Author Archives: vellaunos

Here a few of the more interesting etymologies that I’ve come up with in the course of the research that I’ve been doing for the Vindonian language that I’m creating. Following are entries from my Vindonian wordlist, along with some commentary…

glaghu (glaghwa) m : rain [< *glagwon < *glaghwom ‘clatter, rattle, pitter-patter’; B glav; W glaw] {cf. PGmc *klakkōną > Eng clack & Fr claquer; cf. also Lat clangere (clangō) ‘clang, sound out’}

The usual reconstruction of the Proto-Celtic is *glaw-, but this would be the same as the Proto-Celtic word for “coal” (*glaw-, not *glow-). Note the complete disappearance of medial -g- in Brittonic. [Added August 21, 2022] I eventually realized some time after posting this that Latin clangere does not belong in this etymology due to its initial c- instead of g-. Instead, Latin clangere is related to Gaulish *clocca ‘bell’, OIr cluiche ‘game, play’ (< PrCelt *klokjon) and PrGmc *hlahhjaną > Eng laugh.

gníjom (gníjed) : create [< *gnīje/o– < *gnēje/o– < *g’neh₁-je/o-; G neI anmanbe gniIou ‘I do not create by names’ (Châteaubleau); OIr gníid ‘make, do’] >>> genjom

adnajom (adnajed) : recognize [< *ati-gna– < *ati-gnina– < *g’ṇh₃-na-; PBrit *ati-gna-bute/o– > B anavout, anavezout & W adnabod, nabod; OIr ad-gnin < *ati-gnina-] {cf. PGmc *kunnaną (< *g’ṇh₃-na-) > Eng can, Du kunnen, Ger können} >>> gnájom

This etymological discussion for gníjom should be compared with the discussion of gniIou on page 182 of Xavier Delamarre’s Dictionnaire de la langue gauloise in which gniIou is related to Old Irish ad-gnin ‘know, recognize’ rather than Old Irish gníid (or gniïd). As indicated in the second entry above (adnajom), Old Irish ad-gnin is in fact related to Breton anavout and Welsh adnafod. A Gaulish cognate to these would probably have been either *adgnināmi or perhaps *adgnāmi (or possibly *adgnināIou/*adgnāIou).

kaffom (kaffed) : find [< *kaffe/o– < *kasφe/o– < *kaspe/o-; B kavout < MB caffout; C kavos ‘find, get, have’; W caffael, caffel ‘get, obtain’ (merged with cael)] {cf. PGmc *haspijǭ > Eng hasp} >>> kaghom

kaghom (kaghed) : have, contain [< *kage/o– < *kagh-; B kaout; W cael] >>> kaffom

It is usually thought that the Breton and Welsh verbs without a medial consonant (kaout/cael) are modified forms of the verbs with -v-/-ff- (kavout/caffael). My opinion is that these are in fact two separate verbs – only Breton has kept them totally distinct. Note the development of -ff- from -sφ- in the first verb above.

kuls (kulsa) m : vagina [< *kultson < *kuls-tom; B kourzh ‘vulva’] {cf. PGmc *hulistrą > Eng holster}

I think the semantic connection between “holster” and “vagina” is clear enough.

uzar : cold [< *usaros < *h₁eus– ‘burn, singe’; W oer; OIr úar] {meaning of *h₁eus– transferred in Celtic from sensation of heat to sensation of cold; W and OIr < *usaros rather than **ougros; for G month name Ogronios >>> aireine; OIr úacht ‘cold(ness)’ < *us-axtus (not **ouxtus); cf. Lat ūrere (ūrō) ‘burn’, Gr heuein (heuō) ‘singe’, PGmc *uzjǭ ‘fire’ > *aimuzjǭ > Eng ember; no relation to Lat auctumnus, autumnus < *h₂eug– ‘increase’; possible relation to Latv auksts ‘cold’ < austs (with intrusive –k-) < *aus– or *aus-sk– ?< *h₁ous-(ske/o-)}

I have believed for a long time that the form *ougros as the origin of Old Irish úar and Welsh oer did not in fact exist, all evidence for it being either unconvincing or mistaken.

