I've noted before that in the center of Ceredig's Ceredigion is the Afon Arth, with its tributary the Nant Erthig ('Little Bear'). There was a promontory fort at Aberarth Llandewi, and the medieval castle at Dinerth (confluence of the Afon Arth and the Nant Erthig) may well have had an ancient precursor.
This river and its forts are significant, given that three bear-names appear among Ceredig's descendants:
Ceredigion
[G]uocaun map Mouric map Dumnguallaun map Arthgen map Seissil map Clitauc Artgloys map Artbodgu map Bodgu map Serguil map Iusay map Ceretic map Cuneda.
This is rather amazing, as there only a few bear-names in the remainder of the Harleian pedigrees, e.g. Arthur son of Pedr (Dyfed), Arthgal (Ystrad Clud), Artmail (Gwent) and Artan (Powys).
Arthgen is 'Bear-born' or "Born of the Bear', Artgloys is either 'Beautiful Bear' or 'Pure or Holy Bear' and Artbodgu is 'Bear-crow', the second element being inherited from the name of Artbodgu's father.
With a name like Arthgen, we must naturally ask what bear it was this King of Ceredigion was born from. I've suggested we are talking about the river itself, as such bodies of water are often deified. The mythological Math and Mathonwy names of Caer Dathal, where Uther Pendragon had kin and from which Arthur took a wife, are Irish bear-names.
When it comes to attaching Arthur to Cerdic/Ceredig, we are left with really only one possibility, recently detailed for me by Dr. Simon Rodway:
"As for *Artorīx ['Bear-king'], Marged Haycock is absolutely right to say that it would give *Erthyr. The composition vowel /o/ (short) would become y through final i-affection. Then the final syllable would be lost, and the /a/ would become /e/ through internal i-affection (in which y took part). This is all perfectly well understood from numerous names in -rīx. A comparandum would be *Carantorīx > Cerennyr with /a/ > /e/.
As for Artorius, yes, it could be a deckname for Artorix. Artorius is well-attested. Artorus is a very plausible by-form of Artorius, and that would regularly give Arthur."
Thus we could say that Ceredig of the Afon Arth, a deified river, was the King of the Arth and, by extension, the Bear King. But that at some point Artorius was substituted for *Artorix.
I would again mention the known substitition of Roman names for Celtic ones that were of a perceived semantic match:
Trier (CIL XIII/1.1, no. 3909)
HIC QUIESCIT IN PACE URSULA . . . ARTULA MATER TIT(ULUM) POSUIT
In this case the mother (Artula, Little Bear) and the daughter (Ursula, Little Bear) have the same name, the mother still in Celtic, the daughter already in the Roman tongue.
Among specialists of ancient Italic languages, Artorius is said to be derived as follows (this from Blanca Maria Prosper Perez discussing ARTORRES):
"The most promising etymology would derive artor- in Latin and Messapic from the root meaning "adjust", "assemble", "fit", though it is impossible to reach a really specific meaning."
Professor Stefan Zimmer actually attempted to derive Artorius itself from the Celtic (see
However, Blanca Maria Prosper Perez made an interesting comment regarding Zimmer's theory:
"It could perhaps explain the origin of Arthur's name, but neither the Latin nor the Messapic names."
Here is what Zimmer has to say:
"Recently, Ch. Gwinn (apud Delamarre 2003: 56) has proposed to
understand Artorius directly as a Latinized version of *Arto-rīχs. This
cannot be excluded categorically, but is highly unlikely. The Romans used
to treat all the many Celtic names in -rīχs, well known at least since
Caesar’s commentaries on his Gaulish wars, like their own word rex, regis
because the close relation of the two lexemes (in fact, their etymological
identity) was obvious to them. So, a Celtic *Arto-rīχs should
automatically appear in a Latin context as *Artorix – but it never does.
With due caution, I propose to understand Artorius –
exclusively used as gentilicium as mentioned above – as the Latinized
version of a Celtic patronym *Arto-rīg-i̯os. This is nowhere attested in the
Celtic world. But the basic *Arto-rīχs is, see above OIr. Artrí; the British
forms with second member *-maglos are but a variant of the same.
