I misread the one word in Alford’s text which I queried: when magnified, I see it is not nimiùm but nimirùm.


Nimirum would commonly imply certainty not doubt, although in some contexts it is seemingly (see Sir W. Smith’s Dictionary) used ironically; and Jean Baptiste Gardin Dumesnil (Synonymes latins et leurs différentes significations avec des exemples tirés des meilleurs auteurs, 1777) states that it is also used for scilicet. But in any case it is clear what it’s referring to in the sentence ‘ibi … ubi …… ‘. The battle was fought in Devonshire, Camden conjectures, there (namely) where the river Taw, wider than (or ‘widened by’) the waters of the river Torridge, makes for the Severn sea.

So Alford may not have been expressing scepticism regarding Camden’s suggestion, though neither is he affirming its truth. On the other hand, writing some thirty-odd years earlier, the Devon-born topographer Thomas Westcote, gent (bp. 1567-1637?), in A View of Devonshire in MDCXXX was less non-committal. He refers to the stories surrounding ‘Castle Hennaborough’, ‘Kenith-Castle’, Hubba the Dane and Whibbestow-Hubbastow and remarks:

“But to tell you truly, I find as many places in this county claim the honour of this victory, as cities in Greece for the birth of Homer.”

Interesting to note that he calls the ancient remains ‘Castle Hennaborough’ – almost 200 years before Vidal’s locals informed him that his ‘Kenwith or Kenwic Castle’ had only been been known locally as Henni Castle or Henniborough. In 1630 Westcote knew both names, though whereas remains of Castle Hennaborough may be seen, a certain mythical existence attaches to Kenith-Castle, which is also ‘hereby’. Westcote does not appear to equate the two.

Of these two names, Henniborough and Kenwith, the first presumably goes back to Saxon times, hēan+ēg+burh (as in Great Henny, Essex, Domesday 1086 Heni). Even now standing in the potential flood area of Kenwith Stream, Henniborough would then have been the ‘fortification in a high place (partly) surrounded by water’. When Benjamin Donne marked ‘Henny Castle Olim Kenwith’ on his 1765 map of Devon, it might more correctly have been ‘Kenwith Castle Olim Henny’.

No Viking Ship

Corinna Marion Davis’s Short History of Northam (1967) retells the story of Hubba’s defeat at the hands of Earl Odun’s Saxons, adding ‘for generations people feared to pass Bloody Corner at night, probably the relic of an ancient superstition. The remains of a Viking ship were discovered at Westward Ho!’

If such a ship had been found, it would have been the first – and only – such a find in Britain. That honour appears now to go to a ship located beneath a pub car park, as reported in 2007. Ship and boat burials are of a different kind, I suppose.

However, the appearance of the wreck of a wooden vessel has been recorded periodically since the nineteenth century, above the shifting sands off Westward Ho! In the 1940s this  was called ‘the Viking ship’ or ‘the Spanish galleon’ by local people. But if we are able to believe an expert in the person of a Historic England marine archæologist, this is most likely to be the remains of the Sally, carrying port to Bristol from Oporto and wrecked off Northam Burrows in 1769.


When it appeared last year, a tourist on holiday reported that this was still ‘a Viking ship’ for many local people. And who knows? Perhaps the very ship that ……

A detailed description of the wreck is available, with dendrochronology linking it to the second half of the 18th century.

The front-runner: Iscalis-Charterhouse


Charterhouse-on-Mendip has had the most impressive of advocates, from the time of Smith & Rivet’s  The Place-Names of Roman Britain (1979). But it has only ever been a tentative identification and doesn’t seem to fit the meagre scraps of information we have too well. Yes, ‘Charterhouse Roman Town’ was important as a lead-mining centre but:

1. The Roman name has been reconstructed as ‘Vebriacum’, because the letters VEB are found on a number of lead ingots found nearby and farther afield. I’m not sure why Vebriacum is the supposed form: in France a, probably pre-Roman, toponym was Latinised (and is attested) as Vebritum giving rise to Vebret (Cantal) and  other toponyms suggest a  Celtic Veb- origin. If Celtic and Roman forms were Veb-, why did Ptolemy call it Iscalis?

2. Isca- strongly suggests a name based on ‘isca’ – British ‘water’, perhaps on a river which the Celts called Isca; but Charterhouse isn’t on any significant bit of water at all. The nearest river flowed through Cheddar, the old River Axe (now the Cheddar Yeo), a couple of miles away. Cheddar had a small Roman port and the Axe would have been close enough to be used for transporting lead, but surely not close enough to Charterhouse for it to be characterised as ‘the place by the water’. That would be more likely to be Cheddar itself.

3. Charterhouse was an important centre for the Romans, but if Iscalis was a polis of the Belgae, as Ptolemy said, we are really looking for a place with significant Iron Age archaeological remains. Although there are signs that lead was mined at Charterhouse during the Iron Age, it was the Romans who developed it into a sizable industry. There was an Iron Age hillfort to the east of the village, but I can’t find anything to suggest it had special importance.

