Umb. 2
1 PAnat. *negnas 2 *bhréh2tēr
2x PEBalt. deriv. *brāljas
We have adopted both codings for superstate 2, since it is likely that the East Baltic
innovation (diminutive?) replaced the inherited term directly.

318 brother-in-law
Hitt. 1 Av. 5 Luv. 8 Goth. 13
Arm. 2 OCS 2 Lyc. 9 ON 14
Gk. 2 Lith. 6 TA 10 OHG 15
Alb. 3 OE 2 OPer. 11 Welsh 16
TB 4 OI 7 OPru. 12 Osc. 17
Ved. 2 Lat. 2 Latv. 6 Umb. 18
2 *dayh2wḗr 6 PEBalt. *svainjas
The alternative Lithuanian word is borrowed from German; the Welsh term is a loan-
translation from English (and so for all the in-law terms).

September 2012
 99

319 bull, 322 cattle, 326 cow, AND 379 ox [polymorphic set]
319 bull 379 ox 322 cattle 326 cow
Hitt. 1 2 3 4
Arm. 5 6 7 8
Gk. 9 8 8 8
Alb. 10 10 11 12 [loan]
TB 8+13 14 8/14 8
Ved. 13/14 8 8 8/15
Av. 14 8 8 8
OCS 16 17 8 19
Lith. 20 [loan] 21 22 19
OE 23 14 24/25 8
OI 9 26 8 8
Lat. 9 8 8 15
Luv. 27 28 29 8
Lyc. 30 31 32 8
TA 8+13 14 8 8
OPer. 33 34 35 8
OPru. 36 19 37 38
Latv. 39 [loan] 13 40 8
Goth. 41 14 42 43
ON 44 14 24(+37) 8
OHG 23 14 25 8
Welsh 9 14 45 8
Osc. 9 46 47 48
Umb. 9 49 8 50
8 *gwṓws ‘bovine’, pl. *gwówes 19 PBS *kārv-
9 *tawros ‘bull’ 23 PWG *farr ‘bull’
13 *wr̥sḗn ‘bull’ and derivs. 24 PNWG *nautą ‘bovine’
14 *uksḗn ‘bull’ 25 PWG *hrinþizu ‘cattle’ (pl.)
15 *waḱéh2 ‘cow’ 38 *péḱu ‘cattle’ (coll.)
Though state 10 recurs, it is confined to Albanian.
The East Baltic terms for ‘bull’ appear to have been borrowed from German separately
(not at the Proto-East-Baltic stage). OCS nuta, the alternative term for ‘cattle’, is a
Germanic loanword. On the Albanian words cf. Demiraj 1997 s.vv.
The polymorphism of 319 ‘bull’ could be eliminated by conflated split coding, with

September 2012
 100

13/14 ∪ 14 coded against 8+13 in one character and all three coded together in the other
(on the reasonable hypothesis that 8+13 replaced 13 directly in Tocharian); however, the
character still cannot be used as input to the algorithm, since there is reason to suspect
multiple hidden polymorphism at many internal nodes (see below).
There is a strong likelihood that state 9 reflects a very early loan (cf. Proto-Semitic
*θawru), but the distribution of its cognate set—including words with shifted meaning,
like Russian tur ‘aurochs’—suggests borrowing into Nuclear IE at the latest. States 13
and 14 were already present as words for ‘bull’ at that date (cf. the Tocharian terms); thus
a triple polymorphism, surviving as such in no attested language, must apparently be
posited for 319 ‘bull’ at many internal nodes. Specialization of words meaning ‘bull’ and
‘(head of) cattle’ to mean ‘ox’ and ‘cow’ has also given rise to extensive parallel devel-
opment.

320 buy
Hitt. 1 Av. 4 Luv. 9 Goth. 7
Arm. 1x OCS 5 [loan] Lyc. 10 ON 14 [loan]
Gk. 1y Lith. 6 TA 11 OHG 15 [loan]
Alb. 2 OE 7 OPer. 12 Welsh 3
TB 3 OI 3 OPru. 13 Osc. 8
Ved. 3 Lat. 8 Latv. 6 Umb. 8
1 *wes- 6 PEBalt. *perka
1x, 1y derivs. of derived nominals 7 PGmc. *bugiþi (*bugja-)
3 *k reyh2- (pres. *kwrinéh2ti)
w 8 PItal. *emet
The members of superstate 1 have been coded separately, since the Hittite form is a basic
verb while the Greek and Armenian words are denominatives derived from (different)
nominals which are in turn derived from that verb.
The prehistory of the Old Norse and Old High German words is partly obscure, but it
seems clear that they are ultimately derived from a Germanic borrowing of Latin caupō
‘merchant, peddlar’; the inflection of this verb does not match from language to lan-
guage, so that further borrowing within Germanic is very likely.

September 2012
 101

321 carry
[two characters]
Hitt. 1 Av. 2 Luv. 1 Goth. 2
Arm. 2 OCS 4x Lyc. 5 ON 2
Gk. 2 Lith. 4 TA 2 OHG 7
Alb. 3 OE 2 OPer. 2 Welsh 8
TB 2 OI 2 OPru. 6 Osc. 2
Ved. 2 Lat. 2 Latv. 4 Umb. 2
1 PAnat. *pe-d- 4 PBS *nešeti
2 *bhéreti 4x deriv. of 4
Though the Old Prussian word superficially resembles the Anatolian forms, we can find
no evidence that it should be analyzed the same way (i.e., as the prefix *pe- plus the root
*deh3- ‘take’); thus we do not judge it cognate with them.
We have employed both codings of superstate 4, since direct replacement is likely.

322 cattle
Forms a polymorphic set with 319 bull (q.v.), 326 cow, and 379 ox.

323 collect [polymorphic]

[two characters, both polymorphic]
Hitt. 1 Av. 5 Luv. 10 Goth. 1
Arm. 2 OCS 6 Lyc. 11 ON 1/8
Gk. 3 Lith. 7 TA 4 OHG 1/8
Alb. 3 OE 1/8 OPer. 12 Welsh 9
TB 4 OI 9 OPru. 7 Osc. 13
Ved. 5 Lat. 3 Latv. 1x Umb. 14
1 *lésti 5 PIIr. *čináuti
1x deriv. of 1 7 PBalt. *renka
3 *léǵeti 8 PNWGmc. *samnōþi
4 PToch. *krëwp- 9 cpds. of PCelt. *ela-
We have employed both codings for superstate 1, since 1x appears to be a deverbative
derivative that could have replaced the basic verb directly.
Though the overt polymorphism is confined to Germanic, it appears that other internal
nodes must also be assigned two states in all of the more likely trees.

September 2012
 102

324 comb
Hitt. 1 Av. 7 Luv. 11 Goth. 17
Arm. 2 OCS 8a Lyc. 12 ON 10
Gk. 3 Lith. 9 TA 13 OHG 10
Alb. 4 OE 10 OPer. 14 Welsh 18
TB 5 OI 8b OPru. 15 Osc. 19
Ved. 6 Lat. 3 Latv. 16 [loan] Umb. 3
3 *péḱtn̥ ~ *pḱtén- 10 PGmc. *kambaz (< *ǵómbhos
8a, 8b derivs. of verb *kes- ‘comb’ ‘row of teeth’)
States 8a, 8b clearly represent independent derivatives; we have therefore coded them
separately.

325 cook
Hitt. 1 Av. 3 Luv. 9 Goth. 14
Arm. 2 OCS 5 Lyc. 10 ON 15
Gk. 3 Lith. 6 TA 3 OHG 16 [loan]
Alb. 4 OE 7 OPer. 11 Welsh 17 [loan]
TB 3 OI 8 OPru. 12 Osc. 18
Ved. 3 Lat. 3 Latv. 13 [loan] Umb. 19
3 *pékweti
The variable inflection of the Latvian word strongly suggests that it is a Slavic loan (cf.
Endzelīns 1923:635).

326 cow
Forms a polymorphic set with 319 bull (q.v.), 322 cattle, and 379 ox.

327 darkness
Hitt. 1a Av. 6a Luv. 8 Goth. 12
Arm. 2 [loan] OCS 6a Lyc. 9 ON 13
Gk. 3 Lith. 6a TA 5 OHG 1b
Alb. 4 OE 7 OPer. 10 Welsh 6b
TB 5 OI 6b OPru. 11 Osc. 14
Ved. 6a Lat. 6a Latv. 6a Umb. 15
1a, 1b derivs. of *dhn̥g- ‘dark(-colored)’ 6 derivs. of *temH-
5 PToch. *ọrkəәm- 6a *témHs ~ *tªHés-
6b PCelt. *tem-el-

September 2012
 103

Under 6a we include further extensions of the inherited s-stem.

Since the derivatives of superstate 1 and those of superstate 6 are clearly independent, we
have coded them all separately.

328 daughter
Hitt. 1 Av. 2 Luv. 2 Goth. 2
Arm. 2 OCS 2 Lyc. 2 ON 2
Gk. 2 Lith. 2 TA 2 OHG 2
Alb. 3 OE 2 OPer. 6 Welsh 8
TB 2 OI 4 OPru. 2 Osc. 2
Ved. 2 Lat. 5 Latv. 7 Umb. 9
2 *dhugh2tḗr

329 daughter-in-law
Hitt. 1 Av. 4 Luv. 7 Goth. 12
Arm. 2 OCS 2 Lyc. 8 ON 2
Gk. 2 Lith. 5 TA 9 OHG 2
Alb. 2 OE 2 OPer. 10 Welsh 13
TB 3 OI 6 OPru. 5 Osc. 14
Ved. 2 Lat. 2 Latv. 11 Umb. 15
2 *snusós 5 PBalt. *martī

330 door [polymorphic]

Hitt. 1 Av. 2 Luv. 3 Goth. 2/7
Arm. 2 OCS 2 Lyc. 4 ON 2/7
Gk. 2 Lith. 2 TA 5 OHG 2
Alb. 2 OE 2 OPer. 2 Welsh 2
TB 2 OI 2 OPru. 6 Osc. 8
Ved. 2 Lat. 2 Latv. 2 Umb. 9
2 *dhwór- ~ *dhur- 7 PGmc. *hurdiz
The initial d- of the Vedic cognate reflects the influence of ‘two’; the Iranian merger of
*d and *dh has eliminated potential evidence for that lexical analogy.
The polymorphism is confined to Germanic and is leaf-connected.

