THE « CELTS »: QUESTIONS OF NOMENCLATURE AND IDENTITY

This paper counters doubts raised recently about the validity of the term « Celtic » as a linguistically oriented ethnonym with evidence that the ancient continental peoples so designated in classical sources did indeed call themselves Keltoi and with an etymology of their hitherto problematical name as a formation most unlikely to have been created after the Proto-Celtic period itself. Various attested designations of speakers of closely related « Celtic » languages in Ireland and Britain are then considered. Finally, a brief look at the modern revival of the term after centuries in abeyance leads to the conclusion that it remains valid, linguistically at least, despite various questionable uses to which it has been put since its reintroduction into academic and popular discourse.

in the later first century B.C. by Trogus Pompeius, the Romanised grandson of a Gaul, that young Gaulish males were widely employed in the third century B.C. as mercenaries by Hellenistic Greek monarchs in the Near East.
If the Gauls' initial impact on the Mediterranean world was primarily a military one typically involving fierce young *galatīs (the plural of *galatis, inferred above) hungry for land or employment, it would have been natural for the Greeks to apply this name for the type of *Keltoi that they usually encountered to the nation as a whole as Galátai. Although Celts in the East may eventually have begun to use it of themselves, their compatriots in the West still clung to *Keltoi on the good evidence of Caesar above. The problem of Latin Gallus can be solved by positing an Etruscan intermediary. The first major people of Italy to encounter Celtic invaders, the Etruscans spoke a language that lacked a phonemic distinction between voiced and voiceless stops but had developed intervocalic voiced allophones before undergoing extensive syncope around 500 B.C. (Rix, 2004 : 547-50). Accordingly *Galatis could easily have been adapted into Etruscan as *Kalade and then syncopated to *Kalde. In view of admittedly sporadic correlations such as Lat. gubernare < Gk. kubernân « steer, guide » or Lat. gladius « sword » < Gaul. *kladi-« sword » (corresponding to OIr. claideb, MW cledyf), this could have been borrowed into Early Latin as *Galdos and would then (cf. Lat. Pollux, Polluces < *Poldouces < Gk. Polydeukes) have developed regularly into classical Gallus (McCone, 2006 : 104-7). If so, Caesar was essentially right about Galli being a Latin equivalent of native Celtae.
It seems, then, that Keltoi was the name once used by virtually all continental peoples known, on good inscriptional and/or onomastic evidence, to have spoken what are now called Celtic languages in an arc stretching from Iberia in the West to Scythia in the East, and that Etruscans and Greeks were responsible for making an ethnonym of native *galatīs originally referring to warlike young *Keltoi.
The vocalism of a vr̥ ddhi-derivative in relation to its base was simply [+ length] in the case of a vowel (ē/e and ō/o) and [+ e] otherwise (e.g. er/r̥ , el/l̥ , en/n̥ and em/m̥ ), but this relationship will have been greatly complicated (to ī/e, ā/o, er/ri, el/li, en/an, em/am, etc.) by well-known Proto-Celtic sound changes (McCone, 1996 : 49-51, 59-60). Consequently vr̥ ddhi-derivation can hardly have remained viable beyond the Proto-Celtic period and such formations are correspondingly uncommon in Celtic languages. It follows from the requirement of *kl̥ tos rather than *klitos as its original base that *keltos was formed in the Proto-Celtic period and that speakers of Proto-Celtic actually called themselves *Keltoi « Celts » as descendants of the god of the underworld.
Far from being a Greek term for a vast ethnically and linguistically diverse swathe of people as claimed by Chapman (1992 : 30-2), Keltoi was the native ethnonym of a linguistically rather homogeneous populations that had spread quite rapidly over much of continental Europe by the third century B.C. Earlier Greek and Roman failure to distinguish them clearly from other northern peoples was soon rectified by the likes of Caesar and Tacitus, once the Romans had made closer contact with Germani.