varedom (vareded) : help, assist [< *warete/o-; B gwarediñ ‘shelter, protect’; W gwared, gwaredu ‘save, redeem, deliver, rid’; OIr fo-reith ‘help, aid, succour’] {cf. PGmc *wardāną > Eng ward, as well as PGmc *warjaną ‘ward off’, *warnōną ‘warn’, *warōną ‘watch’} {alternatively explained as from *worete/o– < *uφo-rete/o– ‘run beneath’, which would be a calque of Lat succurrere, on the basis of the (probably artificial) OIr form, but *worete/o– would normally have become **goured– in B and **gored– in W}

There are a few examples of words being artificially modified due to a resemblance with other etymologically unrelated words, such as Old Irish fo-reith in this instance. Such artificially modified words can obviously be misleading. Another example of this would be Old High German weralt “world” which appears to be from wer “man” + alt “age” but is in fact from Proto-Germanic *weruldiz. I dealt with this Proto-Germanic term in my post called “Vellaunos” – https://vellaunos.ca/2021/03/27/vellaunos/

zof : wonderful, delightful [< *soffos < *sosφos < *sos-pos; W hoff ‘dear, fond, favorite’; ? cf. OIr subaigidir (< *sosφ-age/o-) ‘delight in’] {cf. Lat sospes ‘saving, delivering’ < *sos-pets; cf. also PGmc *samftijaz (< *som-p(e)ti-) > Eng soft, Ger sanft}

At first, I thought that the Welsh hoff might be from the Latin sospes, but I eventually realized that Lat sospes probably would have become **sosb in Welsh. Note also that this word shows the development of -ff- from -sφ-, as in kaffom above.

nedjom (nedjed) : fly [< *netje/o– (?); B nijal < MB nigal; C neyja < MC nyge; W neidio ‘jump, leap’] {g in MB & MC for [ʒ] or [dʒ]; W naid < *neid; no relation to G duscelinatia < *dus-kelin(o)-at(o)-ja}

zedjom (zedjed) : flap [< *setje/o-; B hejañ ‘shake, wave’; W hedeg ‘fly’]

zontjom (zontjed) : direct, point to [< *sontje/o-; B heñchañ ‘conduct, direct, guide’] {cf. PGmc *sandijaną (< *sont-eje/o-) > Eng send}

These three terms show the development in Breton (and Cornish) of the affricates -tsh- and -dzh- (later -sh- and -zh-) from -tj- and -dj-, this development not occurring in Welsh.

nit (nitta) m : nest [< *nitton < *nisdom; B neizh; W nyth] {cf. Lat nīdus; cf. also PGmc *nestą > Eng nest} >>> rattom for -sd- > -tt-

rattom (ratted) : scrape [< *ratte/o– < *rasde/o-; B razhañ; W rhathu] {cf. Lat rādere (rādō) < *rasde/o-; cf. also Lat rōdere (rōdō) ‘gnaw’; cf. also Skr radati} >>> nit for -sd- > -tt-

These two terms show the development of -tt- from -sd- (probably) in Proto-Celtic, this later becoming -th- in Brittonic.

jegom (jeged) : say [< *jeke/o-; Late G (Châteaubleau) Iegumi, Iexsete, Iexstumi, IegiIinna] {voicing of intervocalic –k– to –g– in Late Gaulish as also in Brittonic} {cf. Lat jocus ‘joke’; cf. also PGmc *jehaną ‘speak, say’} >>> monegijom

The Châteaubleau tile interestingly shows the voicing of the k sound to a g sound in late Gaulish, as also happened in Brittonic.