September 2012
 104

331 drive AND 361 lead [polymorphic set; see notes for coding]
Hitt. 1b, 1a Av. 3, 1c/8x Luv. 13, 1b Goth. 10, 12
Arm. 2, 3 OCS 9, 8 Lyc. 14, 15 ON 18, 11
Gk. 4, 3 Lith. 9, 8 TA 16, 3 OHG 10, 11
Alb. 5, 6 OE 10, 11 OPer. 17, 1c Welsh 19, 20
TB 7, 3 OI 3, 8 OPru. 9, 8 Osc. 21, 22
Ved. 3, 1c Lat. 3, 12 Latv. 9, 8 Umb. 3, 23
1 *noyh1- ~ *neyh1- ‘lead’ 9 PBS *gen- ~ *gun- ‘drive’
1a hi-conjugation 10 PGmc. *drībidi ‘drive’
1b hi-conj., reduplicated 11 PNWGmc. *laidīþi ‘lead’
1c thematic present 12 *déwketi ‘lead’
3 *h2éǵeti ‘drive’
8 *wédheti ‘lead’; 8x intensive (or causative?)
331 ‘drive’ is monomorphic and can be used as input to the algorithm. For 361 ‘lead’ we
employ both codings for state 1 and its substates, giving two characters; but 8 and 8x are
coded together because the latter is unique.
The meaning ‘drive’ is validated for state 3 by the fact that the root *h2eǵ- is the basis for
derivatives meaning ‘axle’ and *‘pasture’ → ‘field’ even in Greek, the most archaic lan-
guage in which the verb means ‘lead’ (see characters 307 and 337). That 12 was the in-
herited state in Germanic is demonstrated by fossilized compounds in the West Germanic
languages (e.g. OE heretoga, OHG herizogo ‘army-leader’).
The distribution of states 1, 8, and 12 is incompatible with all the better trees, since it
groups Italic and Germanic together, Anatolian and Indo-Iranian together (!), and Celtic
and Balto-Slavic together (!!).

332 duck [with parallel development]

Hitt. 1 Av. 6 Luv. 9 Goth. 14
Arm. 2 OCS 7 Lyc. 10 ON 7
Gk. 1 Lith. 7 TA 11 OHG 7
Alb. 3 OE 7 OPer. 12 Welsh 15
TB 4 OI 8 OPru. 7 Osc. 16
Ved. 5 Lat. 7 Latv. 13 Umb. 17
1 *snéh2n̥tih2 (vel sim.) ‘swimming (fem.)’
7 *h2énh2t-s ~ *h2n̥h2t-
On the cognates assigned to state 1 see Rix 1991 and Katz 2004. The meaning of the
protoform strongly argues parallel development.

September 2012
 105

333 eight
Hitt. 1 Av. 2 Luv. 3 Goth. 2
Arm. 2 OCS 2 Lyc. 2 ON 2
Gk. 2 Lith. 2 TA 2 OHG 2
Alb. 2 OE 2 OPer. 4 Welsh 2
TB 2 OI 2 OPru. 2 Osc. 5
Ved. 2 Lat. 2 Latv. 2 Umb. 6
2 *oḱtṓw

334 elbow
[two characters]
Hitt. 1 Av. 6 Luv. 10 Goth. 14
Arm. 2 OCS 7 Lyc. 11 ON 8x
Gk. 3 Lith. 7 TA 12 OHG 8x
Alb. 4 OE 8x OPer. 13 Welsh 8y
TB 5 OI 8 OPru. 7 Osc. 15
Ved. 6 Lat. 9 Latv. 7 Umb. 16
6 PIIr. *ar(a)tn- 8 *olī̆n-
7 PBS *alku- 8x cpd. with PNWGmc. *bogan- ‘bow’
8y cpd. with Welsh penn ‘head’
All the shared states might be derivatives of an inherited form, but since it cannot be
reconstructed, we have kept them separate. We employ both codings of superstate 8, as
direct replacement of a simplex by compounds is likely.
On the Indo-Iranian forms see Szemerényi 1966:196-9; it is not clear whether Old Per-
sian aršniš ‘cubit’ still meant ‘elbow’ as well.

335 eyebrow
Hitt. 1 Av. 3 Luv. 7 Goth. 11
Arm. 2 OCS 3 Lyc. 8 ON 3
Gk. 3 Lith. 5 TA 3 OHG 3
Alb. 4 OE 3 OPer. 9 Welsh 12
TB 3 OI 3 OPru. 3 Osc. 13
Ved. 3 Lat. 6 Latv. 10 Umb. 14
3 *h3bhruH- and derivs.
The Old Prussian form appears to have been mangled by a German copyist.

September 2012
 106

336 father-in-law
Hitt. 1 Av. 4 Luv. 8 Goth. 4
Arm. 2 OCS 4 Lyc. 9 ON 14
Gk. 3 Lith. 6 TA 10 OHG 4
Alb. 4 OE 4 OPer. 11 Welsh 15
TB 5 OI 7 OPru. 12 Osc. 16
Ved. 4 Lat. 4 Latv. 13 Umb. 17
4 *swéḱuros
The velar stop of the Old Church Slavonic form is surprising (cf. older Lithuanian
šẽšuras); very early borrowing from pre-Proto-Germanic is at least possible.
The Armenian word is a compound meaning ‘mother-in-law’s husband’.

337 field [polymorphic]

[two monomorphic characters by conflated split coding]
Hitt. 1 Av. 6 Luv. 10 Goth. 2
Arm. 2 OCS 7 Lyc. 11 ON 2
Gk. 2 Lith. 8 TA 12 OHG 2
Alb. 3a OE 2 OPer. 13 Welsh 14
TB 4 OI 9 OPru. 8 Osc. 15
Ved. 5 Lat. 3b Latv. 8 Umb. 2/3b
2 *h2éǵros ‘pasture’ 3a–b derivs. of *h2er3- ‘plow’
4 PToch. *miṣ- (?) 3b *h2er3-wo-
8 *lowkos ‘clearing’
On the Armenian form see Olsen 1999:30 with references. Tocharian A miṣi means
‘community’ and should be deleted from the wordlist.
We have coded the substates of 3 separately, as they are independent derivatives.
The polymorphism is confined to Italic and is leaf-connected. We have reduced this
character to two monomorphic characters by conflated split coding, with 3b ∪ 2/3b coded
against 2 in one character and all three together in the other.

September 2012
 107

338 finger
Hitt. 1 Av. 7 Luv. 12 Goth. 9
Arm. 2 OCS 8 Lyc. 13 ON 9
Gk. 3 Lith. 8 TA 5 OHG 9
Alb. 4 OE 9 OPer. 14 Welsh 15
TB 5 OI 10 OPru. 8 Osc. 16
Ved. 6 Lat. 11 Latv. 8 Umb. 17
5 PToch. *prar- 8 PBS *pirštas 9 PGmc. *fingraz

339 fly [noun]

Hitt. 1 Av. 4 Luv. 7 Goth. 11
Arm. 2 OCS 2 Lyc. 8 ON 5
Gk. 2 Lith. 2 TA 9 OHG 5
Alb. 2 OE 5 OPer. 10 Welsh 6
TB 3 OI 6 OPru. 2 Osc. 12
Ved. 4 Lat. 2 Latv. 2 Umb. 13
2 *mus- and related forms 5 derivs. of PGmc. *fleuganą ‘to fly’
4 PIIr. *makši- 6 PCelt. *kuli-
The Armenian and Albanian forms are somewhat divergent in shape, but that is not
necessarily unexpected in the name of an economically unimportant animal. See e.g.
Demiraj 1997:240 with references.

340 follow
Hitt. 1 Av. 3 Luv. 8 Goth. 12
Arm. 2 OCS 6 Lyc. 9 ON 7
Gk. 3 Lith. 3 TA 3 OHG 7
Alb. 4 OE 7 OPer. 10 Welsh 13
TB 5 OI 3 OPru. 11 Osc. 14
Ved. 3 Lat. 3 Latv. 3 Umb. 15
3 *sékwetor 7 PNWGmc. *fulgēþi
On the failure of schwa-rounding to occur in the TA form cf. Ringe 1996:150.

September 2012
 108

341 free
Hitt. 1 Av. 7 Luv. 12 Goth. 10
Arm. 2 [loan] OCS 8 Lyc. 1 ON 10
Gk. 3 Lith. 9 TA 13 OHG 10
Alb. 4 OE 10 OPer. 14 Welsh 10
TB 5 OI 11 OPru. 15 Osc. 3
Ved. 6 Lat. 3 Latv. 16 Umb. 17
1 PAnat. *arawos 10 *priHós ‘beloved’ and cpd.
h
3 *h1léwd eros
The startling shift of ‘beloved’ to ‘free’ strongly suggests loan-translation reflecting very
early contact between Celtic and Germanic (cf. Feist 1939 s.v. freis with references). The
distribution of the Mediterranean word (state 3) might also be a contact phenomenon of
some sort.

342 furrow
Hitt. 1 Av. 7 Luv. 13 Goth. 18
Arm. 2 OCS 8 Lyc. 14 ON 19
Gk. 3 Lith. 9 TA 15 OHG 10
Alb. 4 [loan] OE 10 OPer. 16 Welsh 10
TB 5 OI 11 OPru. 17 Osc. 20
Ved. 6 Lat. 12 Latv. 9 Umb. 21
9 PEBalt. *vagā 10 *pr̥k-

343 go
Forms a polymorphic set with 25 come (q.v.) and 185 walk.

344 goat
Hitt. 1 Av. 6 Luv. 11 Goth. 8
Arm. 2 OCS 7 Lyc. 12 ON 8
Gk. 3 Lith. 4 TA 5 OHG 8
Alb. 4 OE 8 OPer. 13 Welsh 9
TB 5 OI 9 OPru. 4 Osc. 15
Ved. 4 Lat. 10 Latv. 14 [loan] Umb. 10
4 *aǵ- 9 PCelt. *gabros
5 PToch. *asəә 10 PItal. *kapros
8 PGmc. *gait-

September 2012
 109

We have accepted the usual etymology of the Albanian word, with some hesitation; see
Demiraj 1997:160 for discussion, and note that the alternative would assign state 2
(*ayǵ-) to Albanian.

345 gold
Hitt. 1 Av. 6a Luv. 9 Goth. 6c
Arm. 2 OCS 6b Lyc. 10 ON 6c
Gk. 3 Lith. 7 TA 5 OHG 6c
Alb. 4 [loan] OE 6c OPer. 6a Welsh 11[loan]
TB 5 OI 8 [loan] OPru. 7 Osc. 12
Ved. 6a Lat. 7 Latv. 6d Umb. 13
6 derivs. of *ǵhel- ‘yellow’ 5 PToch. *wyəәsa
6a PIIr. *źh̥l anyam 7 *awsom
6b PSlav. *zolto
6c PGmc. *gulþą
6d pre-Latv. *zeltas
We have coded states 6a–d separately, since they are independent derivatives.

346 goose
Hitt. 1 Av. 6 Luv. 8 Goth. 12
Arm. 2 OCS 3 Lyc. 9 ON 3
Gk. 3 Lith. 3 TA 10 OHG 3
Alb. 4 OE 3 OPer. 11 Welsh 7
TB 5 OI 7 OPru. 3 Osc. 13
Ved. 3 Lat. 3 Latv. 3 Umb. 14
3 *ǵhans- 7 PCelt. *gigdos (Pedersen 1909:102-3)
The initial velar of the Old Church Slavonic form is surprising, in view of the initial pala-
tal in all the Baltic forms; early borrowing from Germanic is at least possible.

