The horse goes before the cart (updated)

Germanic *xursa-/*xrussa- 'horse' has been linked by some Indo-Europeanists to Latin currō 'to run'¹. This verb is derived from a Paleo-European lexeme *k^ər^əs- also reflected in Celtic *karro- and Latin currus 'cart'² < *k^ər^əs-o-. However, in chronological terms, the traditional view of deriving 'horse' from a verb 'to run' would be like putting the cart before the horse. That is, the meaning 'horse' has to be older than 'run' and not the other way around.  

In my opinion, the Paleo-European word would be related to East Caucasian *ʁHwo:r[tʃ^ʔ]o (˜ -tɕ^ʔ-,-ə) 'deer, game', probably dating from the Upper Paleolithic. Possibly  also Yeniseian *kuʔs 'horse' belongs here, as the names of domesticated animals are often Wanderwörter.

The Caucasian word is also the source of Kurganic IE *jorko-³ (Celtic *jorko- 'roe deer (Capreolus capreolus)', Greek zórks, dórkas 'gazelle'), with metathesis. 

I also suspect this is the origin of the autochthnous name of the Pyrenean chamois (Rupicapra pyrenaica)⁴: Aragonese chizard(o), ixarzo, Gascon isar(t), idart, Catalan isard < *i-tsardV, with regular merger of sibilants /ts/ and /s/ in Gascon and Catalan and a prefix *i-, probably a fossilized article. There's also the unprefixed western variant *tsarri > Gascon sarri, Aragonés sarrio, the latter borrowed into Basque⁵.

_______________________________________________________________________________

¹ G. Kroonen (2013): Etymological Dictionary of Proto-Germanic, pp. 260-261.

² Latin carrus is a Celtic (Gaulish) loanword.

³ Caucasian ʁ ~ Kurganic IE k and Caucasian tʃ ~ Kurganic IE j.

⁴ A species of mountain goat which lives in the Pyrenees, Cantabrian Mountains and the Apeninnes, considered to be a subspecies of the chamois (Rupicapra rupicapra) by some specialists.

⁵ The lack of a native Basque word for the animal can be explained from its absence from the Basque-speaking area.

Gaulish *agranio- 'sloe, fruit of the blackthorn' (updated)

Sloes

Occitan agranhon, Aragonese arañón, Catalan aranyó 'sloe, fruit of the blackthorn (Prunus spinosa)' < Gallo-Latin *agraniōne-, is a diminutive form borrowed from Gaulish *agranio-. The Gaulish word is also the source of Basque ar(h)an, which means 'sloe' in Biscayan¹ but 'plum' in other dialects, as well as the collective form arantza (B, G), arantze (L), arhantza (LN) 'thorn(bush)'.

A bottle of home-made patxaran

A popular liquor made from sloes is called patxaran (Spanish pacharán) from the High Navarrese name of the fruit, which is a variant of ba(i)saran (G), baixaran (S), maisaran (R), basahan (Z), paxaran (HN, Aezk, R), a compound from basa- 'wild' and aran.

The corresponding form in Insular Celtic is *agriniā (Old Irish áirne, Welsh eirin), although phonetic differences make impossible to reconstruct a single Celtic protoform. IE cognates are Germanic *akarna-/*akrana-/*akirna- 'acorn'², Latin agresta 'green grape', agrestis and Greek ágrios 'wild', which have nothing to do with 'field' (Latin ager, Greek agrós) in despite of their resemblance.

Oak acorns

 
The scarce attestation within IE and the semantic latitude point to a Paleo-European substrate loanword which I'd link (with loss of the initial nasal) to Uralic *nakrV 'cedar nut'³, and possibly also to Altaic *ɲíkrV 'a k. of thorny tree', Eskimo *(Nǝ)k[r]uʁa- (~ *Nǝkǝvʁa-) 'tree, spruce'.

Cedar nuts (with and without shell)

_______________________________________________________
¹ The Gipuzkoan form is txaran. In these Westernmost dialects, 'plum' is okaran, apparently from an anaptyctic form *akarann.