ander : original, primeval, primordial, primitive, pristine [< *anderos < *ṇdheros; G brixtia anderon “by the magic of the primeval ones” (Chamalières)] {cf. PGmc *underaz > Eng under; cf. also Skr adhara ‘low, inferior’}

andern (anderne) f : virgin [< *andernā < *ander(o)-nā “pristine one”; B annoar (annoared, annoarezed) ‘heifer’; W anner (aneirod, anneri, annerau) ‘heifer’; G andernados “of the (sorority of) virgins” (Larzac); OIr ainder ‘virgin, maiden’ > Ir ainnir ‘girl, maiden, lass’] {for the –ernā ending, cf. PGmc *þewernǭ ‘servant girl, maid’} {The word andernados on the lead tablet of Larzac is usually thought to refer to the “infernal ones”. The word ueronadas which also appears on Larzac is usually thought to mean “upper ones” (uer– < *uφer), apparently the antonym of andernados. But just as I connect the word andernados with the word anderna “virgin”, I also connect the word ueronadas with a possible *uerona “married” (literally “manned”), this being a derivative of uer “man”: mnas ueronadas would mean “married women” (literally “manned women”). Rather than an infernal/upper opposition, there may actually be a virginal/married opposition.}

The first of these two entries relates to my recent post about Proto-Celtic ītselos (https://vellaunos.ca/2021/03/27/proto-celtic-itselos/) in which I pointed out that Latin īnferus is certainly not an exact cognate of English under, Sanskrit adhara or Gaulish anderos (under ‘anderos’, p. 47, Dictionnaire de la langue gauloise – “andero– est l’exact équivalent phonétique du Latin īnferus…”). The second entry already has its own rather full commentary…

One of my interests since childhood has been languages. One of my favorite articles from the 1972 edition of Collier’s Encyclopedia that we had when I was a kid was the one on Latin. The best part of that article was its good description of the phonetic development of Latin from Proto-Indo-European.

The languages that I have been interested in over the years have included the following: Latin, Ancient Greek, Sanskrit, Avestan, Gothic, Old English, Old Saxon, Old High German, Gaulish, Welsh, (Old) Irish, Sumerian, Akkadian, Ancient Egyptian, Hebrew, Arabic, Ojibway, Cree. I have acquired many reference works on all of these languages (many of which I no longer have). Unfortunately, I’ve never studied any of these languages enough to gain any kind of fluency.

Along with my interest in actual historical or current languages, I’ve worked on a few artificial, or constructed, languages (i.e. conlangs). The first one, called Reshish, was created entirely from scratch. The few words I remember from this language are him ‘earth’, har ‘sky’ and tom’lael ‘cut’ (actually, this last word was “influenced” by Greek a-tomos ‘uncut’).

After I discovered the Gaulish language (mostly due to a website that Christopher Gwinn used to have), I decided to try to recreate Gaulish, or at least something like it. This “reconstruction” was called Danuvjâcon (i.e. Danubian): the reasoning for this name was the idea that the Proto-Celtic language had developed somewhere in the area of southern Germany/Austria/Hungary and that Gaulish was a direct continuation of Proto-Celtic. Although I still believe that Proto-Celtic originated somewhere in Central Europe, others prefer to think that it developed in the Atlantic regions (where the Celtoid languages remain to this day).

I posted a wordlist and grammatical material for Danuvjâcon on a website that I once had (I think it was an Angelfire website). That website hasn’t existed for a long time but an Ancient Celtic conlang Yahoo! Group that I once belonged to might still have that information. Interestingly, I recently found a reference to Danuvjâcon on a Neo-Druidic website. In hindsight, I have to admit that Danuvjâcon was much more a product of what I thought Ancient Celtic should be, rather than what it probably was.

I have worked on two other conlangs in the past few years (off and on). One of them is an attempt at deriving a distinct Indo-European language directly from reconstructed Proto-Indo-European. It would essentially constitute its own branch of Indo-European and be a cousin of Sanskrit, Ancient Greek and Latin, if you will.

The big problem with this project is that there are often things and concepts for which no common word can clearly be reconstructed to Proto-Indo-European. Words for ‘father’ (pater), for ‘horse’ (ekwos) and for ‘carry/bear’ (ber-) are easy enough, but there are no common Proto-Indo-European words that can clearly mean ‘leg’ or ‘wall’ or ‘itch’.

Working on a Germanic conlang is much easier because the Germanic language family is much more homogenous than most other language families. For this simple reason, the conlang that I’ve been most successful with has been my West Germanic conlang.