NAMES OF CELTOPHONE POPULATIONS IN EARLY BRITAIN AND IRELAND
It must be admitted that classical authors never call the inhabitants of Britain and Ireland Keltoí/Celtae or, for that matter, Galátai/Galli. The plural Brettanikaí Nêsoi « British Isles » is used in the second century B.C. by Polybius (3.57.3), whom Strabo (2.4.1) states to have been referring to an earlier work by Pythias, a Massiliote who claimed to have visited Brettanikḗ « Britain ». These look like derivatives of the ethnonym Brettanoí, which happens not to be directly attested until Diodorus (5.22). Shortly before him, Caesar in his de Bello Gallico had called the largest island Britannia (5.12-3), its inhabitants Britanni (4.21, 5.14), and the second largest island Hibernia (5.13) without naming its people, whom Diodorus (5.32) simply calls « Brettanoí inhabiting Íris [Ireland] ».
Middle Welsh Prydein points to native *Pritanyā (Lat. Britannia), while Middle Welsh Prydyn « Picts » and Old Irish Cruithin indicate the presence on both islands of people called *Pritenoi (Gk. Brettanoí, Lat. Britanni) or *K w ritenoi, which was either the original form before British k w > p or the result of Irish substitution of k w for a p lacking in the language until the fifth century A.D. (McCone, 1996 : 69-70, 92; cf. OIr. cásc « Easter » < *k w askā for Lat. pasc[h]a). An extension of *k w rito-« cut off », presumably from the Continent by the sea, would be a plausible Celtic etymology but a pre-Celtic origin cannot be excluded. Either way, Celtic speaking immigrants to Britain seem to have received a name connected with their new island home by a relatively early stage, presumably in place of some earlier designation that could well have been *Keltoi. Subsequent restriction of Prydyn to inhabitants of the Scottish Highlands may have been due to displacement by new designations for the Welsh, notably native Cymry (< *kombrogī « sharers of territory »; MW bro « land, region »; cf. the Gaulish Allo-broges) and English Wealh (basically « foreigner »). Recent settlement by Belgic tribes, apparently speakers of a form of Gaulish, in the south of Britain was contrasted by Caesar with much earlier occupation of the interior (de Bello Gallico, 5.12-3). This is the area referred to in Tacitus' casual mention of the similarity of British and Gaulish speech at the end of a survey of Britain's various inhabitants based chiefly upon dubious physical and geographical considerations (Agricola, 11). Meyer (1910 : xv) had « no doubt that the bands of Scotti who made common cause with the Picts in the third and fourth centuries in harassing Roman Britain were also called fíana ». Old Irish fían(n) < *wēnnā < *wēd-nā (McCone, 2012 : 20-1) effectively meant « wild bunch » as a derivative of *wēd-u-(> OIr. fíad « (wild) game », OBret. guoid, MW guyd « wild »; cf. *wid-u-, underlying OIr. fid « (wild) wood », O/MW guid/gwyd « trees »; Matasović, 2009 : 408, 420). The Proto-Celtic term for a band of *galatīs seems to have been *koryos (OIr. cuire « band », Gaul. Tri-/Petru-corii) directly inherited from Proto-Indo-European, while fían(n) has no cognates outside Irish. The latter looks like a rather late prehistoric Irish replacement of the former (McCone, 2012 : 20-1) and so Meyer's surmise above seems unlikely to be strictly correct.
The Irish almost certainly borrowed their historic name Goídel « Gael » (pl. Goídil) from British *gwoid-elo-(OW Guoidel « Irishman », MW pl. Gwydyl) < *wēd-elo-referring to the type of Scotti mentioned by Ammianus Marcellinus (20.1.1, 26.4.5, 27.8.5, etc.) as wild raiders upon Roman Britain in the latter half of the fourth century A.D. If so, it can only have entered Irish after long ē had been diphthongised to oi/ui in British in the course of the seventh century A.D. (Jackson, 1953 : 330-5). An older designation may survive in the term Féni < (i-stem) *wēnīs or (yo-stem) *wēniyī, which often refers to free Irishmen in general in legal texts (Kelly & Charles-Edwards, 1983 : 133-4) and has a Welsh cognate in Gwynedd (< *Wēniyā) along with [V]ENEDOTIS (< *Wēniyātis), the direct precursor of the later name Gwyndot for an inhabitant of the area (Jackson, 1953 : 188, n. 1;551, n. 3), on a British Latin inscription dated to the end of the fifth century A.D. Since there had been early Irish settlement in the northwestern part of Wales so designated (Dillon & Chadwick, 1973 : 60-1), it seems likely that the name *Wēniyā was created in the fifth century A.D. and originally meant « [land] belonging to the *wēnīs» (cf. Galatia and galatīs above). As a probable derivative of a PIE root *ṷeḭH « strive after, pursue » underlying the rare MW verb gwyn-« plunder » (< *wi-na-; Rix et al., 1998 : 609-10;cf. Matasović, 2009 : 412-3), *wē-ni(-yo)-would have had an original sense « raider », or the like, probably as an Irish word used to name an occupied part of Wales but possibly a British one adopted by the Irish a couple of centuries before *gwoidel. Either way, the relationship between *wēnīs or *wēdelī (probably Irish and British terms respectively) and *Skotī is reminiscent of that between *galatīs and *Keltoi at an earlier stage.