² Pokorný conflated the Celtic and Germanic words with Lithuanian úoga 'berry', OCS (j)agoda 'fruit', vin-jaga 'grape', which are better linked to Latin ūva 'grape'.

³ The fruit of the Siberian pine (Pinus sibirica) or Swiss pine (Pinus cembra).

Basque gorri 'red'

According to the Vascologist Joseba Lakarra, Basque gorri 'red' derives from a lexeme *gorr- 'raw meat'. However, the isolationist paradigm of academic Vascology precludes further investigation of cognates in other language families. 

Considering the voiced initial stop in Basque was originally voiceless and also the trill rhotic isn't the product of gemination, the most likely candidates would be Altaic *kʰjú:ru ‘red, reddish; brown, dark’ (EDAL 1090) and IE *kreuH₂- ‘blood, gore’ (Greek kréas 'meat', Lithuanian kraûjas 'blood'), thus pointing to a lexeme *kʰruH-¹ ~ *kʰuHr- which would date back to the Upper Palaeolithic and underwent a semantic shift from 'raw meat; blood' to 'red' in Altaic and Basque.

One interesting derivative of gorri is gorringo², which refers either to a kind of mushroom (Amanita caesaria) or the egg's yolk.

______________________________________________________________

¹ IE *e is the ablaut vowel and doesn't play any role in the actual etymology.

² With the variants gorrinko in the Easternmost dialects (Roncalese, Zuberoan) and korrinko in the extinct Alavese.

Galician-Portuguese silva 'bramble' (updated)

Blackberries

Contrarily to what some linguists think, Galician-Portuguese silva, silveira (collective) 'bramble (Rubus)' is unrelated to the homonymous Latin silva 'forest' (which regularly gives Romance selva), but it's cognate to Leonese silba 'service berry', silbar, silbal (collective) 'service tree (Sorbus domestica)'².

Service berries

This is a substrate loanword *silba¹ with parallels (through lambdacism) in Romance serba 'service berry'³ (Catalan serva, Occitan sèrba, Spanish serba, jerba, sierba id., serbal, sierbal, Catalan servera, server (collective) 'service tree'⁴), Lithuanian serbentà 'currant (Ribes)' and dialectal Russian serbalína, serberína 'rose hip', sor(o)balína 'bramble'⁵, Latin sorbus 'service tree' (Spanish sorbo), sorbum 'service berry', from which derive Leonese (Liébana) suerba, French sorbe id., sorbier, Galician sorbeira, solveira (collective), Leonese  (Liébana) suerbal  'service tree'.

Although these forms show the typical IE ablaut e ~ o⁶, we also can find variants with /u/ vocalism: Leonese (Liébana) surba 'service berry', surbu (collective) 'service tree' and regional Spanish (Álava, Bureba, High Rioja) zurba, zurbia 'service berry', zurbal, zurbial, surbial (collective) 'service tree'. 
________________________________________________________ 

¹ The shift -lb- > -lv- is regular in Portuguese.
² J. Oria de Rueda et al. (2006): Botánica forestal del género Sorbus en España, in Investigación Agraria. Sistema y recursos forestales, vol. 15, nº 1,  p. 166-186.  

³ The proposed connection (Pedersen) with Celtic *swerwo- 'bitter' (Old Irish serb, Middle Welsh chwerw) can be ruled out. 
⁴ The forms serbo, jerbo, selbo, jelbo are the product of a contamination with sorbo.
⁵ M. Vasmer (1955): Russisches Etymologisches Wörterbuch, p. 697.

⁶ In fact, some Indo-Europeanists reconstruct a protoform *serbh- ~ *sorbh-.

Paleo-European *abVl- 'apple' (updated)

Several IE languages of North Europe (Celtic, Germanic, Balto-Slavic) reflect a lexeme *abVl- 'apple' which is regarded as a Paleo-European substrate loanword by some specialists¹. In my opinion, a likely cognate would be Hittite sam(a)lu- 'apple (tree)', with denasalization of m and loss of the initial sibilant², which in this and other words such as sākuwa- 'eye' < IE *H₃ekʷ- 'to see' and sankuwāi- 'nail; a unit of linear measure' < IE *H₃n(o)gh- 'nail' would reflect a labialized voiced pharyngeal fricative *ʕʷ (IE *H₃) .