The motivation to create a West Germanic conlang came largely from my discovery of the Old Saxon language through James E. Cathey’s Hêliand: Text and Commentary. After the text selections and the commentary on these is a section with a fairly complete description of Old Saxon grammar as well as a wordlist from the text. Along with Cathey’s Old Saxon resource, I’ve consulted reference works on Old English and Old High German as well as Gothic (Joseph Wright’s works for the last two), and Vladimir Orel’s A Handbook of Germanic Etymology. I’ve also consulted dictionaries of Modern German and Dutch, and I’ve made much use of the etymologies on Wiktionary.

I decided to call my West Germanic conlang “Frankish” but with the following caveat: it would be something like the Old Low Frankish which developed into Dutch rather than the West Central German (i.e. Franconian) dialect that the Carolingians apparently used (no High German consonant shift). Also, my version of “Frankish” is phonetically more conservative than Old Low Frankish probably was (especially in that the diphthongs ai and au are retained rather than becoming long e and long o). Still, I like to call my West Germanic conlang “Frankish” – sa Frenkish spraaka.

Here is a small sample of this “Frankish” language (for nouns and adjectives, the second and third columns are singular and plural; for verbs, the second column is the infinitive, the third column is the present third person singular, and the fourth column is the past third person singular):

apple                     appel                     apla

and                        and

beautiful              shoen                   shoena

big                          mikel                     mikla

bigger, more      maira                     mairen

bird                        fogel                      fogla

black                      swart                     swarta

bone                     bain                       baina

bread                    braud

break                    breken                 breketh                brak

bring                      bringen                bringeth               braahta

brother                broodar                broodra

cloud                     wolken                 wolkena

daughter             dohtar                  dohtra

day                         dag                         daga

dog                        hund                     hunda

drink                      drinken                drinketh               drank

ear                         aura                       auren

earth                     ertha

east                       aust

eat                         eten                      eteth                     aat

eye                        auga                      augen

father                   fadar                     fadra

food                      met

foot                       foot                       foeta

from                      fan

go                           gaan                      gaath                     ida

goose                    gans                       gensa

green                    groen                    groena

hand                      hand                      henda

have                      haven                   haveth                  haveda

head                      hauvid                  hauvda

hear                       heuren                 heureth                heureda

heart                     herta                     herten

house                   huus                      huuza

know                     witen                    wait                       wissa

language              spraaka                spraaken

light                       leoht

light                       liiht                         liihta

make                     maken                  maketh                makeda

month                  maanoth              maandha

moon                    maana

mother                 moodar                moodra

mouse                  muus                     myysa

mouth                  munth                  mundha

new                       neo                        newa

night                      naht                       nehta

north                     north

one                        ain

or                            oth

say                         sagen                    sageth                  sagda

see                         seen                      seeth                    sah

sister                     swestar                swestra

sky                         himil                      himla

son                         sun                         suna

south                    sunth

speak                    spreken                spreketh              sprak

star                        sterra                    sterren

sun                         sunna

take                       nimen                   nimeth                 nam

the                         sa                            dha

this/these           dhis                        dheza

three                     thrii

tongue                 tunga                    tungen

tree                       treo                       trewa

two                        twai

water                    watar

west                      west

white                    hwiit                      hwiita

with                       mid

wolf                       wulf                       wulva

word                     word                     worda

world                    werold                  werolda

year                       jaar                        jaara

young                   jung                       junga

There were at least four towns in Western Europe that were called Mediolanum in Antiquity. The most important of these is the Mediolanum in northwestern Italy that is now known as Milano in Italian, i.e. Milan. Other towns called Mediolanum were: Mediolanum Santonum (“Mediolanum of the Santones”), which is modern Saintes in western France; Mediolanum Aulercorum (“Mediolanum of the Aulerci”), which is modern Evreux in northern France; and a Mediolanum in what is now England on the site of Whitchurch in Shropshire.