September 2012
 110

347 grain [polymorphic, with parallel development]

[two monomorphic characters by conflated split coding]
Hitt. 1 Av. 6 Luv. 13 Goth. 9
Arm. 2 OCS 7 Lyc. 14 ON 9
Gk. 3 Lith. 6/8 TA 5 OHG 9
Alb. 4 OE 9 OPer. 15 Welsh 11
TB 5 OI 10a/11 OPru. 9/10b Osc. 16
Ved. 6 Lat. 12 Latv. 8 Umb. 17
y
5 PToch. *w əәsarë 9 *ǵr̥h2nóm
6 *yéwos 10a, 10b derivs. of *h2erh3- ‘plow’
8 PEBalt. *graud- ~ *grūd- 11 PCelt. *itu
We have coded states 10a, 10b separately, since they are independent derivatives; the Old
Irish and Old Prussian polymorphisms thus become ineffective.
The East Baltic polymorphism is local and leaf-connected. We have reduced the charac-
ter to two monomorphic characters by conflated split coding, with 8 ∪ 6/8 coded against
6 in one character and all three coded together in the other.
However, note that the distributions of states 6 and 9, which split the Baltic subgroup,
enforce recognition of parallel development, or else an additional polymorphism at
internal nodes.

348 grind
Hitt. 1 Av. 6 Luv. 8 Goth. 1
Arm. 2 OCS 1 Lyc. 9 ON 1
Gk. 2 Lith. 1 TA 10 OHG 1
Alb. 3 OE 7 OPer. 11 Welsh 1
TB 4 OI 1 OPru. 12 Osc. 13
Ved. 5 Lat. 1 Latv. 1 Umb. 1
1 *molh2- ~ *melh2- 2 *h2elh1- (*alh1- ?)

349 half
Hitt. 1 Av. 7 Luv. 13 Goth. 10
Arm. 2 OCS 8 Lyc. 14 ON 10
Gk. 3 Lith. 9 TA 15 OHG 10
Alb. 4 OE 10 OPer. 16 Welsh 18
TB 5 OI 11 OPru. 17 Osc. 19
Ved. 6 Lat. 12 Latv. 9 Umb. 20

September 2012
 111

9 PEBalt. *pus- 10 PGmc. *halbaz

350 have
Forms a polymorphic set with 77 hold (q.v.).

351 hide [polymorphic]

Hitt. 1 Av. 6 Luv. 11 Goth. 14
Arm. 2 OCS 7 Lyc. 12 ON 9/14
Gk. 3 Lith. 8 TA 5 OHG 9
Alb. 4 OE 9/10a OPer. 6 Welsh 9/10b
TB 5 OI 9 OPru. 13 Osc. 15
Ved. 6 Lat. 9 Latv. 8 Umb. 16
5 PToch. *tpukəәsk- (*tukəәsk- ?) 9 *kéleti and derivs.
6 PIIr. *gūźh- 10a, 10b derivs. of *kewdh-
8 PEBalt. *slepja 14 PGmc. *filhidi (*felha-)
We have coded states 10a, 10b separately because of the unexplained anomaly in the
vocalism of the OE root (hȳd- < *hūd-ij- < *kūdh-, not *kewdh-), which strongly suggests
that its connection with apparent cognates elsewhere (including Homeric Gk. κεύθειν
‘cover, conceal’) is indirect. The polymorphism of OE and Welsh thereby becomes
ineffective.
The remaining polymorphism is confined to Germanic and is leaf-connected.

352 honey AND 405 sweet [polymorphic set]

Hitt. 1, 1x Av. 7, 8 Luv. 1, 12 Goth. 1, 19
Arm. 1, 2 OCS 5, 9 Lyc. 13, 14 ON 10, 6b
Gk. 1, 3 Lith. 5, 9 TA 15, 6a OHG 10, 6b
Alb. 1, 4 OE 10, 6b OPer. 16, 17 Welsh 1, 1y
TB 5, 6a OI 1, 1y OPru. 5, 18 Osc. 20, 21
Ved. 5, 5x Lat. 1, 11 Latv. 5, 9 Umb. 22, 23
1 *mélit ‘honey’ 6 *swād- ~ *swad- ~ *sud- ‘pleasant’
1x *mlitús ‘sweet’ 6a *swadrós
1y PCelt. *melissis ‘sweet’ 6b PNWGmc. *swōtijaz
h
5 *méd u ‘sweet’ (neut.) 9 PBS *saldus ‘sweet’
5x recent deriv. adj. 10 PNWGmc. *hunagą ‘honey’
The meaning of state 5 is validated by the survival of an adj. mádhus ‘sweet’ in Vedic,
though the specialization of the neut. to refer to honey (and alcoholic drinks: cf. OE

September 2012
 112

meodu ‘mead’, Gk. µέθυ ‘wine’, etc.) was early and/or widespread. The meaning of state
6 is established by the earlier-attested languages, in both the adj. and the related verbs.
Despite the general similarity of Gk. γλυκύς and Lat. dulcis, any attempt to posit regular
sound correspondences encounters so many difficulties that we have preferred to reject
the connection. The Luvian word appears to be a participle (see Melchert 1993 s.v.);
whether it has any connection with state 5 is unclear.
We code 1x, 1y separately, but use both codings for superstate 6; this yields two alterna-
tive codings for 405 ‘sweet’. (Substates 5x and 5 do not occur in the same character.)
Both codings of 405 ‘sweet’ are monomorphic and can be used as input to the algorithm.

353 horse
Hitt. 1 Av. 3 Luv. 3 Goth. 3
Arm. 2 OCS 5 Lyc. 3 ON 7
Gk. 3 Lith. 6 TA 3 OHG 7
Alb. 4 [loan] OE 7 OPer. 3 Welsh 10
TB 3 OI 3 OPru. 8 Osc. 11
Ved. 3 Lat. 3 Latv. 9 Umb. 12
3 *éḱwos 7 PNWGmc. *hrossą
The Greek word has been severely deformed, but still seems to belong with the cognate
set coded as 3. Note that its initial h-, at least, is a very recent innovation, to judge from
names like Ἄλκιππος (not “Ἄλχιππος”).
Welsh ceffyl is a Latin loan (though march is a native word).

354 house [polymorphic (ineffectively), with parallel development]

[two characters]
Hitt. 1 Av. 2x Luv. 1 Goth. 6b
Arm. 2 OCS 2 Lyc. 1 ON 8
Gk. 3 Lith. 7 TA 5 OHG 8
Alb. 4 [loan] OE 8 OPer. 2x Welsh 9
TB 5 OI 9 OPru. 7 Osc. 10
Ved. 2/6a Lat. 2 Latv. 2 Umb. 11

September 2012
 113

1 PAnat. *pḗr ~ *pr̥n- 7 PBalt. *butan

2 *dom- 8 PNWGmc. *hūsą
2x PIr. *dmānam 9 PCelt. *tegos (← ‘roof’)
5 PToch. *wostəә (< *wā́stu ‘settlement’)
6 set orig. meaning ‘enclosure’
6a *ghr̥dhós
6b *ghórdhos
We have coded states 6a, 6b separately, since they are clearly different formations from
the same root; the Vedic polymorphism thereby becomes ineffective. However, we em-
ploy both codings for superstate 2.
The distribution of states 2 and 7, splitting the East Baltic subgroup, strongly argues
parallel development.

355 hundred
Hitt. 1 Av. 3 Luv. 5 Goth. 3
Arm. 2 OCS 3 Lyc. 6 ON 3
Gk. 3 Lith. 3 TA 3 OHG 9
Alb. 4 [loan] OE 3 OPer. 7 Welsh 3
TB 3 OI 3 OPru. 8 Osc. 10
Ved. 3 Lat. 3 Latv. 3 Umb. 11
3 *ḱm̥ tóm and derivs.
The first vowel of the OCS word is unexpected; a loan from some Iranian language is at
least possible.

356 jaw
Hitt. 1 Av. 2 or 2x Luv. 9 Goth. 14
Arm. 2y OCS 5 Lyc. 10 ON 6b
Gk. 2z Lith. 2w TA 2v OHG 2u
Alb. 3 OE 6a OPer. 11 Welsh 2
TB 4 OI 7 OPru. 12 Osc. 15
Ved. 2x Lat. 8 Latv. 13 Umb. 16
2 *ǵḗnus ~ *ǵénu- and derivs.
2x unexpected PIIr. or pre-Skt. *źh- (as if < *ǵh-)
6a, 6b derivs. of PGmc. *kef- ~ *kaf- (‘chew’?)
We have coded states 6a, 6b separately, since they are independent derivatives.
For superstate 2 we could have adopted both codings, since it is likely that the Indo-Ira-

September 2012
 114

nian form with an innovative initial consonant (like all the derivatives) replaced the
inherited word directly. However, because PIIr. *ź and *źh merged in Iranian, we cannot
tell whether Avestan should be assigned state 2 or state 2x. Thus all the suffixed sub-
states of 2 could be unique, and in that case offer no information about the tree. We
accordingly code superstate 2 as a single state.

357 join
Hitt. 1 Av. 6 Luv. 10 Goth. 15
Arm. 2 [loan] OCS 7 Lyc. 11 ON 16
Gk. 3 Lith. 6 TA 5 OHG 8
Alb. 4 OE 8 OPer. 12 Welsh 17
TB 5 OI 9 OPru. 13 Osc. 18
Ved. 6 Lat. 6 Latv. 14 Umb. 19
5 PToch. *tsuwəәsk- 8 PWGmc. *fōgiþi
6 *yewg- (pres. *yunégti)
The Armenian verb is derived from a noun zoygkh ‘pair’ borrowed from Syriac, which in
turn borrowed the word from Greek; see Olsen 1999:931 and (on the derivational pattern)
15-7.

358 king
Hitt. 1 Av. 7 Luv. 11 Goth. 14
Arm. 2 OCS 8 [loans] Lyc. 11 ON 10
Gk. 3 Lith. 9 [loan] TA 5 OHG 10
Alb. 4 OE 10 OPer. 7 Welsh 15
TB 5 OI 6 OPru. 12 [loan] Osc. 16
Ved. 6 Lat. 6 Latv. 13 [loan] Umb. 17
5 PToch. *wəәlo, *lant- 10 PNWGmc. *kuningaz
6 *h3rḗǵs 11 PLuv. *Hantowot-
7 PIr. *xšayah and deriv.

September 2012
 115

359 knot
Hitt. 1 Av. 7 Luv. 13 Goth. 18
Arm. 2 OCS 8 Lyc. 14 ON 10
Gk. 3 Lith. 9 TA 15 OHG 10
Alb. 4 OE 10 OPer. 16 Welsh 19
TB 5 OI 11 OPru. 17 Osc. 20
Ved. 6 Lat. 12 Latv. 9 Umb. 21
9 derivs. of *mezg- ‘tie’ 10 PNWGmc. *knu…
The Germanic words appear to reflect sound symbolism; we have coded them together
because they appear to share enough phonological material to reflect common ancestry,
but they could be independent developments.