Scot(t)i is apparently first recorded as a name for the Irish by Ammianus (above), being also used by Saint Patrick in the fifth century and quite widely in Latin thereafter. Unlike Picti « the painted ones », it has no obvious Latin source and so seems likely to have been borrowed from Irish, where it apparently became obsolete not long after. The Irish ā-stem noun scoth « flower, pick » provides the key on the reasonable assumption that in Ireland a Celtic speaking upper class became known as *skotās « flowers » or a masculine equivalent *skotī as the « pick » of the population, this term then being borrowed into Latin.
Scot(t)i sometimes appears with Aticotti, who also seem to have come from Ireland and appear on no less than ten occasions as auxiliary troops serving in the Western Roman Empire in the late fourth-century Notitia Dignitatum (Freeman, 2002). Since it need not originally have been identical with Old Irish aithech « vassal » (Vendryes, 1959 : 54-5), the first element of Medieval Irish aithechthúatha « subject peoples » may be reconstructed as *atiko-and taken as an adjective formed by suffixing *-ko-to the well-attested Celtic prepositional element *ati. Its sense « ex-, former », as applied to people, could easily become negative as in the case of the apparently Insular Celtic *ati-wiros « ex-man » (> MIr. aith(f)er, MW adwr « coward »). The adjective *ati-ko-meaning something like « ex-, demoted » would suit peoples deprived of their full rights by subjugation to a Celtic speaking élite of *skotās. Groups of such subjects and hence the category as a whole could then have been denoted by an early compound *Atiko-tōtās contrasted with superior *Skotās, thereby triggering a shift in the application of the term *K w ritenoi in Ireland. Be that as it may, Primitive Irish *Átiko-tōtās would presumably have been borrowed into Late Latin as *Aticótoti parallel to Scoti, and then become Aticotti as a result of an observable Vulgar Latin tendency to syncopate post-tonic vowels (Grandgent, 1907 : 99-102).
The status of Scoti and Aticotti as an Irish ruling élite and its subjects respectively in the fourth century A.D. would account not only for their twinning on Irish expeditions but also for the role of Aticotti as mercenaries in the Roman army, since paid service abroad would appeal most to the underprivileged (cf., for instance, Polybius, 1.68.7, on the significant proportion of slaves and deserters among Carthage's mercenaries at the end of the First Punic War). Scot(t)i seems to have acquired a geminate t under the influence of Aticotti before the latter fell out of use after the fourth century A.D.

THE MEDIEVAL AND EARLY MODERN PERIOD
Medieval speakers of what are now called Celtic languages in Ireland and Britain plus Brittany had no reason to regard the ancient Celts or Gauls on the Continent as close relatives. Latin Gallus was borrowed into Irish as Gall but connotations of barbarism rather than ethnic affinity soon made it a term for wild Viking intruders and then, after the Norman invasion, for any non-Gaelic settler from outside. « Celt » simply dropped out of use because the continental peoples once so named had been culturally as well as linguistically assimilated to Latin, Greek or Germanic environments in the Roman imperial period or its immediate aftermath and because Celtae was much rarer than Galli in Latin sources, although Keltoí or Kéltai was on a more equal footing with Galátai in Greek ones.