The origin of the name Mediolanum is undoubtedly Celtic (all four towns called Mediolanum were in Celtic territories that were conquered by the Romans). The Celtic form of the name was most certainly Mediolanon but beyond this the etymology is considered uncertain. Two possible etymologies that are sometimes given are:

• *medjo– “middle” + *lanon “enclosure” – the second element being considered a cognate of Welsh llan, which can mean “yard, churchyard, enclosure”. The big problem with this is that the n was originally double, as can be seen in the plural llannau, as well as the Breton cognate lann (plural lannoù).  The Welsh llan and Breton lann actually come from *landā which is cognate with Germanic *landą “land”. That being said, I consider the possibility of a Celtic *lanon meaning “enclosure” quite doubtful.

• *medjo– “middle” + *lānon < *φlānon “plain” – the second element in this case being considered a cognate of Latin plānum. Although it is certainly not impossible that Celtic had an exact cognate of the Latin word with the same meaning, I consider it unlikely that Celtic had a word for “plain” which was virtually identical to their word for “full” – *φlānos (cf. Welsh llawn, Breton leun). I suppose that one might imagine that Mediolanum meant “Middle Full(ness)”, but I’m really not keen on that possibility myself. [Added on August 18, 2022] It has recently occurred to me that the decisive argument against relating *-lānon with Latin plānus is the fact that the cognate of Latin plānus (< *plh₂t-nos) is actually Proto-Celtic *φlitanos (< *pḷth₂-nos). See the entry for platno with etymological information in my Indo-European conlang called Woks Teuteka (https://vellaunos.ca/2021/08/02/woks-teuteko/).

The etymology for Mediolanon that I personally fancy is this:

• *medjolo- “middle, centre” + *-ānon – the first element in this etymology being a noun *medjolon derived from *medjos “mid-” which is cognate with Germanic *midilą (< *midja– + *-la-) “middle”. The second element is the neuter of an adjectival ending *-ānos making *medjolānos “central”. The name Mediolanon would quite simply mean “(The) Central (One)”.

The username that I chose for this blog – vellaunos – is a rendering of a name of a Celtic Deity, more usually rendered as Vellaunus. This name appears on two inscriptions, one in Britain (Wales) and the other in southern Gaul. In the first inscription, the compound name Ocelus Vellaunus appears as an alternate name of Mars Lenus. In the second inscription, the God Mercury is surnamed “(the) Victor Macniacus Vellaunus”.

The name also appears as part of the name of the Celtic Goddess Icovellauna; as part of the name of the British ruler Cassivellaunus (Welsh Caswallawn) who famously confronted Caesar in 54 BC; and as part of the name of the Catuvellauni tribe of Britain which may have been led by Cassivellaunus and was definitely led by the famous British king Cunobelinus (rendered as Cymbeline by Shakespeare).

The name of the God Vellaunus – *Wellaunos in Celtic – has been compared to the Old Irish follamnaigid ‘rule’ which is derived from *wollamn–agje/o– (the change of –mn– to –un– is usual in Celtic, and in Brittonic as well, but is not found in Gaelic). The name of the God would therefore mean “Ruler”. Here are some further etymological connections that I came up with many years ago:

Celtic *wellaunos < *wellamnos < *werl-ṃnos “ruler”; Germanic *weruldiz (Old Saxon werold, worold & Old English weorold, worold, woruld > English world) < *werḷ-tis “dominion” (not *weraldiz < *weraz + *aldiz i.e. “man-age”); Greek Hellēnes “Greeks” < *wellānes < *wellamn– < *werl-ṃn– “rulers”; Sanskrit Varuṇa < *werl-ṃnos “ruler” (Varuṇa being the original supreme God in the Indo-Aryan (Vedic) religion).