The patronymic type in *-i̯os is well known, cf. Gaulish names such as
Tarbeisonios ‘son of Tarbeisu’; and of course, outside Celtic, especially
in Greek. It should be safe to assume as a working hypothesis that the *-gunderwent
a kind of early lenition (whether a Latin or Celtic phenomenon
need not detain us) to the spirant -ʒ -/-j-, giving, with subsequent
assimilation of [ʒj] > [jj] > [j], *Artorījos.7 The Latinization implied two
simple adaptations to Latin:8
3.1. According to Latin writing conventions, *[artori:ʒjos] or
*[artori:jos] was spelled, with automatic replacement of the Celtic ending
by Latin -us, as Artorius.
3.2. Following the obvious and frequent pattern of Latin nouns in
-ōrius, -a, -um, regular derivative adjectives to agent nouns (including
proper names) in -tōr, the short -ŏ- in *Artŏrius was replaced by -ō-, and
the long -ī- shortened, thus producing a totally Latin-sounding Artōrius.
3.3. The subsequent phonetic development from Latin Artōrius to
Early Welsh Arthur is perfectly regular (cf. L labōrem > W llafur, etc.)."
However, I had an idea: what if we allow for a standard Latinization of a British *Arto-rīg-i̯os to Artorius? If we could do this, then we don't have to propose that Artorius was substituted as a deckname. We could look at Arthur merely as a natural development of *Arto-rīg-i̯os through early Latinization.
I took this idea to Dr. Simon Rodway (Welsh expert), Professor Ranko Matasovic (Proto-Celtic/Celtic expert) and Alan James (Brythonic expert). [As I hear back from more linguists, I will add their contributions below.] Their responses?
Rodway:
"You can only get Arthur from *Arto-rig-ios by supposing conflation with Latin Artorius. This is not impossible, but is uneconomical, and it is easier to derive it straight from *Artorus, an unattested but quite plausible byform of Artorius."
Matasovic:
"I agree with you that there may have been an early British name *Artori:gyos, and it may have been Latinized in Britain at an early date.
If the name *Artori:gyos was heard by Latin-speakers after the fricattivization of -g- they would have adopted it as *Artorius (with short -o-); then they would have to change it to Arto:rius on the analogy with other names in -o:rius (as Zimmer says), and then it would have to be borrowed into Welsh again as Arthur. Complicated, but maybe possible."
James:
"The short answer is, yes, Arthur could be a regular reflex of *Arto-rīg-i̯os. I don't know that I 'favour' a Celtic origin for the name, only that I see no problem with it being one.
Patrick S-W has a typically detailed and thorough discussion of names in -orios in The Celtic Inscriptions of Britain, pp. 33-4, where he supports the view that that suffix is a 'by-form' of -orīx, Latinized as -orius, at an early date (pre-200). But he does note that - unlike in several Brittonic names of this type, and more like Roman ones such as Victōria - the vowel is long, *Artōrius.
Patrick just signals that as a 'proviso', a possible reason for doubting whether*Artōrius would be exactly comparable with the Brittonic -orius names he's discussing.
As to the ms records, we have, as I said, Arthur, with various Latin inflexions tagged on, in Harley 3859, from somewhere in the range 1100x1200. It's reasonable to infer that the name, in that form, was current at the time HB was put together, i.e. early 9th ct. And the same name appears in Llyfr Aneirin, from the late 13th ct, in the 'A series' of awdlau, whose language is no earlier than around 1100; again, we may infer that the name was current in the late 1st millennium. Whether or not it's a 'late interpolation' in some earlier version, and whether or not that supposed earlier version was part of an imagined 6th ct poem, are irrelevant, all we have is evidence for a name that was probably current in the Old Welsh period, roughly 800-1100.
But I'm still wondering when and where in the manuscript record Artorius first pops up? I must say I'm coming to suspect that it might be a relatively late invention of a medieval monk or scribe, a name sounding more agreeably Latin than the rather clumsy Art(h)urus. Which is not to deny that the name Arthur can be derived from an early Brittonic *Artorijos, which in turn may be a by-form of *Artorix."
In other words, the Latin Artorius need not come from some Roman era man of that name (like L. Artorius Castus), nor do we need to see Artorius so much as a substitition for an earlier British 'Bear-king" name, but instead merely as a natural product of Latinization, which brought *Arto-rīg-i̯os in line with Artorius.
Arthur, then, could be, essentially, *Arto-rīg-i̯os pronounced like Latin Artorius and, hence, indistinguishable from Artorius.
While ordinarily it would be impossible to make anything other than a highly speculative and somewhat weak case for such development, if Ceredig of the Arth Water and the Arth- personal names is Arthur, the argument becomes considerably stronger for an original *Arto-rīg-i̯os.