4. (And least likely to be of any importance at all) My reconstruction of the various Ptolemaic coordinates gave this:

Crossing ptsIt’s a bit rough and has obvious drawbacks (Iscalis wouldn’t have been in the middle of the Severn estuary for one thing); but allowing for triple inaccuracies, Ptolemy’s, the transcriber’s and mine – an inexactitude of inaccuracies – it does suggest something. Of the 10 crossing points, 9 are closer to the coastal possibilities near the mouth of the Axe or the mouth of the Parrett, real ‘watery places’. Against that, the one inland point is where the alignments London-Bath-Iscalis and Canterbury-Bath-Iscalis meet. I might suppose that London, Canterbury and Bath would be more likely to be accurately located than other towns around the country.

But, on the whole, I’m going to rule out Charterhouse as being not important enough to the Belgae and too far from any water. Also that the evidence seems to be that it was called something else.

Not very accurate (2)

Ptolemy’s coordinates are in many cases obviously inaccurate (Silchester and Exeter being two examples): the mistakes may have been his or later transcribers’ but, either way, we have to go by the manuscripts. I would think that combining a number of these coordinates would not necessarily result in a sort of “accurate” average, but it would be more accurate than the least accurate figures. It may be less accurate than the most accurate, but we can’t be sure which ones are the most accurate. We can, though, look for ‘patterns’ which work.

Bearing in mind that my own reconstruction hasn’t been realised with 100% accuracy (oh, no, it hasn’t: I was a bit flighty), I looked for alignments of three towns, one of which in each case was Iscalis and the other two were places whose positions we know. If these lines cross each other, how closely would they show Iscalis in the same place when transferred to a modern map?

So these were *approximate* (NB! Really!) alignments that I used on my reconstruction:






And this is how the lines converged:

Names lined up

It looks as if there’s  a very rough agreement among Ptolemy’s Roman towns as to where Iscalis was located. But what happens if the alignments are drawn on a modern map? As it happens, we’re still swimming in the Bristol Channel, but not nearly as far adrift as we were with the previous attempt:


The two red circles show the mouth of the river Axe by Brean Down and, slightly further south, the mouth of the river Parrett. The two small black dots show Cheddar and, to the north-east, Charterhouse-on-Mendip. More follows …

Not very accurate

I’ve been looking at Ptolemy’s information on the island of Britannia and have constructed a version of the southern part of England based on his coordinates. Whether it’s oriented correctly or whether it should be tilted slightly doesn’t matter for the moment. Some of his geographic locations (as far as they’ve been identified) are obviously wide of the mark; but, notwithstanding that there may have been a number of inaccuracies introduced between Ptolemy’s time and the late medieval copies that we have, some of it must have been at least what he intended – even if not necessarily accurate.

This is my rough (all right, very rough) map, with a few speculations, based on his latitude and longitudes:

Named map1

And just for a bit of fun I took a modern map of the part of Somerset where Iscalis is likely to have been. Ptolemy’s latitudes for Bath (Aquae Calidae) and Iscalis are the same (53°40); and the longitude shows Iscalis somewhat to the east of Pumsaint (Luentinum), the former at 16*00, the latter at 15*45. So taking Ptolemy at his word, X marks the spot – the site of Iscalis, one of the key cities or towns of the Belgae:


Iscalis Square

However, I shall abandon that line of reasoning.

The black rectangle above is the area where I think Iscalis might have been. The top and bottom lines are, respectively, the rough latitudes of London (Londinium) and Winchester (Venta Belgarum), to the north and south of the latitude for Bath; and those to the left and right are the longitudes of Carmarthen (Maridunum) and Usk (Bullaeum/Burrium), which extend the western and eastern limits of the longitude of Pumsaint.

Within this rectangle, the two lower red dots mark the Roman ports at Combwich, where there was a ford across the river Parrett, and Crandon Bridge, a little further up the river. The area round the mouth of the Parrett is favoured by some scholars as a likely site. The other red dot is at the mouth of the river Axe whose name suggests a form similar to Isca, from a Celtic word meaning water – Isca Dumnoniorum was Exeter, on the river Exe.

All these four possibilities (well, three if we discount the middle of the Bristol Channel) are to the west of Charterhouse on Mendip (not Hinton Charterhouse – I got a bit mixed up), another important Roman site and centre for lead-mining in Roman times, which some have suggested. I’m going to rule it out for the moment as well as Cheddar, another important Roman site, because I had another idea. Which will follow.

All Gaul divided into three – or perhaps four – parts

I’ve been studying – for no particular reason – the place names of Gaul, or rather France. From studying some of the old works on French toponymy, mainly Houzé and Nègre, it seems that something like 60% of the départments have place names (cities, towns, communes, villages) which clearly derive from Gaulish Condate, and which stand at the confluences of large or small water courses.

Extent of the place names Condat, Condé, Contes &c

Extent of the place names Condat, Condé, Contes &c

Group 1 consists of places called Condat, in the southern half of the country; Group 2 consists of places called Condé, in the north. This represents the normal phonological difference between the Langue d’Oc and the Langue d’Oïl. Free accented a, derived from Latin, remains in the south e.g. vadu > ga, amare > amar; but became the –ay– sound like –é– in the north e.g. vadu> gué, amare > aimer. According to Bourciez, this probably occurred some time in the Gallo-Roman period – perhaps the end of 7th century. So Condat and Condé are the southern and northern variants of the same name.