360 lamb [with parallel development]

Hitt. 1 Av. 5 Luv. 9 Goth. 8
Arm. 2 OCS 6 Lyc. 10 ON 8
Gk. 2 Lith. 7 TA 11 OHG 8
Alb. 3 OE 8 OPer. 12 Welsh 6
TB 4 OI 6 OPru. 7 Osc. 13
Ved. 2 Lat. 6 Latv. 7 Umb. 14
2 *wr̥h1én- 7 PBalt. *erj-
w
6 *ag nós 8 PGmc. *lambaz, *lambiz-
The preform of state 6 is not secure; the Celtic forms seem to demand *gwh, while the
others demand *gw. Note that the same word appears in Greek (though not in Classical
Attic, the basis of our coding); thus it seems clear that polymorphism or parallel devel-
opment is widespread in the tree.

361 lead
Forms a polymorphic set with 331 drive (q.v.).

September 2012
 116

362 leave
[two characters]
Hitt. 1 Av. 2 Luv. 7 Goth. 3b
Arm. 2 OCS 5 Lyc. 8 ON 3b
Gk. 2 Lith. 2 TA 4 OHG 3b
Alb. 3a OE 3b OPer. 9 Welsh 12
TB 4 OI 6 OPru. 10 Osc. 13
Ved. 2 Lat. 2 Latv. 11 Umb. 14
w w
2 *leyk - (pres. *linék ti) 4 PToch. *arṣəә(ṣəә)
3 *leh1d-
3a present with nasal suffix
3b PGmc. *lētidi
We have employed both codings for superstate 3, since it is a basic verb.
Though the Old Irish word resembles set 2, it is difficult to account for its root vocalism
and stem vowel on the hypothesis that it is a cognate of that set. The OCS and Latvian
compounds also resemble one another, but the stem formation is different; it seems likely
that the Latvian word is a “loan translation” based on Russian.

363 lick
Hitt. 1 Av. 2 Luv. 5 Goth. 2
Arm. 2 OCS 2 Lyc. 6 ON 2
Gk. 2 Lith. 2 TA 7 OHG 2
Alb. 3 OE 2 OPer. 8 Welsh 2
TB 4 OI 2 OPru. 9 Osc. 10
Ved. 2 Lat. 2 Latv. 2 Umb. 11
2 *léyǵhti and derivs.
The inherited verb has been deformed by sound symbolism in many daughters; for exam-
ple, that is probably the source of the Old Norse initial s- (see e.g. Feist 1939:91, de Vries
1962 s.v. sleikja).

September 2012
 117

364 lift [polymorphic]

[two monomorphic characters by conflated split coding]
Hitt. 1 Av. 8 Luv. 13 Goth. 11
Arm. 2 OCS 9 Lyc. 14 ON 11
Gk. 3 Lith. 10 TA 5 OHG 11
Alb. 4 OE 11 OPer. 15 Welsh 17
TB 5/6x OI 12 OPru. 16 Osc. 18
Ved. 7 Lat. 6 Latv. 10 Umb. 19
5 PToch. *musnatəәr 10 PEBalt. *kelja
6 *tl̥néh2ti 11 PGmc. *habidi (*habja-)
6x PToch. *təә́ləәṣṣəә(ṣəә)
The polymorphism is confined to Tocharian and is leaf-connected. We have resolved this
polymorphism by conflated split coding.

365 light [polymorphic, but not effectively so]

Hitt. 1 Av. 1 Luv. 5 Goth. 1
Arm. 1 OCS 4 Lyc. 6 ON 1
Gk. 2a Lith. 4 TA 1 OHG 1
Alb. 3 OE 1 OPer. 7 Welsh 1
TB 1 OI 1 OPru. 8 Osc. 10
Ved. 1/2b Lat. 1 Latv. 9 Umb. 11
1 derivs. of *lewk- 4 derivs. of PBS *šveit-
h
2a, 2b derivs. of *b eh2- ‘shine’
We have coded states 2a, 2b separately, because the shape of the root is actually some-
what different (cf. Homeric Gk. φάος < *φάϝος, root φαϝ- also in pres. πιφαύσκειν
‘make clear, declare’); the Vedic polymorphism thereby becomes ineffective.

366 lip [polymorphic]

Hitt. 1 Av. 6 Luv. 11 Goth. 9
Arm. 2 OCS 6 Lyc. 12 ON 9
Gk. 3 Lith. 7 TA 5 OHG 8
Alb. 4 OE 8/9 OPer. 13 Welsh 14
TB 5 OI 10 OPru. 9 Osc. 15
Ved. 6 Lat. 8 Latv. 7 Umb. 16

September 2012
 118

5 PToch. *lyəәmẹnë (dual) 8 *leb-

6 *austh2o- (*ou-?) 9 *wer-
7 PEBalt. *lūpā
The forms belonging to state 9 exhibit a variety of suffixes which do not match.
Though the overt polymorphism is confined to Old English, the distribution of the states
shared by OE is striking, including not only Germanic but also Latin and Old Prussian.
Parallel development is possible (cf. Hitt. lilipai ‘(s)he licks’, which suggests that state 8
might represent a verb root), but the situation remains unclear.

367 livestock
Hitt. 1 Av. 6 Luv. 10 Goth. 16
Arm. 2 [loan] OCS 7 Lyc. 11 ON 6
Gk. 3 Lith. 8 TA 12 OHG 6
Alb. 4 OE 6 OPer. 13 Welsh 17
TB 5 OI 9 OPru. 14 Osc. 18
Ved. 6 Lat. 6 Latv. 15 Umb. 6
6 *péḱu(s)

368 make [polymorphic]

Hitt. 1 Av. 6 Luv. 1 Goth. 2b
Arm. 2a OCS 7 Lyc. 1 ON 13
Gk. 3 Lith. 8 TA 1 OHG 2b/9
Alb. 4 OE 2b/9 OPer. 6 Welsh 10
TB 5 OI 10 OPru. 12 Osc. 11
Ved. 6 Lat. 11 Latv. 8 Umb. 11
1 *h1yeh1- 8 PEBalt. *darā
2 derivs. of *werǵ- 9 PWGmc. *makōþi
2a *worǵeyeti 10 PCelt. *gnī-
2b *wr̥ǵyéti 11 PItal. *fakit (*fakyo-)
6 PIIr. *kr̥náuti
On the lexical analogies that have deformed the Welsh word see Pedersen 1913:544-6,
Schumacher 2004:345-6.
We have coded states 2a, 2b separately, since they are very different derivatives (state 2a
could even be denominative).
The polymorphism is confined to Germanic and is leaf-connected.

September 2012
 119

369 middle
Hitt. 1 Av. 2 Luv. 6 Goth. 2
Arm. 2 OCS 4 Lyc. 7 ON 2
Gk. 2 Lith. 5 TA 3 OHG 2
Alb. 2 OE 2 OPer. 8 Welsh 10 [loan]
TB 3 OI 2 OPru. 9 Osc. 2
Ved. 2 Lat. 2 Latv. 5 Umb. 11
2 *médhyos and derivs. 5 PEBalt. *vid-
3 PToch. *yəәwarcəәka-

370 milk [with parallel development?]

[two characters]
Hitt. 1 Av. 6a Luv. 8 Goth. 5x
Arm. 2 OCS 7 [loan] Lyc. 9 ON 5x
Gk. 3a Lith. 6b TA 5 OHG 5x
Alb. 4 OE 5x OPer. 10 Welsh 12 [loan]
TB 5 OI 5 OPru. 11 Osc. 13
Ved. 6a Lat. 3b Latv. 6b Umb. 14
3a Gk. γάλακτ- 6 derivs. of *peyH- ‘be fat’
3b Lat. lact- 6a PIIr. *páyas
5 *h2melǵ- ‘to milk’ and derivs. 6b PEBalt. *pienas
5x PGmc. *meluk-
The meanings reconstructable for states 5 and 6 suggest parallel development, but it
should be borne in mind that no noun ‘milk’ is reconstructable in any case.
We have coded states 3a, 3b separately because the shapes of the Latin and Greek words
simply do not match; this item could easily be a loan from some non-IE language in both
Latin and Greek (Porzig 1954:132). We have also coded states 6a, 6b separately because
they are clearly independent derivatives.
Superstate 5 is a less clear case. With some hesitation we have adopted both codings,
since a direct connection between the Germanic words and the others is not unlikely,
even though the Germanic *-u- is unexpected.
OCS mlěko MUST be a loan from Germanic: not only does it exhibit a velar stop in place
of an expected palatal, but the stop has been devoiced by Grimm’s Law!

September 2012
 120

371 mother-in-law
Hitt. 1 Av. 5 Luv. 8 Goth. 2
Arm. 2 OCS 2 Lyc. 9 ON 14
Gk. 3 Lith. 6 TA 10 OHG 2
Alb. 2 OE 2 OPer. 11 Welsh 15
TB 4 OI 7 OPru. 12 Osc. 16
Ved. 2 Lat. 2 Latv. 13 Umb. 17
2 *sweḱrúh2

372 mouse
Hitt. 1 Av. 4 Luv. 7 Goth. 12
Arm. 2 OCS 2 Lyc. 8 ON 2
Gk. 2 Lith. 5 TA 9 OHG 2
Alb. 2 OE 2 OPer. 10 Welsh 6
TB 3 OI 6 OPru. 11 Osc. 13
Ved. 2 Lat. 2 Latv. 5 Umb. 14
2 *mū́s 6 PCelt. *lukūs, *lukot- (Joe Eska, p.c.; cf. Pedersen
5 PEBalt. *pelē 1909:376, Thurneysen 1946:206)
The Tocharian B word does not belong with set 2 (Adams 1999 s.v. maścītse).

373 (finger)nail
Hitt. 1 Av. 5 Luv. 6 Goth. 9
Arm. 2 OCS 3w Lyc. 7 ON 3y
Gk. 3 Lith. 3x TA 3v OHG 3y
Alb. 4 OE 3y OPer. 8 Welsh 3z
TB 3v OI 3z OPru. 3w Osc. 10
Ved. 3 Lat. 3 Latv. 3x Umb. 11
3 *h3nogh(w)- and derivs. 3x PEBalt. *nagas
3v PToch. *mëkuwa (pl.) 3y PGmc. *naglaz
3w PBS *nagutis 3z PCelt. *angwīnā
We have coded the states of superstate 3 separately, simply because if they are coded
together it is the only shared state. On the Armenian form see Olsen 1999:138; we reject
the connection.

September 2012
 121

374 naked
Hitt. 1 Av. 1 Luv. 4 Goth. 1y
Arm. 1 OCS 1x Lyc. 5 ON 1y
Gk. 1 Lith. 1x TA 6 OHG 1y
Alb. 2 OE 1y OPer. 7 Welsh 1z
TB 3 OI 1z OPru. 8 Osc. 10
Ved. 1 Lat. 1 Latv. 9 Umb. 11
1 *negwnós and derivs. 1y PGmc. *nakwadaz
1x PBS *nōgas 1z PCelt. *noktos
Many cognates of this set have been irregularly reformed (for tabu reasons?).
We have coded the states of superstate 1 separately, simply because if they are coded
together it is the only shared state.

