The Origins of the British: The New Prehistory of Britain (15 page)

In the first of two collaborative papers (see also pp. 292–3), Gray and Atkinson’s estimate for the Goidelic/Brythonic split is 2,900 years ago – during the Bronze Age. This could be conservative, since their estimates for the break-up of Romance and Germanic languages are only 1,700 and 1,750 years ago respectively, and in the latter case would seem to be an underestimate (see
Figure 6.2
).
⁶¹

Just before Gray and Atkinson’s paper appeared in
Nature
, the geneticist Peter Forster at the McDonald Institute for Archaeological Research, University of Cambridge, published a comparable figure (3,200 years) for the break-up of Continental and insular celtic, using a completely different dataset and method of estimation.
⁶²
Whereas the Dyen dataset included only well-attested, living and thus data-rich insular-celtic languages, Forster bravely took information from Continental Gaulish inscriptions and compared this linguistically with insular-celtic languages, both living and extinct.
⁶³
However, his tree explicitly acknowledges borrowing, resulting in ambiguity of some branches, thus causing the tree to become more of a ‘network’. Possibly as a result of this, Forster’s tree shows the two insular-celtic language branches Brythonic and Goidelic to be more closely related to each other than to Gaulish, but only just, and with a (branch ambiguity) at the break-up point.

The closest relatives to Gaulish are the insular-celtic Old Irish, Welsh, Scots Gaelic and Breton languages – in that order.
⁶⁴
In Forster’s tree, Gaulish is separated from the others by a very rough estimate of 5,200 years, consistent with the inference from Himilco’s
Periplus
that insular celtic could have arrived in the British Isles rather early. In Forster’s words:

For the fragmentation of Gaulish, Goidelic, and Brythonic from their most recent common ancestor, the … tree yields [an age of 5,200 years
⁶⁵
], but this … should be regarded as exploratory because it is based on only three estimators, i.e., three descendent branches. The [age of 5,200] years would represent an oldest feasible estimate for the arrival of Celtic in the British
Isles, and indeed is expected to be close to the actual date if the phylogenetic split between Gaulish and Insular Celtic was caused by the migration of the Celtic language to Britain and subsequent independent development in Britain.

On a deeper scale, Forster’s network tree also ‘yields a date for Indo-European fragmentation in Europe, as a whole,’ of 10,100 years ago,
⁶⁶
which is at least comparable to the figure arrived at by Gray and Atkinson (7,300 years), who used a much larger dataset. These old estimates are consistent with Colin Renfrew’s theory that the Indo-European language family originally expanded into Europe on the back of the arrival of the first agriculture there.
⁶⁷
I shall return to these dating methods later (see
Figure 6.2a
).

Coming back to the celtic branch, these two studies – which used completely different sources and methods – seem to agree that the break-up of individual celtic languages happened perhaps twice as long ago as the separation of each of the two largest West European groups, Germanic and Romance, and roughly the same time as the separation of Lithuanian from Slavic (although obviously these conclusions have to be acknowledged as provisional). Such datings would be more consistent with a Neolithic or Bronze Age celtic expansion than with the Iron Age one previously suggested under the southern German homeland theory (see
Chapter 1
). Not only that, but since Irish (Goidelic) and British (Brythonic) seem to be on opposite sides of a deep divide (
Figure 2.4
), the possibility of two early movements of celtic languages to different parts of the British Isles before 1000
BC
is not disproved. Finally, on an even deeper timescale, Gray and Atkinson’s estimate of the separation of the ancestor of the
celtic branch from the common ancestor of the other two West European language groups, Romance and Germanic, is very old indeed, with a Neolithic scale estimate of 6,100 years – or even older, if we accept Forster’s estimate.

Such a deep history of linguistic colonization, if real, might resolve an interesting apparent anomaly which lies at the core of Irish legend and culture. According to the orthodox academic view of ‘Iron Age Celtic invasions’ from Central Europe, ‘Celtic’ cultural history should start in the British Isles no earlier than 300
BC
. Yet Irish legend has
all
the so-called mythological cycle of invasions from the Continent done and dusted by 1700
BC
, which is the time of some of the earliest copper-mining in Ireland at Mount Gabriel in Cork. On this basis again, the first of the mythological invasions would have been at about the same time as the very earliest copper mining in Ireland, at Ross Island. These legendary estimates are based on two Medieval documents which list names and dates of Irish kings following the last Gaelic invasion, both pre- and post-Christian.
⁶⁸

The last three of the six invasions in the mythological cycle were over within the comparatively short period of 234 years.
⁶⁹
The first of these last three groups of invaders were the Firbolgs, whose name is generally given a celtic derivation by linguists. The Firbolgs are said to have fled Greece, where they had been enslaved and made to carry earth in bags. They were supposed to have made ships out of these bags and sailed to Spain before arriving in Ireland, where their dynasty of nine kings lasted only thirty-seven years. Around 1900
BC
the
Tuatha Dé Danann
, skilled artisans, replaced the Firbolgs and supplied another
series of nine kings, who ruled Ireland for a further two hundred years, until 1700
BC
.
⁷⁰
The
Tuatha Dé Danann
worshipped Dana, who was synonymous with Brigit, the Celtic mother-goddess.