On this map, the dark red and the mottled red départements all have Condat/Condé and are differentiated (dark/mottled) only because I found them noted in different sources. Of course, it’s only accurate in marking the names we know about, but it may be significant that there are barely any examples in the west and north west (Brittany), nor in the south. Does that indicate a movement of people into ‘Gaul’ from the north east?

On the left is the south coast of England: Dubris is Dover, Iscadumnoniorum, Exeter. To the right, Condate, Rennes. All distances, and directions, are approximate.

Tabula Peutingeriana (4th/5th c.): On the left is the south coast of England: Dubris (upper circle) is Dover, Iscadumnoniorum (lower circle), Exeter. To the right, Condate is Rennes. All distances, and directions, are approximate…

Group 3 consists of a few places which bear quite different names now, but for which there is written evidence that there was an early settlement called Condate. The most notable examples are Rennes, which stands at the confluence of the rivers Ille and Vilaine and which is marked as Condate on the Tabula Peutingeriana (the original presumed to date from the 4th or 5th century); and Lyons which had two older settlements, Condate and Lugdunum. Condate stood at the confluence of the Rhône and the Saône, and Lugdunum overlooked it on the other side of the Rhône. The number of names in this group is unknown because unless there is written evidence in some form we wouldn’t know that the ancient name had been Condate.

There is a small fourth group of places which have a modern form that is neither Condat nor Condé but for which the evidence is that Condat(e) was an earlier form. Contz les Bains in the east of France, Condes and Conte in Jura – again in the east, Contes in Pas de Calais, Conty in Somme,  Condes in Haute Marne and several examples of Cosne.

These forms (except Conty) seem to derive from forms where the Gaulish accent was preserved on the antipenultimate: Cóndate > Conde(s); whereas the forms Condat and Condé come from Condáte, the later Latin/French accentuation.

What all names have in common (as far as I have been able to check) is that they all stand at or by a confluence, usually of at least one large river though the second is sometimes more of a stream – at least nowadays. So Condate itself must just mean confluence (as Koblenz derives from Latin confluentes, the confluence of the Rhine and the Moselle). I wonder if everywhere stands near a confluence if you look hard enough?

I believe there is just one inscription, at Allonnes, Sarthe, apparently addressed to a god (not clear if this is Condatis – I can’t find the text), rather than just the actual confluence; there are several more dedications in Britain.

Back to England again next time. Probably.

Amateur etymologist

Not knowing anything whatsoever about the esoteric subject of Old Welsh/Celtic historical linguistics, I’ve tried to read up on it a bit and hope tagging might attract here the attention of someone who knows better than me.

A steep learning curve

A steep learning curve


Looking at the (Romano-British?) name Cunetus, which seems most authoritatively to be at the root of the Welsh name Cynuit, Cynwyd and perhaps Breton Conet, relevant facts seem to be:

1. Cunetus was stressed on the normal penultimate syllable Cu-‘ne-tus

2. Some stressed vowels, in some circumstances, diphthongised (as in Romance); and in Old Welsh the unchecked -e- would diphthongise to -ui- (cf Common Celtic *ɸlētos > *luïd  > llywyd) – not quite sure about this.

3. The final syllable being unstable (as in Romance), the hypothetical development might be:

Cu-‘ne-tus > *Cu-‘nuito

4. Final syllables dropped (again as in Romance) so:

*Cu-‘nuito > *Cun-‘uit and perhaps at this point the first (unstressed) syllable weakened from the Latinate oo [ʊ] to Welsh y [ə]?

5. The stress then shifted forward, so that it was once more on the penultimate:

Cyn-‘uit > ‘Cyn-uit

6. I think final unvoiced vowels became voiced (unlike Romance):

‘Cyn-uit > ‘Cyn-uid, ‘Cyn-wyd

These seem to be standard developments as outlined in David Willis, Old and Middle Welsh (2008?).

The point about diphthongisation would explain Breton Conoit (Conet), Langunuit, Langenewit (Langunnett), Llangynwyd (Llangoned); and the (unsourced) form St Knuet, which sounds like *Cy-‘nuit.

I am wondering if, when the stress shift took place, the (then) unstressed final diphthong simplified into a single vowel? This would explain why Langunuit/Langenewit became Langunnett and why Llangynwyd is also found in the  form Llangoned. But there seem to be two important points:

i) that the original stressed vowel, that -e-, survives in some form: LangunnETT, LlangynWYD, LangonnET (Brittany, and still the stressed vowel there), LlandigwinnET. There is still a distinct syllable between the N and the T, even when it loses the stress.

ii) Not much evidence here, but at the time of Domesday, 1086, the spelling appears to have still included the diphthong (DB Langenewit, Exon Langunuit).

I expect it is clear now where all this is leading. If not I will stumble on in the next post.