375 navel
Hitt. 1 Av. 3 Luv. 8 Goth. 12
Arm. 2 OCS 6 Lyc. 9 ON 3x
Gk. 3 Lith. 7 TA 10 OHG 3x
Alb. 4 OE 3x OPer. 11 Welsh 13
TB 5 OI 3 OPru. 3 Osc. 14
Ved. 3 Lat. 3 Latv. 3 Umb. 15
h
3 *h3nob - and derivs. (most of which are unique)
3x PNWGmc. *nabVlan-
We have coded the states of superstate 3 separately, simply because if they are coded
together it is the only shared state.

376 nine
Hitt. 1 Av. 2 Luv. 3 Goth. 2
Arm. 2 OCS 2x Lyc. 2 ON 2
Gk. 2 Lith. 2x TA 2 OHG 2
Alb. 2 OE 2 OPer. 2 Welsh 2
TB 2 OI 2 OPru. 2 Osc. 4
Ved. 2 Lat. 2 Latv. 2x Umb. 2
2 *(h1)néwn̥ 2x initial *d- by lexical analogy with ‘ten’
We have coded the states of superstate 2 separately, simply because if they are coded
together it is the only shared state.

September 2012
 122

377 now
[two characters]
Hitt. 1 Av. 1xx Luv. 1 Goth. 1
Arm. 2 OCS 1 Lyc. 5 ON 1
Gk. 1 Lith. 3 TA 6 OHG 1
Alb. 1 OE 1 OPer. 1xx Welsh 9
TB 1 OI 4 OPru. 7 Osc. 10
Ved. 1x Lat. 1 Latv. 8 Umb. 11
1 *nū́ and derivs. (most unique or probably parallel)
1x PIIr. *nūnám 1xx PIr. *nūram (with dissimilation)
We have adopted two codings, one in which states 1x and 1xx are coded together (against
the remainder of superstate 1), the other in which all states are coded separately, since
that captures the development of this character in Indo-Iranian.

378 orphan
Hitt. 1 Av. 6 Luv. 11 Goth. 17
Arm. 2 OCS 7 Lyc. 12 ON 18
Gk. 2 Lith. 8 TA 13 OHG 19
Alb. 3 OE 9 OPer. 14 Welsh 20
TB 4 OI 10 OPru. 15 Osc. 21
Ved. 5 Lat. 2 Latv. 16 Umb. 22
2 *orbhos

379 ox
Forms a polymorphic set with 319 bull (q.v.), 322 cattle, and 326 cow.

380 pig [polymorphic]

[two monomorphic characters, in part by split coding]
Hitt. 1 Av. 3 Luv. 8 Goth. 3x
Arm. 2 OCS 3x Lyc. 9 ON 3x
Gk. 3 Lith. 5 TA 10 OHG 3x
Alb. 4 OE 3x OPer. 11 Welsh 6
TB 3 OI 6 OPru. 3x Osc. 13
Ved. 3 Lat. 3/7 Latv. 12 Umb. 3/7
3 *sū́s 6 PCelt. *mokkus
3x *su-īno- (orig. adj.?) 7 *pórḱos

September 2012
 123

The polymorphism is confined to Italic (though cognates of state 7 do occur elsewhere in

other meanings) and is leaf-connected.
We have employed both codings for superstate 3; for the polymorphism 3/7 and super-
state 3 we employ split coding.
Both Albanian words are of uncertain etymology; cf. Demiraj 1997:131-2, 397-8.

381 plow
Hitt. 1 Av. 6 Luv. 7 Goth. 3
Arm. 2 OCS 3 Lyc. 8 ON 3
Gk. 3 Lith. 3 TA 9 OHG 3
Alb. 4 OE 3 OPer. 10 Welsh 3
TB 5 OI 3 OPru. 3 Osc. 11
Ved. 6 Lat. 3 Latv. 3 Umb. 12
3 *h2erh3-, pres. *h2éryeti 6 *kwélsti ‘make a furrow’
The assignment of state 3 to Old Prussian is uncertain, since only the noun for ‘plow’ is
attested in that language. However, that does not affect the shape of the tree, since the
alternative is to assign it a unique state.

382 pour
Hitt. 1 Av. 5 Luv. 1 Goth. 3x
Arm. 2 OCS 6 Lyc. 8 ON 10
Gk. 3 Lith. 6 TA 3 OHG 3x
Alb. 4 OE 3x OPer. 9 Welsh 11
TB 3 OI 7 OPru. 6 Osc. 12
Ved. 5 Lat. 3x Latv. 6 Umb. 13
1 PAnat. *laHu- (Melchert 1994:72-3) 5 PIIr. *sinčáti
h
3 *ǵ ew- 6 *leyH-
3x “extended” root *ǵhewd-
We have cautiously coded states 3 and 3x separately; though there is a clear etymological
connection between them, the fact that the shape of the actual root has been altered makes
it unclear whether the developmental link between them is direct, since the “extended”
root might at first have had a different meaning.

September 2012
 124

383 put [polymorphic, with parallel development]

[two monomorphic characters by split coding]
Hitt. 1d Av. 1a Luv. 1b Goth. 5/6a
Arm. 1e OCS 4a/5 Lyc. 1b ON 5/6a
Gk. 1a Lith. 1a/4b TA 1c OHG 5/6a
Alb. 2 OE 5/6a OPer. 1a Welsh 4c
TB 1c OI 7 OPru. 6b Osc. 1f
Ved. 1a Lat. 8 Latv. 9 Umb. 10
h
1 *d eh1- 4a–c derivs. of *steh2- ‘stand’
1a pres. *dhédheh1ti 5 *loghéyeti ‘cause to lie down’
1b PLuv. *tuwV- 6a *sodéyeti ‘cause to sit down
1c PToch. *tas- 6b other deriv. of *sed- ‘sit down’
1d–f other present stems
We have coded states 4a–c, 6a, 6b separately, since they are clearly independent develop-
ments. (The OCS and Lith. polymorphisms thereby become ineffective.) For superstate
1 we employ both codings, since the differences between the substates mainly involve
stem-formation.
For the polymorphism 5/6a we employ split coding.
The meanings reconstructable for (super)states 4–6 argue massive parallel development.
However, none of the states is (precisely) shared with any of the other verbs involved,
since it is typically causatives of the latter that develop into ‘put’.
The meaning of the Umbrian word listed is probably ‘put’ (cf. Vetter 1953:248), but
other interpretations are possible (cf. e.g. Ernout 1961:125, Ancillotti and Cerri 1996:
391); in any case Umbrian does not exhibit a cognate of any term in any of the other
languages.

384 remember
Hitt. 1 Av. 6 Luv. 7 Goth. 3
Arm. 2 OCS 3 Lyc. 8 ON 3
Gk. 3 Lith. 3 TA 5 OHG 11
Alb. 4 [loan] OE 3 OPer. 9 Welsh 3
TB 5 OI 3 OPru. 10 Osc. 3
Ved. 6 Lat. 3 Latv. 3 Umb. 12
3 *men- (pf. *memóne) and derivs. 6 PIIr. *smárati
5 PToch. *epiyacəә kəәllaṣṣəә(ṣəә) ‘call to mind’

September 2012
 125

385 roof
Hitt. 1 Av. 7 Luv. 9 Goth. 14
Arm. 2 OCS 8 Lyc. 10 ON 3c
Gk. 3a Lith. 3b TA 11 OHG 3c
Alb. 4 OE 3c OPer. 12 Welsh 3d
TB 5 OI 3d OPru. 3b Osc. 15
Ved. 6 Lat. 3e Latv. 13 Umb. 16
3a–e derivs. of *(s)teg- ‘cover’ 3c PNWGmc. *þaką
3b PBalt. *stāgas 3d PCelt. *togiā
We have coded the states of superstate 3 separately, simply because if they are coded
together it is the only shared state.

386 row
Hitt. 1 Av. 7 Luv. 10 Goth. 15
Arm. 2 OCS 8 Lyc. 11 ON 9b
Gk. 3 Lith. 9a TA 12 OHG 16
Alb. 4 OE 9b OPer. 13 Welsh 17 [loan]
TB 5 OI 9b OPru. 14 Osc. 18
Ved. 6 Lat. 9c Latv. 9a Umb. 19
9 derivs. of *h1reh1- 9b o-grade *h1roh1-
9a PEBalt. *irja 9c denom. verb ← compound agent noun
(‘rower’) ← instrument noun (‘oar’)
We have coded states 9a–c, which are very different independent formations from an
inherited root, separately.

387 send
Hitt. 1 Av. 7 Luv. 13 Goth. 10
Arm. 2 OCS 8 Lyc. 14 ON 10
Gk. 3 Lith. 9 TA 5 OHG 10
Alb. 4 OE 10 OPer. 6 Welsh 17
TB 5 OI 11 OPru. 15 Osc. 18
Ved. 6 Lat. 12 Latv. 16 Umb. 19
5 PToch. *luwa- 10 PGmc. *sandīþi
6 PIIr. *iš-

September 2012
 126

388 seven
Hitt. 1 Av. 1 Luv. 2 Goth. 1x
Arm. 1 OCS 1 Lyc. 3 ON 1x
Gk. 1 Lith. 1 TA 1 OHG 1x
Alb. 1 OE 1x OPer. 4 Welsh 1
TB 1 OI 1 OPru. 1 Osc. 5
Ved. 1 Lat. 1 Latv. 1 Umb. 6
1 *septm ̥́
1x → *septm ̥ ́t (by lexical analogy with ‘ten’) > *sepm
̥ ́t > PGmc. *sebun
(cf. Szemerényi 1960:35, 127 fn. 53, Stiles 1985-6, part 3, pp. 6-7)
We have coded the states of superstate 1 separately, simply because if they are coded
together it is the only shared state.

389 shadow [polymorphic]

Hitt. 1 Av. 3 Luv. 8 Goth. 5a
Arm. 2 OCS 3 Lyc. 9 ON 6
Gk. 3 Lith. 4 TA 10 OHG 5a/6
Alb. 3 OE 5a/6 OPer. 11 Welsh 5b
TB 3 OI 5b OPru. 12 Osc. 14
Ved. 3 Lat. 7 Latv. 13 Umb. 15
3 *sḱéh2ih2, *sḱh2iéh2- (cf. Ringe 1996:18-20)
5 *skot- ~ *skōt- 6 PGmc. *skuwwan- (cf. Goth. skuggwa
5a PGmc. *skaduz ‘mirror’)
5b PCelt. *skātom
We have coded states 5a, 5b separately, since they are very different formations.
The polymorphism is confined to Germanic and is leaf-connected.

390 sheep [polymorphic]

[two monomorphic characters by conflated split coding]
Hitt. 1 Av. 7 Luv. 6 Goth. 13
Arm. 2 OCS 6 Lyc. 6 ON 14
Gk. 3 Lith. 6 TA 5 OHG 8
Alb. 4 OE 8 OPer. 10 Welsh 15
TB 5/6 OI 9 OPru. 11 Osc. 16
Ved. 6 Lat. 6 Latv. 12 Umb. 6

September 2012
 127

5 derivs. of PToch. *śaw- ‘live’ 8 PWGmc. *skāp

6 *h2ówis (Kimball 1987:185, 189)
The polymorphism is confined to Tocharian and is leaf-connected; we have reduced the
character to two monomorphic characters by conflated split coding.