A further lack of any sense of particular ethnic affinity between the medieval Irish and British themselves is confirmed by origin legends first recorded in a Latin history of the Britons apparently compiled in the ninth century A.D. by a certain Nennius (Morris, 1980 : 19-20, 60-2), based as they are upon divergent choices between the two main starting points for such accounts in the Christian West at the time (Juaristi, 2000 : 99-109, 118-21). Having lost most of the Roman province once inhabited by them to the incoming Anglo-Saxons, the Britons understandably invoked an eponymous ancestor, Brito, and linked him with the prestigious Roman myth of Aeneas. He was later replaced by Brutus with a famous Roman name probably extrapolated from Isidore of Seville's unflattering suggestion that the British (Britones) were so called eo quod bruti sunt (« because they are stupid »; Etymologiae, 9.2.102) by reinterpreting it as « because they are of Brutus ». The Irish turned to the no less prestigious biblical account of Israel's exodus from Egypt by introducing a virtually eponymous Scythian, or Scythus, married to a Scotta, and an intermediate Spanish staging post doubtless suggested by the influential Isidore's view that Ireland owed its name (Hibernia) to the fact that « its nearer parts look towards Iberia [Hiberia] » (Etymologiae, 14.6.6).
In the early eighth century A.D. the Northumbrian monk Bede (Historia ecclesiastica gentis Anglorum, 1.1) divided Britain between the four linguistically distinct ethnic groups of English (Anglorum), British (Brettonum), Irish (Scottorum) and Picts (Pictorum), all sharing the ecclesiastical culture of « Latins » (Latinorum). The Picts faded from the scene in the following centuries as their now virtually unknown language(s) gave way to Irish and Norse in the Highlands, but the other three ethno-linguistic identities endured.
In Bede's day « British » could be applied straightforwardly to the language and inhabitants of Brittany, Southwestern Britain, Wales and Cumbria plus Strathclyde, where it had apparently been replaced by English or (in Galloway) Gaelic by the twelfth century. Around the end of that century it was still clear to Giraldus Cambrensis (Descriptio Cambriae, 1.6) that « in Cornwall and Brittany they speak almost the same language as in Wales […] but probably closer to the original British speech ». Despite increasing divergence thereafter, an awareness of these shared origins persisted in learned circles at least.
Although the charters of Scotland's king William the Lyon (1165-1214) still referred to « faithful subjects, French, English, Scots, Welsh and Gallovidian » (Smout, 1972 : 31), the term « Scot » had acquired its typical modern sense by the thirteenth century (Barrow, 1981 : 153). Soon afterwards Robert the Bruce could ignore his Norman ancestry and baldly assert in a short letter to the Irish that « our people and your people […] share the same national ancestry » (Barrow, 1988 : 314). Once his throne was secure, the so-called « Declaration of Arbroath », seeking papal recognition of Scottish independence from England in 1320, made the unprecedented claim that the Scotti had conquered Northern Britain directly from Spain, without reference to Ireland (Duncan, 1970).
There thus arose a difference of opinion as to which island was first settled by the Scotti. The Irish, of course, stuck to the original version enshrined in their Leabhar Gabhála or « Book of Invasions » (Carey, 1993) and the hardening of lines between Protestantism and Catholicism in the sixteenth century lent it added significance, to judge from the statement that « nine out of ten in the island are Catholic » and « say that Kingdom belongs to Your Majesty because it was originally Spanish » in the report of a visit to Ireland presented by a Basque seacaptain to Philip II of Spain in 1574 (Valdés Miyares & Tazón Salces, 1997 : 219). An authoritative revision of Leabhar Gabhála was completed in 1631 and it was a major foundation for Geoffrey Keating's history of Ireland, Foras Feasa ar Éirinn, which quickly acquired classic status after its appearance in 1635 (Cunningham, 2000 : 218).

THE REVIVAL OF « CELTIC » AND « CELT »
« Celt » was still available to be picked up from classical sources, especially from the fifteenth century onwards as western scholars rediscovered ancient Greek literature. The Italian Annius of Viterbo gave Japhet a great-grandson called Celtes in 1497 (Piggott, 1975 : 124) and the sixteenth century saw the beginnings of a French interest in the ancient Gauls that was to play a significant and often politically motivated role in France's intellectual life thereafter (Juaristi, 2000 : 234-42). The Scyths had been the great barbarian people of the East in classical ethnography and, unlike their Celtic counterparts in the West, had continued to figure in the Middle Ages. Indeed, the Irish had adopted a Scythian ancestor early on, presumably because Scottus sounded like Scythus. By the end of the medieval period Germanic peoples were widely regarded as Scyths and the sixteenth century saw « Scythomania » in full swing (Juaristi, 2000 : 157-228). It is to be borne in mind that, « when scholars in this period referred to Celts and Gauls, they had no modern sense that these were linguistic categories closely tied to Wales, Scotland and Ireland » but « were variously arguing that the classical Gauls, Celts or Germans were their own ancestors and had spoken whatever language was appropriate to that ancestry » (Chapman, 1992 : 203).