All these terms are derived from a Proto-Indo-European root *werl– which apparently meant “rule, command”. I have wondered if this *werl– was not somehow connected with the root *welh₁– “choose, want” which is reflected in words like Latin velle (volō) “wish, want” and English will (from Proto-Germanic *wiljô, *wiljaną) – “to rule/command” being essentially “to impose one’s will/wishes” on others. I might even suggest that the root *welh₁– was really *werl– (which usually became *well-)…

[NB: The name of the Greek God Ouranos is usually considered to be cognate with Sanskrit Varuṇa. But in fact, Ouranos is from *osuranos, a derivative of *onsuros which became asura in Sanskrit and ahura in Avestan (*onsuros in turn being a derivative of *onsus which became *ansuz in Germanic (Old English ōs “a God” found in names such as Oscar and Oswald, and Old Norse Áss “a God” (plural Æsir “The Gods”). See https://vellaunos.ca/2021/12/23/the-reflexes-of-proto-indo-european-h1ensus-h1onsus-and-h1onsuros/ ]

I was working on the demonstratives and articles for my Vindonian language one day. I eventually shifted to the personal pronouns and the possessives. As I thought about these, I decided it might be useful to make my own reconstructions of the Proto-Indo-European 1^st and 2^nd person plural pronouns and the plural demonstratives (i.e. 3^rd person plural pronouns). These are given in the following table along with their Proto-Celtic reflexes:

                                    PIE pronouns                                       PC pronouns

                                    nom                 acc                               nom                 acc

1pl incl                        hnes               hnṇs (hṇs)                 snīs                  ans

1pl excl                      wejes               wejṇs                           —                      —

2pl                               juhes                juhṇs                           swīs                 was

3pl                               hejes                hejṇs                            ijes                   ijas

(Of course, the h’s in the PIE forms stand for laryngeals, but without the usual subscript numerals. The 1pl incl forms are usually not reconstructed with an initial laryngeal, but I think the accusative form should require it. The laryngeal in the 2pl forms are apparently undefined. But the larygneal in the 3pl forms was most probably h₁.)

Perhaps I should explain the changes between PIE and PC in nominative forms of the 1^st and 2^nd persons:

1pl          hnes      >             nes         >             nēs         >             nīs          >             snīs

2pl          juhes     >             (i)wes    >             wēs        >             wīs         >             swīs

The lengthening of the vowels is paralleled in Latin (nōs, vōs). By the way, the addition of the initial s in both forms is unique to the Celtic languages.

Although the accusative forms of the PC pronouns aren’t found in Celtoid, the Old Irish possessives ar ‘our’ and for ‘your’ come from *ans-ros and *was-ros which are derived from the accusative forms of the personal pronouns (as also in Latin and in Germanic).

I should also point out that I believe that there was originally clusivity in the 1^st person plural pronouns in PIE but that this was eventually lost somewhere along the way. This inclusive/exclusive distinction in 1^st person plural pronouns is found in many languages in several language groups. Clusivity in PIE would obviously account for the two morphologically irreconcilable 1^st person plural forms shown in the table above. I might speculate that the exclusive form (*wejes) is related to the *wi– prefix which denoted separation/apartness in PIE.

One of the more idiotic etymologies regularly given by linguistic experts is the one for the usual Celtoid word for “low”. This word appears as izel in Breton, as isel in Welsh and as ísel in Old Irish. (A Celtic (i.e. “Gaulish”) cognate has apparently not yet been found, but it would most certainly have been *īsselos/*ītselos.)

The immediate origin for these words is obviously Proto-Celtic *ītselos. But beyond this, the linguistic experts imagine that this Proto-Celtic *ītselos is derived from the Proto-Indo-European locative plural of the word for “foot” – *pōts – which would have been *petsu and which meant “at the feet” (in which case the Proto-Celtic would presumably have been *φītselos).

This etymology is not entirely devoid of sense, but I consider it quite ridiculous. It’s really just silly, not to mention phonetically impossible – there is no way that a long i can result from a reflex of *petsu.

The important cognate of Proto-Celtic *ītselos that the linguistic experts routinely disregard is Latin īnferus “lower” (the source of English inferior). Latin īnferus is always considered to be a cognate of Proto-Germanic *underaz (> English under) and Sanskrit adhara, but these terms are from an original *h₁ṇdher-, a form that cannot have produced the long ī in the Latin word, nor can it have produced the medial f. The exact reflex of PIE *h₁ṇdher– in Latin would have been **enderus.