391 ship [polymorphic]

[two monomorphic characters, in part by conflated split coding]
Hitt. 1 Av. 3 Luv. 11 Goth. 9
Arm. 2 [loan] OCS 7 Lyc. 12 ON 9
Gk. 3 Lith. 8 TA 5/6 OHG 9
Alb. 4 OE 9 OPer. 3 Welsh 10
TB 5/6 OI 3/10 OPru. 13 Osc. 15
Ved. 3 Lat. 3 Latv. 14 Umb. 16
3 *neh2u- 9 PGmc. *skipą
y
5 PToch. *ọl i- 10 PIns.Celt. *longā
6 PToch. *kọlm-
The Armenian word is probably an Iranian loan, since it is inflected as an a-stem (Olsen
1999:896). On the Celtic forms see McCone 1993:245-8.
All the polymorphisms are local and leaf-connected. Since states 6 and 7 always appear
together, we have coded them as a single state. The remaining polymorphism (in Old
Irish) has been obviated by conflated split coding, with 11 ∪ 3/11 coded against 3 in one
character and all three coded together in the other.
On the preform of state 3 see Szemerényi 1956:185-6 with references; however, unless
and until the absence of *s- can be explained, the derivation of this word from *sneh2-
‘bathe, swim’ (ibid. p. 186, fn. 1 with references) remains doubtful.

392 shoulder [polymorphic, with parallel development]

Hitt. 1 Av. 4 Luv. 10 Goth. 2
Arm. 2 OCS 5 Lyc. 11 ON 7
Gk. 2 Lith. 6 TA 2 OHG 7/8
Alb. 3 OE 7/8 OPer. 12 Welsh 13
TB 2 OI 9 OPru. 6 Osc. 14
Ved. 2 Lat. 2 Latv. 5 Umb. 2
2 *ṓmsos / *ómsos 7 PNWGmc. *ahslu
5 PBS *pletjan 8 PWGmc. *skuldru
6 PBalt. *petijas (vel sim.)

September 2012
 128

The overt polymorphism is confined to West Germanic and is leaf-connected.

The distribution of states in Balto-Slavic shows hidden polymorphism or parallel devel-
opment, since the clear East Baltic subgroup is split.
The preform of state 2 is an unsolved problem, though the words are clearly cognate.
On the difficult problems surrounding the Albanian form see Demiraj 1997:155-6.

393 silver [with parallel development?]

Hitt. 1 Av. 3b Luv. 7 Goth. 6c
Arm. 2 [loan] OCS 6a Lyc. 8 ON 6c
Gk. 3a Lith. 6b TA 5 OHG 6c
Alb. 4 [loan] OE 6c OPer. 3b Welsh 3b
TB 5 OI 3b OPru. 6d Osc. 3b
Ved. 3b Lat. 3b Latv. 6e Umb. 9
3a–b derivs. of *h2erǵ- ‘white’ 6a–e *siCVbrom vel sim. (where *C is a
3b *h2r̥ǵn̥tóm voiced coronal)
5 PToch. *nəәkəәnt- (*ñ-) 6c PGmc. *silubrą
Both the reconstructable meaning of state 3 and the marked variation in the forms
grouped under state 6 suggest parallel development; state 6 probably reflects multiple
borrowing of a non-IE word and/or repeated borrowing between IE languages. Note that
even the Tocharian forms (state 5) do not match perfectly.
On the other hand, it is possible that state 3b arose only once.
We have coded the substates of 3 and 6 separately, but that does not obviate all the diffi-
culties.

394 sister
Hitt. 1 Av. 2 Luv. 1x Goth. 2
Arm. 2 OCS 2 Lyc. 5 ON 2
Gk. 3 Lith. 2 TA 2 OHG 2
Alb. 4 OE 2 OPer. 6 Welsh 2
TB 2 OI 2 OPru. 2 Osc. 8
Ved. 2 Lat. 2 Latv. 7 Umb. 9
1 PAnat. *negos and deriv. 2 *swésōr
1x deriv. of *negnos ‘brother’ ← *negos
We have coded states 1 and 1x together, both because direct replacement is likely and
because they are otherwise unique.

September 2012
 129

395 sister-in-law
Hitt. 1 Av. 7 Luv. 11 Goth. 16
Arm. 2 OCS 3 Lyc. 12 ON 17
Gk. 3 Lith. 8 TA 13 OHG 18
Alb. 4 OE 9 OPer. 14 Welsh 19
TB 5 OI 10 OPru. 15 Osc. 20
Ved. 6 Lat. 3 Latv. 8 Umb. 21
3 *ǵl̥Hōw- (??) 8 PEBalt. *svainē
The preform of state 3 is an unsolved (and probably unsolvable) problem.

396 six
Hitt. 1 Av. 2 Luv. 3 Goth. 2
Arm. 2 OCS 2 Lyc. 4 ON 2
Gk. 2 Lith. 2 TA 2 OHG 2
Alb. 2 OE 2 OPer. 5 Welsh 2
TB 2 OI 2 OPru. 2 Osc. 6
Ved. 2 Lat. 2 Latv. 2 Umb. 7
2 *swéḱs

397 son [polymorphic, with parallel development]

Hitt. 1 Av. 1/3b Luv. 7 Goth. 3b
Arm. 2 OCS 3b Lyc. 8 ON 3b
Gk. 3a Lith. 3b TA 3a OHG 3b
Alb. 4 OE 3b OPer. 1 Welsh 5
TB 3a OI 5 OPru. 3b Osc. 1
Ved. 1/3b Lat. 6 Latv. 9 Umb. 10
1 *putlós 3 derivs. of *suH- ‘give birth’
5 PCelt. *makwkwos 3a *suHyús
3b *suHnús
We have coded states 3a, 3b separately, since they are independent derivatives.
The meaning reconstructable for state 3 argues parallel development; thus the Indo-Ira-
nian polymorphism may or may not extend through many internal nodes, as the distribu-
tion of states suggests.

September 2012
 130

398 son-in-law
[two characters]
Hitt. 1 Av. 3b Luv. 7 Goth. 13
Arm. 2 OCS 3c Lyc. 8 ON 13
Gk. 3a Lith. 3c TA 9 OHG 5
Alb. 3a OE 5 OPer. 10 Welsh 14
TB 4 OI 6 OPru. 11 Osc. 15
Ved. 3b Lat. 3a Latv. 12 Umb. 16
3 *ǵem- (?) 5 PWGmc. *aiþam
3a *ǵm̥ rós (vel sim.) 13 PGmc. *mēgaz ‘kinsman’
3b PIIr. *źāmātar-
3c PBS *žent-
We employ both codings for superstate 3, since we seem to be in the presence of modifi-
cations of a single inherited word.
For discussion of these difficult forms see e.g. Frisk 1960 s.v. γαµβρός. We suggest that
the Latin and Balto-Slavic forms have been influenced by *ǵenh1- ‘be born’ and/or
*ǵneh3- ‘recognize’; the Latvian word, however, has either been replaced by a participle
of the latter root or has been so thoroughly remodelled that assignment of a unique state
is advisable in any case.

399 spin
Hitt. 1 Av. 7 Luv. 11 Goth. 10
Arm. 2 OCS 8 Lyc. 12 ON 10
Gk. 3 Lith. 9 TA 13 OHG 10
Alb. 4 OE 10 OPer. 14 Welsh 3
TB 5 OI 3 OPru. 15 Osc. 16
Ved. 6 Lat. 3 Latv. 9 Umb. 17
3 *snéh1yeti and derivs. 10 PGmc. *spinnidi
9 PEBalt. *verpja
The analysis and etymology of Armenian niwthê are unclear. It is possible that the first
two segments reflect *sneh1-, but Pedersen 1906:426, 436-7 suggests an alternative
analysis n- + *hiwth-, the latter < *pi-ubh-t- (PIE *webh- ‘weave’). We have cautiously
assigned Armenian a separate state.

September 2012
 131

400 spring
[two characters]
Hitt. 1x Av. 1 Luv. 6 Goth. 10
Arm. 1 OCS 1 Lyc. 7 ON 1
Gk. 1 Lith. 1y TA 8 OHG 4
Alb. 2 [loan] OE 4 OPer. 1 Welsh 1
TB 3 OI 5 OPru. 9 Osc. 11
Ved. 1 Lat. 1 Latv. 1y Umb. 12
1 *wésr̥, *wesn- and derivs. 4 PWGmc. *langitīn-
1x compound with *h2ent-
1y PEBalt. *pavasarjas (cf. 403 ‘summer’)
We have employed both codings for superstate 1, since direct replacement of the inher-
ited word by its derivatives seems likely.
The analysis of the Hittite form accepted here is not completely certain; see Puhvel 1991:
73-5 for discussion. The form of the Latin cognate must be the result of a sequence of
sound changes and paradigmatic levellings, roughly *wesor, *wesn- > *wesor, *wēn- →
*wēr, *wēn- → vēr, vēr-.

401 stay [polymorphic, with parallel development]

[two monomorphic characters by conflated split coding]
Hitt. 1 Av. 7 Luv. 11 Goth. 4
Arm. 2 OCS 6x Lyc. 12 ON 14
Gk. 2 Lith. 8 TA 5 OHG 15
Alb. 3 OE 9 OPer. 13 Welsh 16
TB 4/5 OI 10 OPru. 8 Osc. 17
Ved. 4/6 Lat. 2 Latv. 8 Umb. 18
2 *men- 6 *steh2- ‘stand’
4 *h2wes- ‘stay overnight’ 6x reflexive of cpd. of deriv.
y
5 PToch. *kəәl əәtəәr ‘stand’ 8 PBalt. *(pa)leik- (< *leykw- ‘leave’)
We have coded states 6 and 6x separately, since there is no indication of any direct his-
torical connection between them; the overt polymorphism, which is confined to Vedic,
thereby becomes ineffective.
We have obviated the Tocharian B polymorphism by conflated split coding.
The meanings reconstructable for states 2 and 4 argue parallel development.

September 2012
 132

402 steal [polymorphic]

[two monomorphic characters by double conflated split coding]
Hitt. 1 Av. 7 Luv. 13 Goth. 3/10
Arm. 2 OCS 8 Lyc. 14 ON 10
Gk. 3 Lith. 9 TA 5 OHG 10
Alb. 4 OE 10 OPer. 15 Welsh 18 [loan]
TB 3/5 OI 11 OPru. 16 Osc. 19
Ved. 6 Lat. 12 Latv. 17 Umb. 20
3 *klep- 10 PGmc. *stilidi (*stela-)
5 PToch. *kəәrka-, pres. *kəәrna(ṣəә)
The polymorphisms are local and leaf-connected. We have obviated them by double
conflated split coding, as follows. In one character, 5 ∪ 3/5 is coded against 3 ∪ 3/10 ∪
10; in the other, 5 ∪ 3/5 ∪ 3 is coded against 3/10 ∪ 10.