The Dutch Scythicist Boxhorn did break new ground in the mid-seventeenth century with linguistic arguments for a close relationship between Gaulish and British (Juaristi, 2000 : 246). Meanwhile the term « Celt » came into limited circulation as an alternative to « Gaul » in accordance with an essentially classical usage still seen in a book by Pezron tracing Bretons from Gauls within a Scythian framework that was published in 1703 with the title Antiquité de la nation et de la langue des Celtes, autrement appelés Gaulois (Juaristi, 2000 : 247-8).
Lhuyd's attempt to trace the « original language » of the British Isles in his Archaeologia Britannica of 1707 not only placed the long recognised relationship between Welsh, Cornish and Breton as well as their more recently surmised link with Gaulish on a firmer footing, but also first demonstrated a close connection between Irish and British and tended, albeit inconsistently, to use « Celtic » as an umbrella term rather than a mere synonym for « Gaulish » or « Gallic », with its increasing French associations (McCone, 2008 : 27).
The union of the Scottish and English crowns in 1603 led the Stuart wearers of both to coin the new term « Great Britain », which was turned into a political reality by the Act of Union in 1707 and attracted terms such as « Briton » and « British » into its orbit, the resultant gap soon being filled by Lhuyd's « Celtic » in the works of antiquarians with a druidic bent (Juaristi, 2000 : 253-6;McCone, 2008 : 27). The romantic appeal of Macpherson's works, which he published in the 1760s and fraudulently claimed to have translated from ancient Gaelic originals by Ossian, helped to fuel emergent « Celtomania » in Europe (Juaristi, 2000 : 253-6). Ireland, however, remained largely immune, not least because of resentment at Macpherson's denigration of Irish culture, along with the appropriation of the heroic Fianna for Scotland (Juaristi, 2000 : 267-9). Although the traditional connection with Spanish coreligionists initially retained its appeal in the face of the proposed link with the Protestant Welsh and Ledwich had been attacked for espousing Celtic origins in 1790, the likewise Protestant but fervently nationalist Thomas Davis managed to establish « Celt » as a synonym for the Gaelic Irish by the middle of the nineteenth century (Comerford, 2003 : 69-70).
« Celt » or « Celtic » had thus gained fairly general currency as labels applicable not only to certain ancient continental peoples but also to the inhabitants of areas where a Gaelic or a British language was still or had recently been spoken. Further encroachment by English in Ireland, the Isle of Man, Northern Scotland, Wales and Cornwall or by French in Brittany has since resulted in the traditional local idiom being no longer spoken by a majority of those living within their boundaries and often still regarded as Celts by themselves and others. Once language was not insisted upon as an essential criterion, a Celtic identity could be claimed in regions where no Celtic idiom had been spoken since ancient times, as by some Romance-speaking Gallegos whose name continues that of the presumably Celtic Gallaeci inhabiting the Spanish province of Galicia in Antiquity. Switzerland's former Celtic inhabitants, the Helvetii, were the basis for the seventeenth-century creation of a female symbol for the Swiss Federation named Helvetia, who appears regularly on Swiss coins and postage stamps, and have since provided the abbreviation CH (Confoederatio Helvetica) as a further means of avoiding the cumbersome claims of several national languages where space is limited.
The ancient Celts' remoteness in time and politically un-centralised occupation of a large area leaves scope for the imagination, as does the marginal location of most modern speakers of Celtic languages on Europe's Atlantic seaboard. Accordingly, they have been given many different roles in relation to language, culture, politics or even religion on levels ranging from the local or regional to the national or even European. Their potential as a prototype of cultural and political harmony over much of what is now the EU has, for instance, been exploited in the volume accompanying the major 1991 exhibition in Venice entitled I Celti: la prima Europa in Italian and The Celts: the Origins of Europe in English. The foreword to this refers to « the great impending process of the unification of western Europe, a process that pointed eloquently to the truly unique aspect of Celtic civilization, namely its being the first historically documented civilization on a European scale » (Moscati et al., 1991 : 14).