The last of the legendary invaders were the Milesians, who, as mentioned above, have been identified by both archaeologists and traditionalists as being Gaelic, and as coming from Spain and possibly ultimately from Asia Minor. I have already noted David Rankin’s discussion on this theory, which partly relies on Thomas O’Rahilly’s reinterpretation of the traditional or legendary texts; I am now coming back to it again in the context of linguistic dating. O’Rahilly accepted the orthodoxy of the Iron Age as the period for all ‘Celtic’ invasions and simply changed the dates of the mythological invasions to suit, by cutting out 1,600 years between the last invasion and the birth of Christ. He re-dated the Gaelic invasion at 100
BC
, thus effectively killing off nearly all the numerous pre-Christian Irish kings. But he still accepted the older mythological cycle of invasions as valid evidence from before those king lists. O’Rahilly (and Rankin) also accepted that there was an association between the multiple earlier invasions, which O’Rahilly now dated over the previous 600 years, and Brythonic languages arriving from the nearby Continent. In arguing for these dates, and for the final Milesian invasion representing the arrival of Goidelic or Gaelic from Spain, O’Rahilly clearly creates a dating problem or anachronism by comparison with the much older language splits discussed above.
⁷¹
O’Rahilly, if he were alive today, might not accept the new linguistic dates. However, the anachronism can be addressed by replacing the southern Germanic homeland theory, for which there is less and less evidence, with a Spanish homeland theory, for which – as we shall see – there is
ample genetic data and even archaeological evidence of cultural connections.
⁷²

If celtic languages moved into the British Isles not in the Iron Age but, as suggested by linguistic dating evidence, in the Bronze Age or earlier, then the dates of nearly 2000
BC
implied by the Irish king lists may not be so fantastic.
⁷³
It should be remembered that the earliest text relating to these dates
⁷⁴
was written in the twelfth century
AD
. However, it was based on a compilation of older texts now lost to us, and gives considerably more credible detail and cross-references for the last two thousand years than do any of the chroniclers, such as Gildas,
⁷⁵
whose works are still relied upon in reconstructing the Anglo-Saxon invasion.

Barry Cunliffe gives parallel archaeological evidence for such an early cultural spread up the Atlantic coast from Spain to Britain and Ireland. In spite of this, in his more recent book
Facing the Ocean
he still seems to sit on the fence when it comes to origins of ‘the Celts’ and their languages. In apparent deference to the orthodoxy, he still uses a map (
Figure 1.3
),
⁷⁶
reproduced from his 1987 book,
⁷⁷
of Celtic tribal arrows coming out of southern Germany. But elsewhere he acknowledges its evidential weakness and describes an alternative archaeological southern-origin view which argues that cultural continuity along the Atlantic coast could have fostered the development of a celtic lingua franca along the coast. He adds:

It could be further argued that the language had developed gradually over the four millennia that maritime contacts had been maintained, perhaps reaching its distinctive form in the
Late Bronze Age … the archaeological and linguistic evidence support each other without being dependent.

Cunliffe then suggests that the famous Beaker pottery archaeological phase, which moved with early metallurgy into Ireland, might be a marker for this cultural-linguistic spread: ‘In this model Beaker prospectors were the carriers of the Celtic language.’
⁷⁸

The specifics of the first metal mining dates are critical in Ireland, in view of the legendary last three ‘invasions’. Cunliffe points out that, after Spain, where copper was being extracted as early as 3000
BC
,

[t]he main source of copper in the northern part of the Atlantic zone in the third and second millennia [
BC
] was Ireland … Two major mining complexes have been located in the south of the country, together spanning the period from c.2400 to 1500
BC
… The earliest so far known is at Ross Island … Co. Kerry … the mines were in operation from 2400 to 2000
BC
… A later more extensive series of mines was opened at Mount Gabriel … west Cork… [They] were in use over about the two centuries from 1700 to 1500
BC
… Finally we must turn to [north] Wales, where some thirty mining sites are now known … [F]our … were in operation in the first half of the second millennium.
⁷⁹

Significantly Cunliffe notes that ‘the earliest [Irish] metallurgy seems to be coeval with the appearance of the earliest beakers’,
⁸⁰
which confirms his view of their association as cultural markers.

The distributions of so-called Maritime Bell Beakers, a distinct form of pottery traded along the Atlantic and western Medi terranean coasts of France, Britain, Spain and northern
Italy in the early third millennium
BC
(see
Chapter 5
and
Figure 5.12b
) are remarkably coincident with the map of inscriptional evidence for celtic languages over two thousand years later. More specifically, when tin later came to be used to make bronze, it was mined from the same key metal-rich locations of the Atlantic coast (western Spain, Brittany, Cornwall, north Wales and southern Ireland) that made up the later tin trading networks of the first millennium
BC
discussed earlier (see pp. 40–1).

Finally, I should like to reiterate a point about dates and the Irish legends. There is a recurring theme in Irish tradition that the ancestors of the last two invasions arrived in Spain, by sea from an ultimate homeland in the eastern Mediterranean. The earliest linguistic dates given by Forster and Gray appear to suggest an early Neolithic date for the separation of the common celtic ancestor, well before the splitting of the other two West European language branches, Italic and Germanic.

Cunliffe has something to say on this Mediterranean trail. When discussing the theory of Beaker–celtic association (see also
Chapter 5
), he notes that ‘Other archaeologists have taken a more radical line, in considering Celtic languages to have been carried to the west much earlier by Neolithic cultivators.’
⁸¹
‘Other archaeologists’ refers to Colin Renfrew’s theory of the agricultural spread of Indo-European languages and the role of celtic languages in that spread.
⁸²
Cunliffe refers to this Mediterranean ‘roots’ possibility himself in terms of Neolithic trade between the Mediterranean and Atlantic coast from the fifth to the third millennium
BC
. For evidence of this, he cites the trade in Cardial Impressed Ware, a distinctive type of pottery which, during the early Neolithic, spread from Italy
along the Mediterranean coast to southern France and through his favoured Garonne river route to the Atlantic Coast (see p. 203 and
Figure 5.1
). As we shall see, there are many genetic parallels to such events during the Bronze Age, the Neolithic and even earlier.