403 summer
[two characters]
Hitt. 1 Av. 2 Luv. 9 Goth. 13
Arm. 2 OCS 6 Lyc. 10 ON 2x
Gk. 3 Lith. 7 TA 2y OHG 2x
Alb. 4 OE 2x OPer. 11 Welsh 2
TB 2y OI 2 OPru. 12 Osc. 14
Ved. 5 Lat. 8 Latv. 7 Umb. 15
2 *semH- ~ *sm ̥ H- and derivs. 7 PEBalt. *vasarā
2x PGmc. *sumaraz
2y PToch. *ṣəәmay-
We have employed both codings for superstate 2.

404 sweat [with parallel development]

Hitt. 1 Av. 2 Luv. 6 Goth. 10
Arm. 2 OCS 4 Lyc. 7 ON 2
Gk. 2 Lith. 5 TA 8 OHG 2
Alb. 3 OE 2 OPer. 9 Welsh 2
TB 2 OI 1 OPru. 5 Osc. 11
Ved. 2 Lat. 2 Latv. 2 Umb. 12
1 *all- 5 PBalt. *prakait-
2 *sweyd-

September 2012
 133

Note that both the Celtic subgroup and East Baltic are split; such a distribution of states
argues parallel development.

405 sweet
Forms a polymorphic set with 352 honey (q.v.).

406 ten
Hitt. 1 Av. 2 Luv. 3 Goth. 2
Arm. 2 OCS 2 Lyc. 4 ON 2
Gk. 2 Lith. 2 TA 2 OHG 2
Alb. 2 OE 2 OPer. 5 Welsh 2
TB 2 OI 2 OPru. 2 Osc. 6
Ved. 2 Lat. 2 Latv. 2 Umb. 2
*déḱm̥t

407 thousand
[two characters]
Hitt. 1 Av. 3b Luv. 9 Goth. 6
Arm. 2 [loan] OCS 6 Lyc. 10 ON 6
Gk. 3a Lith. 6 TA 5 OHG 6
Alb. 4 [loan] OE 6 OPer. 11 Welsh 12 [loan]
TB 5 OI 7 [loan] OPru. 6 Osc. 13
Ved. 3b Lat. 8 Latv. 6 Umb. 14
3 derivs. of *ǵhéslo- 5 PToch. *wyəәltsë
3a *ǵhéslio- (adj.?) 6 *túHsn̥t-
3b PIIr. *saźháslam < *sm ̥ -ǵhéslo-m
We have employed both codings for superstate 3, since a direct historical connection be-
tween the substates seems probable.
That Lat. mīlle belongs with state 3 remains unprovable (and improbable, since it in-
volves positing a zero-grade feminine derivative of *ǵhéslo- compounded with *sémih2
→ *smíh2 ‘one’).

September 2012
 134

408 twenty
Hitt. 1 Av. 2 Luv. 6 Goth. 5
Arm. 2 OCS 4 Lyc. 7 ON 5
Gk. 2 Lith. 4 TA 2 OHG 5
Alb. 3 OE 5 OPer. 8 Welsh 2
TB 2 OI 2 OPru. 9 Osc. 10
Ved. 2 Lat. 2 Latv. 4 Umb. 11
2 *wī́ḱm
̥ tih1
4 PBS phrase *dvai dešimtī (vel sim.) ‘two tens’
5 PGmc. phrase *twai tigiwiz (vel sim.) ‘two decads’
On the Albanian form see Demiraj 1997:425; on Lycian kbisñtãta see Melchert 2004 s.v.

409 udder
Hitt. 1 Av. 6 Luv. 10 Goth. 15
Arm. 2 OCS 7 Lyc. 11 ON 3
Gk. 3 Lith. 8 TA 12 OHG 3
Alb. 4 OE 3 OPer. 13 Welsh 16
TB 5 OI 9 OPru. 14 Osc. 17
Ved. 3 Lat. 3 Latv. 8 Umb. 18
3 *ówdhr̥ ~ *uHdhén- 8 PEBalt. *tešmen-

410 wasp
Hitt. 1 Av. 7 Luv. 10 Goth. 15
Arm. 2 OCS 8 Lyc. 11 ON 16
Gk. 3 Lith. 7 TA 12 OHG 17
Alb. 4 OE 7 OPer. 13 Welsh 18
TB 5 OI 9 OPru. 7 Osc. 19
Ved. 6 Lat. 7 Latv. 14 Umb. 20
h
7 *wob seh2
It is possible, but far from certain, that Latvian lapsene belongs with set 7 and owes its
initial consonant to some lexical analogy.

September 2012
 135

411 wear [polymorphic]

[two monomorphic characters, in part by conflated split coding]
Hitt. 1 Av. 1 Luv. 1 Goth. 1x
Arm. 1 OCS 4 Lyc. 8 ON 11
Gk. 2x Lith. 5 TA 1 OHG 12
Alb. 3 OE 1x OPer. 9 Welsh 1
TB 1/2 OI 6 OPru. 10 Osc. 13
Ved. 1 Lat. 7 Latv. 4/5 Umb. 14
1 *wéstor and derivs. 4 derivs. of PBS *neš- ‘carry’
1x *woséyeti ‘clothe’ (→ intrans.) 5 derivs. of PEBalt. *velk-
h
2 *b éreti ‘carry’
2x intensive *bhoréyeti ‘carry around’
We have employed both codings for superstate 1, since a direct replacement of the basic
intransitive verb by its causative appears likely. We code states 2 and 2x separately; the
Tocharian B polymorphism thus becomes ineffective.
The remaining polymorphism is confined to East Baltic and is leaf-connected. We have
obviated it by conflated split coding, with 5 ∪ 4/5 coded against 4 in one character and
all three coded together in the other.

412 weave [with parallel development]

Hitt. 1 Av. 5 Luv. 10 Goth. 13
Arm. 2 OCS 6 Lyc. 11 ON 3
Gk. 3 Lith. 7 TA 3 OHG 3
Alb. 4 OE 3 OPer. 12 Welsh 8
TB 3 OI 8 OPru. 6 Osc. 14
Ved. 4 Lat. 9 Latv. 7 Umb. 15
h
3 *web - 7 PEBalt. *audja
4 *weyh1- ‘plait’ 8 PCelt. *weg-
6 PBS *tuk-
The (probable) reconstructed meaning of state 4 argues parallel development.

September 2012
 136

413 wheel [polymorphic, with parallel development]

Hitt. 1 Av. 6a Luv. 8 Goth. 11
Arm. 2 OCS 6b Lyc. 9 ON 6a
Gk. 3 Lith. 7 TA 5 OHG 7
Alb. 4 [loan] OE 6a OPer. 10 Welsh 7
TB 5 OI 3/7 OPru. 6c Osc. 12
Ved. 6a Lat. 7 Latv. 7 Umb. 13
3 *dhroghós ‘runner’ 6a–c derivs. of *kwel- ‘turn’
5 PToch. *wyVrk(w)əәnt- 6a *kwékwlos
7 *(H)rotós ‘runner’, coll. *(H)róteh2
We have coded states 6a–c separately, since they are independent derivatives.
The Tocharian forms do not match perfectly, and the Hittite word is so dissimilar that we
have coded it separately. On the Armenian form see Olsen 1999:23.
The meanings reconstructable for most states argue massive parallel development; there-
fore polymorphism, which is overt in OIr., may or may not also be present at numerous
internal nodes.

414 widow
Hitt. 1 Av. 2 Luv. 6 Goth. 2
Arm. 2 OCS 2 Lyc. 7 ON 11
Gk. 3 Lith. 5 TA 8 OHG 2
Alb. 2 OE 2 OPer. 9 Welsh 2
TB 4 OI 2 OPru. 2 Osc. 12
Ved. 2 Lat. 2 Latv. 10 Umb. 13
h h
2 *h1wid éwh2 ~ *h1wid wéh2-
Ablaut of the vowel before the second *w argues a proterokinetic paradigm (Lionel
Joseph, p.c.). On the Armenian form see Cowgill 1983.

415 winter
Forms a polymorphic set with 153 snow (q.v.).

September 2012
 137

416 wolf [probably with parallel development]

Hitt. 1 Av. 3 Luv. 5 Goth. 3
Arm. 2 OCS 3 Lyc. 6 ON 3
Gk. 3 Lith. 3 TA 7 OHG 3
Alb. 3 OE 3 OPer. 3 Welsh 8
TB 4 OI 2 OPru. 3 Osc. 9
Ved. 3 Lat. 3 Latv. 3 Umb. 10
2 *waylos 3 *wĺ̥kwos
On the etymology of the Armenian form cf. Hübschmann 1897:431, Olsen 1999:34; the
alternative we have adopted here seems preferable to positing an otherwise unsupported
*wĺ̥yos or the like.
If it is true that state 2 is derived from onomatopoeic *wáy (Olsen, loc. cit.), the character
exhibits parallel development.

417 wood [polymorphic]

Hitt. 1 Av. 1 Luv. 1 Goth. 1
Arm. 2 OCS 1 Lyc. 7 ON 1/5
Gk. 3 Lith. 4 TA 1 OHG 5
Alb. 1 OE 1/5 OPer. 1 Welsh 9
TB 1 OI 5 OPru. 4 Osc. 10
Ved. 1 Lat. 6 Latv. 8 Umb. 11
1 *dóru and derivs. 5 *widhus
4 PBalt. *malkā
The polymorphism is confined to Germanic and is leaf-connected.

418 wool
Hitt. 1 Av. 1 Luv. 1 Goth. 1
Arm. 2 OCS 1 Lyc. 6 ON 1
Gk. 3 Lith. 1 TA 7 OHG 1
Alb. 4 OE 1 OPer. 8 Welsh 1x
TB 5 OI 1x OPru. 9 Osc. 10
Ved. 1 Lat. 1 Latv. 1 Umb. 11
́
1 *h2wl̥h1no-, collective *h2wl̥h1neh2
1x PCelt. *wlanā, with unexpected short vowel in first syllable
The Irish word appears to have been deformed by lexical analogy (cf. Pedersen 1909:
179).

September 2012
 138

We have coded substates 1 and 1x separately, since otherwise the character is unin-
formative.

419 yoke
Hitt. 1 Av. 4 Luv. 5 Goth. 1
Arm. 1 OCS 1 Lyc. 6 ON 1
Gk. 1 Lith. 1 TA 7 OHG 1
Alb. 2 OE 1 OPer. 8 Welsh 1
TB 3 OI 1 OPru. 9 Osc. 10
Ved. 1 Lat. 1 Latv. 1 Umb. 11
1 *yugóm and derivs.
The initial l- of the Armenian word must be the result of some lexical analogy. Various
languages have introduced -n- under the influence of the related verb *yunégti ‘join’;
since that can easily have been a parallel development, we have not taken account of it in
coding.