CHALLENGES TO THE « CELTIC CONSTRUCT », AND THE INDO-EUROPEAN FRAMEWORK
The Celtic identity constructed over the last three centuries has developed extensive ramifications with the help of romanticism, prejudice, wishful thinking or sheer fantasy in various proportions. It is small wonder that a reaction set in a couple of decades ago, when Chapman (1992) and James (1999) impugned « Celt(ic) » as a modern construct lacking a real historical foundation.
One is bound to agree with them that the medieval or modern populations concerned are hardly « Celtic » in any meaningful biological, in effect racial, sense. Not only is considerable genetic continuity with older Neolithic populations on Europe's Atlantic fringe demonstrated by recent samplings of DNA, but simplistic correlations between language and genetic makeup are also undermined by the recurring phenomenon of language shift. Hence « Celt(ic) » is inadmissible as a racially oriented term, but hardly more so than many other ethno-linguistic labels.
Its complete lack of currency in the Middle Ages also calls into question the common practice of applying the label « Celtic » to the medieval kingdoms, Christianity, society, literature, art or the like, of Ireland, parts of Britain and Brittany (e.g. Dillon & Chadwick, 1973). This back-projection of an essentially modern concept tends to detract from profound political and cultural affinities and interaction within the British Isles, including Anglo-Saxon England. This matrix might be more suitably termed « insular » without thereby denying a significant broader European input or the existence of the broadly « Irish », « British », « Pictish » and « English » regional variants identified by Bede (above).
That said, sceptics are on much shakier ground in questioning the validity of « Celtic » as a linguistic term, chiefly on account of the Indo-European hypothesis established and conclusively validated in the nineteenth century. Whatever their inadequacies by today's standards, works by Prichard (1831), Pictet (1837) and Bopp (1838) largely dispelled doubts about the Celtic family's Indo-European credentials. The foundations for the scientific study of its earliest attested stages, especially Old Irish, were then laid by Zeuss (1853), while further progress in comparative Indo-European linguistics led to and in turn flowed from the rigorous « Neogrammarian » approach initiated in the 1870s. Both strands converged in Pedersen's great work (1909Pedersen's great work ( , 1913 and subsequent advances in historical Celtic and Indo-European linguistics include a significant increase in information about ancient Continental Celtic, thanks to inscriptions discovered in recent decades.
As a result, it is a firmly established scientific fact that Irish or Goedelic and British or Brythonic belong with Gaulish and Celtiberian to a well-defined language family conventionally called Celtic, after Lhuyd, and itself a subgroup of the large Indo-European family. This is a serious stumbling block that revisionists like Chapman and James must try to negotiate.
In addition to expressing utterly unfounded doubts about the linguistic affiliation of the ancient Celts, both suggest that « Celtic » was hardly an automatic choice as the modern umbrella term. The obvious reply is Shakespeare's « What's in a name? ». One might as well question the conventional modern labels « Germanic » and « Indo-European » on the grounds that their respective speakers cannot possibly have called themselves by the Latin word Germani and a term reflecting a much later expansion. Indeed, « Celtic » or « Celt » is more valid than these and other similarly established terms such as (American) « Indian », insofar as speakers of the reconstructed « Proto-Celtic » language apparently called themselves *Keltoi.
Since people were the only effective means of spreading a language in the absence of modern mass media and Celtic speech apparently emerged in continental Europe around the turn of the first millennium B.C., its subsequent presence in Britain and Ireland must be put down to the migration of Celtic speakers thither from the Continent. That said, the incomers' numbers need not have been unduly large, if they established themselves as a dominant elite liable to be imitated by their more numerous subjects.

CONCLUSION
To conclude, in the centuries before Christ's birth peoples generally calling themselves Keltoi and speaking closely related languages occupied a broad arc of territory from Iberia to Asia Minor. Migration and invasion demonstrably furthered their expansion on the Continent, and can be safely assumed to have brought Celtic speakers first to Britain and then to Ireland. A distinctive subfamily of Indo-European can be identified as « Celtic » not only on grounds of well established modern usage but also because it reflects the name actually applied to themselves by speakers of its reconstructed « Proto-Celtic » source. Neither the reality of the linguistic continuity involved nor the unique suitability of « Celtic » as the name for it are open to serious doubt.