AN APOLOGY FOR ARTICLE RETRACTION: WHY I WAS WRONG ABOUT PTOLEMY'S SELGOVAE SITES

Ptolemy's Britain from Alastair Strang, With Four Selgovae Sites Highlighted in Red

Sometimes, no matter how hard on tries not to, you still end up going down a rabbit hole.  The consequences of such a journey can be embarrassing.

In an effort to pinpoint the four cities of the Selgovae mentioned by Ptolemy, I resorted to a new study by Keith Fitzpatrick-Matthews.  Unfortunately, after publishing several findings based upon my research, I've come to realize that the methods he employed are faulty.  And, as a result, his conclusions are neither compelling nor satisfying.  

The realization came when I recognized that his identification of the Ravenna Cosmography's Loxa with Aberlosk was incorrect. I did this by drawing upon the earlier, much more careful work of Rivet and Smith.  They had identified the Loxa with the Lossie in far northern Scotland.  To confirm their idea I had only to find the Locatreue/Locatrebe which immediately follows the Loxa in the RC list.  And there it was on the map - Loch Trevie, a small lake that is part of the source of the Lossie.

What this meant is that other names were also properly placed in the far north.  An example is the Smetri (*Smerti), a tribe located there by Ptolemy, but which Fitzpatrick-Matthews tries to find as a place-name in Lowland Scotland.  Other RC place-names are equally in doubt, and many are hopelessly corrupt or seriously scrambled in a geographical sense or lost and so can't possibly be associated with extant place-names.  Even the Lindum Fitzpatrick-Matthews would related to the Lyne fort on the Tweed is almost certainly Ptolemy's Lindum at Drumquhassle in Stirlingshire.  Etc.

So... I had to go back to the drawing board.  And the best source to use was once again Alastair Strang's masterful "Explaining Ptolemy's Roman Britain" (Britannia, Vol. 28 (1997), pp. 1-30).  This study aims to "adjust" Ptolemy's map by applying very complex geographical principles.  The result is a splendid rearrangement of Ptolemy's sites which has a very high probability of being accurate.

While a proponent of etymology, this the case of the RC positional computation has a much better chance of yielding valuable results.  A good example of the limitations of toponymastics in this particular application can be demonstrated by the place-name Alitacenon.  Rivet and Smith discuss their belief on why this name should be emended as *Alaunacelum, 'the headland/promontory/spur of [a river called] Alauna.'  I wrongly linked Alaunacelum to Castle O'er at Allangill Burn, while Fitzpatrick-Matthews opted for the Ale Water.  Of course, there are many Alaunas, so even this proposed form doesn't get us very far.  But years ago I asked Dr. Graham Isaac, a noted Celtic linguists and place-name specialist about Alitacenon.  This was his take on the problem:

The Ravenna Cosmography’s Alitacenon could be corrupt beyond redemption, but if it is accurate, then both elements are unproblematically found elsewhere: alita- 'shining [river]' gives W Aled (RN), and -cenon is a common toponym element, of admittedly uncertain meaning. I asked Dr. Isaac about –cenon in the context of Alitacenon. If Alita- meant originally the "Shining" (-river), could not -cenon be from Proto-Celtic *cen-je/o, "rise (from)"? In other words, Alitocenon was at the headwaters of a stream called Alito, the place where the waters of the river rose from. To which he responded, “This is not impossible.”

Needless to say, we know of no Alito- site in the North.  As with *Alaunocelum, we are at a loss when it comes to placing this name on the map.

Now to turn our attention to Strang's identications of the Selgovae sites.  He quite reasonably makes the following identifications:

Carbantoritum - Raeburnfoot

Uxellum - Rubers Law?

Corda - Crawford

Trimontium - Newstead (replacing the native hillfort of Eildon North Hill; this is a confirmed site)

Newstead is beyond doubt Trimontium, named for the three Eildon Hills. Some have thought the Tweed to be the border river between the Selgovae and the Votadini.  However, given that both Traprain Law and Edinburgh, the chief seats of the Votadini, are north of the Lammmuir Hills, it is likely the Selgovae controlled both sides of the Tweed basin. 