420 young
[two characters]
Hitt. 1 Av. 5 Luv. 6 Goth. 5x
Arm. 2 OCS 5 Lyc. 7 ON 5x
Gk. 2 Lith. 5 TA 8 OHG 5x
Alb. 3 OE 5x OPer. 9 Welsh 5x
TB 4 OI 5x OPru. 10 Osc. 11
Ved. 5 Lat. 5 Latv. 5 Umb. 5
2 *néwos ‘new’ and derivs. 5 *h2yuh1én-
5x *h2yuh1n̥ḱós
We have adopted both codings for superstate 5, since a direct replacement of the basic
word by its derivative is plausible.

421 tear
Hitt. 1 Av. 2b Luv. 5 Goth. 2a
Arm. 2a OCS 4 Lyc. 6 ON 2a
Gk. 2a Lith. 2b TA 2b OHG 2a
Alb. 3 OE 2a OPer. 7 Welsh 2a
TB 2b OI 2a OPru. 8 Osc. 9
Ved. 2b Lat. 2a Latv. 2b Umb. 10

September 2012
 139

2a *dáḱru 2b *áḱru
There is an obvious relation between the forms represented by the two large states, but its
exact nature remains obscure. Though the Hittite word clearly resembles them (mainly
because it ends in -ru), it is too different from either set to be assigned the same state.
We have coded states 2a, 2b separately, simply because there is otherwise only one
shared state.

Bibliography.
Adams, Douglas. 1999. A dictionary of Tocharian B. Amsterdam: Rodopi.
Ancillotti, Augusto, and Romolo Cerri. 1996. La civiltà degli Umbri. Perugia: Jama.
Bartholomae, Christian. 1979. Altiranisches Wörterbuch, zusammen mit den Nachar-
beiten und Vorarbeiten. Berlin: de Gruyter.
Braune, Wilhelm, and Ingo Reiffenstein. 2004. Althochdeutsche Grammatik I: Laut-
und Formenlehre. 15th ed. Tübingen: Niemeyer.
Buck, Carl Darling. 1949. A dictionary of selected synonyms in the principal Indo-
European languages. Chicago: U. Chicago Press.
Cardona, George, and Norman H. Zide (edd.). 1987. Festschrift for Henry Hoenigs-
wald. Tübingen: Narr.
Clackson, James. 1994. The linguistic relationship between Armenian and Greek.
Oxford: Blackwell.
Cowgill, Warren. 1960. Greek ou and Armenian oč‘. Language 36.347-50.
—. 1973. The source of Latin stāre, with notes on comparable forms elsewhere in
Indo-European. JIES 1.271-303.
—. 1983. Intervocalic *dh and *t in Armenian and the Armenian endings of the 1st and
2nd plural pronouns. Paper read at the 3rd East Coast Indo-European Conference,
Philadelphia.
Demiraj, Bardhyl. 1997. Albanische Etymologien. Amsterdam: Rodopi.
Eichner, Heiner. 1992. Anatolian. Gvozdanivić, Jadranka (ed.), Indo-European numer-
als (Berlin: de Gruyter) 29-96.
Endzelīns, Jānis. 1923. Lettische Grammatik. Heidelberg: Winter.
Ernout, Alfred. 1961. Le dialecte ombrien. Paris: Klincksieck.
Evans, D. Simon. 1964. A grammar of Middle Welsh. Dublin: Institute for Advanced
Studies.
Feist, Sigmund. 1939. Vergleichendes Wörterbuch der gotischen Sprache. 3rd ed.
Leiden: Brill.
Frisk, Hjalmar. 1960. Griechisches etymologisches Wörterbuch. Band I. Heidelberg:

September 2012
 140

Winter.
—. 1970. —. Band II. Heidelberg: Winter.
Hilmarsson, Jörundur. 1996. Materials for a Tocharian historical and etymological
dictionary. Reykjavík: Málvísindastofnun Háskóla Íslands.
Hübschmann, H. 1897. Armenische Grammatik. I. Theil. Armenische Etymologie.
Leipzig: Breitkopf & Härtel.
—. 1899. Self-review of Hübschmann 1897, II. Abteilung. Anzeiger für indogerman-
ische Sprach- und Altertumskunde 10.41-50.
Jasanoff, Jay H. 1989. Language and gender in the Tarim basin: the Tocharian 1 sg.
pronoun. TIES 3.125-47.
Katz, Joshua T. 1997. Ein tocharisches Lautgesetz für Monosyllaba. TIES 7.61-81.
———. 1998. Topics in Indo-European personal pronouns. Harvard dissertation.
———. 2004. The ‘swimming duck’ in Greek and Hittite. Penney, J. H. W. (ed.),
Indo-European perspectives: studies in honour of Anna Morpurgo Davies.
Oxford: OUP.
Kim, Ronald. 2000. ‘To drink’ in Anatolian, Tocharian, and Proto-Indo-European. HS
113.151-70.
Kimball. Sara. 1987. *H3 in Anatolian. Cardona and Zide (edd.) 1987:185-92.
Krause, Wolfgang. 1953. Handbuch des Gotischen. Munich: Beck.
Lewis, Henry, and Holger Pedersen. 1961. A concise comparative Celtic grammar. 3rd
ed. Göttingen: Vandenhoeck & Ruprecht.
Lloyd, Albert L., Rosemarie Lühr, and Otto Springer. 1998. Etymologisches Wörter-
buch des Althochdeutschen. Band II. Göttingen: Vandenhoeck & Ruprecht.
Mayrhofer, Manfred. 1992. Etymologisches Wörterbuch des Altindoarischen. I. Band.
Heidelberg: Winter.
McCone, Kim. 1993. Zisalpinisch-gallisch uenia und lokan. Heidermanns, Frank, Hel-
mut Rix, and Elmar Seebold (edd.), Sprachen und Schriften des antiken Mittel-
meerraums: Festschrift für Jürgen Untermann (Innsbruck: Institute für Sprach-
wissenschaft der Universität) 243-9.
Melchert, H. Craig. 1989. PIE ‘dog’ in Hittite? MSS 50.97-101.
—. 1992. Relative chronology and Anatolian: the vowel system. Beekes, Robert, Alex-
ander Lubotsky, and Jos Weitenberg (edd.), Rekonstruktion und relative Chrono-
logie (Innsbruck: Institut für Sprachwissenschaft der Universität) 41-53.
—. 1993. Cuneiform Luvian lexicon. Chapel Hill (desktop publication).
—. 1994. Anatolian historical phonology. Amsterdam: Rodopi.
—. 2004. A dictionary of the Lycian language. Ann Arbor: Beech Stave Press.

September 2012
 141

Nussbaum, Alan. 1986. Head and horn in Indo-European. Berlin: de Gruyter.

Olsen, Birgit. 1999. The noun in Biblical Armenian. Berlin: Mouton de Gruyter.
Pedersen, Holger. 1906. Armenisch und die Nachbarsprachen. ZVS 39.334-485.
—. 1909. Vergleichende Grammatik der keltischen Sprachen. I. Band. Göttingen:
Vandenhoeck & Ruprecht.
—. 1913. —. 2. Band. Göttingen: Vandenhoeck & Ruprecht.
Peters, Martin. 1991. Ein tocharisches Auslautproblem. Die Sprache 34.242-4.
Pinault, Georges-Jean. 2002. Tocharian and Indo-Iranian: relations between two
linguistic areas. Sims-Williams, Nicholas (ed.), Indo-Iranian languages and
peoples (= Proceedings of the British Academy 116; Oxford: OUP) 243-84.
Porzig, Walter. 1954. Die Gliederung des indogermanischen Sprachgebiets. Heidel-
berg: Winter.
Puhvel, Jaan. 1991. Hittite etymological dictionary. Volume 3. Berlin: Mouton de
Gruyter.
Quin, E. G., et al. 1983. Dictionary of the Irish language. Dublin: Royal Irish Acad-
emy.
Ringe, Don. 1996. On the chronology of sound changes in Tocharian. Vol. 1. New
Haven: American Oriental Society.
Ringe, Don, and Ann Taylor. Forthcoming. The development of Old English. (= A
linguistic history of English, Vol. 2.) Oxford: OUP.
Rix, Helmut. 1991. Nochmals griech. νῆττα/νῆσσα/νᾶσσα. HS 104.186-98.
Rix, Helmut, et al. 2001. Lexikon der indogermanischen Verben. Revised ed. Wies-
baden: Reichert.
Schindler, Jochem. 1966. Hethitisch lišši „Leber“. Die Sprache 12.77-8.
Schmidt, Gernot. 1978. Stammbildung und Flexion der indogermanischen Personalpro-
nomina. Wiesbaden: Harrassowitz.
Schmidt, Klaus T. 1994. Zur Erforschung der tocharischen Literatur: Stand und Auf-
gaben. Schlerath, Bernfried (ed.). 1994. Tocharisch. (Reykjavík: Málvísinda-
stofnun Háskóla Íslands) 239-83.
—. 1996. Das tocharische Maitreyasamitināṭaka in Vergleich mit der uigurischen
Maitrisimit. Emmerick, Ronald, et al. (edd.), Turfan, Khotan, and Dunhuang
(Berlin: Akademie Verlag) 269-78.
Schumacher, Stefan. 2004. Die keltischen Primärverben. Innsbruck: Institut für
Sprachwissenschaft der Universität
Seebold, Elmar. 1970. Vergleichendes und etymologisches Wörterbuch der german-
ischen starken Verben. The Hague: Mouton.

September 2012
 142

Sommer, Ferdinand. 1948. Handbuch der lateinischen Laut- und Formenlehre. 2nd and
3rd ed. Heidelberg: Winter.
Stang, Christian. 1966. Vergleichende Grammatik der baltischen Sprachen. Oslo:
Universitetsforlaget.
Starke, Frank. 1987. Die Vertretungen von uridg. *dhugh2tér- „Tochter“ in den
luwischen Sprachen und ihre Stammbildung. KZ 100.243-69.
Szemerényi, Oswald. 1956. Latin rēs and the Indo-European long-diphthong stem
nouns. KZ 73.167-202.
—. 1960. Studies in the Indo-European system of numerals. Heidelberg: Winter.
—. 1966. Iranica II. Die Sprache 12.190-226.
Thurneysen, Rudolf. 1946. A grammar of Old Irish. Dublin: Dublin Institute for
Advanced Studies.
Vetter, Emil. 1953. Handbuch der italischen Dialekte. I. Band. Heidelberg: Winter.
Vries, Jan de. 1962. Altnordisches etymologisches Wörterbuch. 2nd ed. Leiden: Brill.
Watkins, Calvert. 1969. Indogermanische Grammatik. Band III: Formenlehre. Erster
Teil: Geschichte der indogermanischen Verbalflexion. Heidelberg: Winter.
—. 1987. Two Anatolian forms: Palaic aškumāuwa-, Cuneiform Luvian wa-a-ar-ša.
Cardona and Zide (edd.) 1987:399-404.
Winter, Werner. 1985. “Left” or “right”? Fisiak, Jacek (ed.), Historical semantics and
historical word formation (Berlin: de Gruyter) 583-95.

September 2012