[By the way, the Celtic language (usually called “Gaulish”) had an exact cognate of Proto-Germanic *underaz and Sanskrit adhara, this appearing as anderon on the inscribed lead tablet found at Chamalières. A related Celtic andernados appears on the inscribed lead tablet found at Larzac. Celtic anderos possibly meant “under” or “nether”, although I consider it equally possible that it meant “original” or “primeval”…]

The existence of f in the middle of a non-compound Latin word is very peculiar, and it has had a few people scratching their heads. The f sound that appears at the beginning of many Latin words usually comes from original bh and dh, but these voiced aspirates normally became b and d in the middle of Latin words. For this reason, a hypothetical Pre-Proto-Italic *īndheros would normally have become **īnderus in Latin rather than īnferus. So there was something special in the proto-form of this word that caused the f sound to develop in medial position.

In my view, this peculiar medial f sound is the result of a dh being immediately followed by an s, as in a possible Pre-Proto-Italic *īndh-s-eros. In my opinion, the following s influenced the development of f from dh in a medial position. And this Pre-Proto-Italic *īndh-s-eros would nearly correspond to a possible Pre-Proto-Celtic *īndh-s-elos which would have become Proto-Celtic *īntselos, and which would have lost the n at some point for some reason, yielding Proto-Celtic *ītselos.

Of course, the *īndh– element in the possible Pre-Proto-Celtic *īndh-s-elos and Pre-Proto-Italic *īndh-s-eros would be related to the *h₁ṇdh– element in the *h₁ṇdher– that became English under and Sanskrit adhara. But the existence of the long i in the former must necessarily indicate two different forms.

It is usual to find Proto-Celtic reconstructions in Wiktionary that assume that PIE [ei] became [ē] in Proto-Celtic. I don’t know who came up with this (Matasović?) or why, but I must strongly disagree.

There is evidence that [ei] became [ē] in Gaulish (probably late Gaulish), an apparent instance of this being deuo– ‘god’, i.e. dēuo– from *deiwos. But I hope this isn’t the basis for assuming that [ei] had already become [ē] in Proto-Celtic. This would be like saying that Latin must have had the [ø] sound because French has it, which would obviously be absurd.

It may also be that [ei] became [ē] at some stage in Proto-Goidelic. This [ē] would have shifted to [ea] and then to [ia] as is seen in Old Irish (written ía): e.g. OIr íasc ‘fish’ from *eiskos (< *φeiskos < *peiskos). But note that the [ē] (written é) remained in contexts where i-affection inhibited the shift: e.g. OIr éisc, genitive singular and nominative plural of íasc, both from *eiskī; also OIr léicid ‘to leave’ from *leikwī– (not *linkwī-; cf. PGmc *līhwaną < *leikw-). The [ē] in Proto-Goidelic obviously came from [ei] but this development occurred in Proto-Goidelic, not in Proto-Celtic.

The Brittonic languages give strong evidence that [ei] did not in fact become [ē] in Proto-Celtic. The reflex of PIE [ei] in Brittonic was [ui] as represented in Welsh by wy and in Breton by oue. It is quite unlikely that [ui] could have developed from [ē]. Much more likely is the following transition: Proto-Celtic [ei] > [oi] > [ui].

By the way, this resembles the development of Old French [ei] into Modern French [wa] : [ei] > [oi] > [oe] > [we] > [wa] with a shift from falling to rising during the [oe] stage (although this development occurred later than the aforementioned Brittonic  development).

I might also mention just for interest’s sake that PIE [ejV] (where V stands for any vowel) became [ijV] in Proto-Celtic. Of course, this does not exclude the possibility that PIE [ei] became Proto-Celtic [ē] (but it didn’t happen).

For what it’s worth, my Vindonian language has the following development of Proto-Celtic [ei]: [ei] > [ē] > [ie] with a shift from falling to rising during the [ē] stage. The Vindonian word for “food” is bied which may be compared with Welsh bwyd and Breton boued, all of these being from Proto-Celtic *beiton (not *bēton).