Therefore, I'd reconstruct a Nostratic³ Wanderwort *ʕu-malV also reflected in Basque udare, udari, madari 'pear'⁴ (with denasalization and further delabialization), and which I'd link to Nakh-Daghestanian *mhalV- ~ *mhanV- 'warm', with a straightforward semantic drift from 'warm (season)' to 'fruit'. This makes sense because the apple tree is originary of Kurdistan, precisely in the area where Nostratic was presumably spoken.
______________________________________
¹ For example, Theo Vennemann links it to Afrasian *ʔa-bul- 'male genitals', which (in his own words) is "semantically unsatisfactory although phonetically perfect". See T. Vennemann (1998): Andromeda and the Apples of the Hesperides, in Europa Vasconica, Europa Semitica, p. 591-652.
² Explained by ortodox IE-ists such as Kloekhorst as the result of a "s-mobile".

³ My own concept of "Nostratic" isn't the one of a very large macro-family including IE, Uralic, Altaic, Kartvelian, etc. but a language spoken in the Taurus-Zagros mountains, where several species of plants and animals were domesticated in the Neolithic.
⁴ A variant *ʕu-manV would be the origin of Basque umao (B), umo 'ripe, seasoned' and Uralic *omena ~ *omVrV 'apple'.

Tartessian Aiburis (updated)

Tartessian aibᵘuris /Aiburis/ appears at the beginning of a funerary inscription (J.3.1) and is frequently interpreted as an anthroponym whose second element -ris would correspond to Celtic *-rīxs 'king' > -rix, which in fact is the most frequent element in Gaulish anthroponymy¹. However, the first element aibu- (likely reflected in Iberian aibe), doesn't have a clear Celtic etymology, although it appears in the Gallaecian theonym Aebosocelensis² and possibly also the toponym Aipora (modern Évora)³.

Prósper proposes a link to Greek aipús 'steep, craggy', aípos 'hill, crag', whose etymology is disputed⁴. If the labial stop comes from a labiovelar, we could compare it to the Italic toponyms Aequāna, Aequum Faliscum and Aequimelium (a village on the hillside of the Capitoline), the latter being a calque of the Gallaecian form. Although homonymous to Latin aequus 'equal; fair, even' < *aikʷo-, we can hardly expect Aequimelium to mean 'confraternity hill' or the like.

On the other hand, we've got Gallic (Gaulish) *aikʷo-randā 'boundary' as the source of the toponyms Aigurande, Eguérande, Eygurande, Iguerande, Ingrande(s), Yvrande(s), La Guirande, Eurande, etc⁵. The first element *ekwo- has been alternatively linked to Latin aqua 'water' (absent from Celtic) or aequus, but in my humble opinion it would make more sense to be a foreign loanword related to the Italic toponyms. 

_________________________________________

¹ X. Delamarre (2008): Dictionnaire de la langue gauloise, 2nd edition, pp. 259-260.
² B.Mª. Prósper (2000): Lenguas y religiones prerromanas del occidente de la Península Ibérica, p. 112.

³ J.T. Koch (2013): Celtic in the South-west at the Dawn of History, 2nd edition, p. 141.
⁴ P. Chantraine (1968): Dictionnaire étymologique de la langue grecque, p. 37-38.

⁵ X. Delamarre, op. cit., pp. 163-164.

Basque (k)abia 'nest'

The Basque word for 'nest' appears with variants such as abia (B), habia (Z), kabia (G), kafia (G, HN), abira (S), abi (B), kabi (G, HN), api (G, HN), aapi (G). For this word, Mitxelena -the founder of modern Vascology- proposed more than 50 years ago an etymology from Latin cavea 'cage', which turns to be semantically inadequate.