As for Carbantoritum or 'Chariot for' at Raeburnfoot, I think he may well be correct. The Rae Burn is a tributary of the Esk, and the Esk for the historical period has always been a border river.  In fact, it is known as the Border Esk precisely because it divided NW England from Scotland.  It is probable that the Selgovae border with the Novantae lay here.  Incidentally, there are a few major hillforts just south of Raeburnfoot. We know the Anavionenses lived in Annandale, the next river valley to the west, and the former are believed to be a sub-group of the Novantae.  Hence the Selgovae lands could not have extended to the Annan.

Strang rightly only tentatively suggests Rubers Law.  I'm not sure he is right here, as his rearrangement of Ptolemy's sites leaves Uxellum on almost exactly the same latitude as Raeburnfoot and more to the west of the longitudinal line on which we find Trimontium.  This does not fit Rubers Law.  

A better fit would be Burgh Hill, site of a hillfort, stone circle and numerous satellite settlements, some of which continued in use up through the Dark Ages.  For the sake of comparison only, I have plotted these sites, as well as Milton in Annandale (which while it looks good in one sense, cannot have been Selgovae for the reason I've already mentioned above). 

I would, then, solely on the basis of Strang's meticulous correction of Ptolemy's map, choose Burgh Hill for Uxellum and his Raeburnfoot for Carbantoritum.  According to Canmore (https://canmore.org.uk/site/54016/burgh-hill):

"The nearby hills were quite well settled in the iron age by British tribes, and subsequently by Northumbrians for whom the Catrail (NT 4804- 4904: a considerable linear earthwork just visible from Burgh Hill towards The Pike, south-east across Dod Burn) may have been a temporary 'frontier' before the mid 12th century."

For more on the Catrail, see https://canmore.org.uk/site/86604/catrail.

If we MUST have a Roman site associated with a native one, then the only Uxellum I can think of in the right general area if Woden Law hillfort with its multiple Pennymuir Roman camps.  This is, however, quite a bit further to the east than Rubers Law. [https://canmore.org.uk/site/83741/pennymuir, https://canmore.org.uk/site/58068/woden-law]. The Pennymuir camps on Dere Street sit at the foot of the Cheviots, and it is probable that this range of hills marked the southeastern boundary of the Selgovae,  No Roman fort has been found at Rubers Law, either.

Which leaves us with Corda (or Coria, if one accepts Rivet and Smith's emendation).  Crawford has never made sense to me for a Selgovae 'hosting place' or administrative center.  It is in upper Clydesdale, and while it's always dangerous to try and see in Dark Age kingdom boundaries the earlier Roman period tribal boundaries, the very name Strathclyde suggests strongly that the Clyde belonged to the Dumnonii.  On the other hand, the Selgovae seem to be confined to the Tweed Catchment Basin.  Lyne/Easter Happrew on the upper Tweed - which, ironically, is Rivet and Smith's choice for Carbantoritum! - looks to be a very good Corda.  

Granted, on Strang's reconstructed map, Corda is lower in latitude than Trimontium.  This fact definitely favors Crawford.  But we've seen that Uxellum and Carbantoritum are on the same latitude according to Strang, yet it is clear Raeburnfoot is somewhat lower than Burgh Hill.  If we allow for a similar, but opposite latitudinal error for Lyne, it would be exactly where we find it on Strang's map.

Alternately, we could opt for Dreva Craig further south on the Tweed as Corda.  This is an area replete with hillforts and standing stones.  It might better satisfy the requirement of Strang's map, although he doubtless would have preferred it be further west.  See https://canmore.org.uk/site/49881/dreva-craig.  According to Richard Feachem (GUIDE TO PREHISTORIC SCOTLAND), Dreva Craig commanded "the eastern mouth of the Biggar Gap, the valley 7 miles in length which affords easy communication between the upper reaches of the Rivers Clyde and Tweed."

The four cities of the Selgovae could then be situated thusly, with what may have been the royal cemetery of Yarrow at the center of the kingdom:

Tweed Catchment Basin