 

By contrast, I propose a loanword from Celtic *āwjā 'brood; nest', a femenine form corresponding to the masculine *āwjo- 'egg' (Old Welsh ui, Middle Welsh wy, Middle Breton and Old Cornish uy)¹. Therefore, both k- and -r- in some of the Basque forms would be epentic.

On the other hand, the forms abe (B) 'honeycomb', kabe (*L) 'nest; beehive' are the result of a semantic contamination with derivatives of Romance *kofanu 'deep basket' (Latin cophinus) such as Biscayan abao, abau 'honeycomb' and Bearnese càben 'beehive'. The semantic drift can be explained because artificial beehives in the shape of a deep basket are used in some places.

_____________________________________________________________________
¹ R. Matasović (2008): Etymological Dictionary of Proto-Celtic, p. 50.

Etruscan viscri 'penis' (updated)

Italian (Tuscan) bischero 'peg (on a string instrument); fool, idiot' derives from Etruscan viscri 'penis', attested in a votive inscription: eti viscri ture Arnθ Alitle Pumpuś (TLE 685) '... gives this penis', engraved on a figure of a man offering his male organ found at Paterno di Vallombrosa (Arezzo). 

Although many specialists have linked the Etruscan word to Latin vīscus, vīsceris 'entrails' (itself a word of problematic etymology), I think the latter would correspond to Etruscan visc 'center' (see here). In my opinion, viscri would be an Italic loanword akin to Latin virga 'twig, rod' (with rhotacism) < *wizg-ā, from a lexeme *wizg- also attested in Germanic *wiskō 'bundle, besom'. The semantic shift 'rod' > 'penis' also happened in Spanish verga < Latin virga.

Basque ol(h)a 'forge, foundry'

Basque ol(h)a 'forge, foundry; workshop' is a word conflated with its homonymous 'hut; dwelling' (geographically restricted to a few dialects), a confusion also extending to the widespread toponymic suffix -ola¹. However, upon a closer scrutiny they reveal themselves as two different words. Firstly, Roncalese õla 'hut' has a nasal vowel, a remnant of a former initial nasal consonant. And secondly, the respective derivated verbs are different, as on the one hand we've got ol(h)atu (LN, Z) 'to hit with violence', and on the other, olhatü (Z) 'to stay on huts; to go to the pastures (in transhumance)'.

I think the Basque word would be a loanword from Celtic *ordā (f.) 'hammer' (Middle Welsh orth, Breton orz)², possibly through retroflexion of the dental stop (as in Sweden and Norwegian), which ultimately evolved into a lateral. However, as no plausible IE etymology can be posited for the Celtic word, a non-IE origin is likely. A likely cognate would be Etruscan urθan 'to make, to manufacture', attested in the past form urθanice '(he) made'. 

This way, we're left with a Tyrrhenian verbal root *ur-d- which I'd link on the one hand to IE *wer-g´- 'to work'³ (zero Ablaut), where IE *-g´- ~ Tyrrhenian *-d-, and on the other to Urartian ur-/or- 'to make, to work'⁴, which Diakonoff-Starostin⁴ relate in turn to Nakh-Daghestanian *=ahwV(r) 'to do' (NCED 1826).

___________________________________________________________________

¹ Compare for example Loiola 'pottery', from lohi 'mud', and Gisasola 'broom bush', from (g)isats 'broom'.
² There's also a masculine form *ordo- (Old Irish ord, Gaulish Ordo-). See R. Matasović (2009): Etymological Dictionary of Proto-Celtic, p. 300.
³ Incidentally, the amateur linguist Gianfranco Forni (who regards Etruscan as an IE language) derives urθan from an IE participle *H₃er-to- 'risen' < *H₃er- 'to rise' (Latin orior). However, to me this is not only semantically questionable but also phonetically inconsistent, as it wouldn't explain the voiced stop in Celtic, of whose connection with the Etruscan verb he's of course unaware.
⁴ N. Diakonoff & S. Starostin (1986): Hurro-Urartian as an East Caucasian Language, §163.