The P-Celtic Place-Names of North-East England and South-East Scotland

This is a preprint of an article whose definitive and final version is available free-access as The Heroic Age, 10 (2007), http://www.heroicage.org/issues/10/fox.html.

Bethany Fox

Introduction

The processes whereby the p-Celtic dialects died as large areas of medieval Britain became English-speaking have been a major matter of debate for over a century now, a crux being the apparent dearth of place-names of p-Celtic etymology surviving in most of Anglo-Saxon England.[1] Against a backdrop of change and development in historical and archaeological research (e.g. Dark 2000; Higham 1993; Koch 1997; Lucy 1999; Rollason 2003; Thomas–Stumpf–Härke 2006; Woolf 2003), historical linguists are now bringing increasingly subtle, informed and fruitful models of language change to bear on the issue (e.g. the articles in Filppula–Klemola–Pitkänen 2002; Hough 2004). Given its importance, however, the surveying of surviving p-Celtic names in areas which became English-speaking during the Middle Ages has been surprisingly lacking in intensity (the works by Coates and Breeze listed in the bibliography representing the main contributors in this area). Moreover, the study of Anglo-Saxon Northumbria specifically has been further impeded by an anachronistic tendency in place-name studies to treat England and Scotland as discrete regions. The prevailing assumption is therefore of ‘a near wholesale replacement of British place-names throughout Northumbria in the Anglo-Saxon period’ (Higham 1993, 100).

Focusing on the region between the Firth of Forth to the north and the River Tees to the south, this study gathers together those names identified by past scholars as possibly p-Celtic, providing a new assessment as to whether each contains p-Celtic elements. With the identification of eighty-four names probably containing p-Celtic elements, and forty-five further possible examples, it emerges that p-Celtic toponyms in the region are surprisingly numerous. Although statistical comparisons cannot meaningfully be made, the frequency of these names relative to our total corpus of surviving early medieval names is suggested by the fact that the earliest identifiable stratum of English settlement names in the region, those ending in -hām and -ingahām, number no more than forty, while there are probably between ten and fifteen names containing the p-Celtic settlement-term caer (‘fort, fortified building’) and between sixteen and nineteen containing tref (‘farm’)—not a very much smaller number.[2] Moreover, there is reason to think that the distribution of p-Celtic names is historically significant: generally speaking, the distribution of -hām and -ingahām names is mutually exclusive of p-Celtic names. It is hoped that foregrounding these p-Celtic names in this way will encourage more detailed reconsiderations of the more problematic etymologies, and stimulate further investigations of the p-Celtic component in English and Scottish toponymy, along with its significance for medieval British history.

Methods

P-Celtic dialects were at one time spoken throughout Britain. How different they were from one another is hard to judge, but our early medieval evidence shows enough distinctive features for us to distinguish between dialects such as Cornish, Welsh, Cumbric and Pictish. The dialect relevant to this study is Cumbric, the term Cumbric being applied, insofar as there is evidence to apply it to, to p-Celtic throughout what is now northern England and Scotland south of the Forth-Clyde line.

In the past, the p-Celtic place-names of North-East England and South-East Scotland have generally only been studied by scholars whose speciality is another linguistic stratum: Mawer (1920), the most respected scholar of Northumberland and Durham place-names, shows the Old English bias characteristic of English place-name studies of his time, and though more knowledgeable about Celtic philology, Ekwall’s later revisions (1960) are not without bias either (Coates 2004, 5). Besides the general surveys of Scottish place-names by Watson (1926), Mackenzie (1931) and Nicolaisen (1976)—none of which pays a great deal of attention to p-Celtic names south of the Forth—the place-names of South-East Scotland have received some specific attention (e.g. Macdonald 1941). However, Robb has shown that works on Scottish place-names, particularly the more popular ones, tend to be biased towards the study of Gaelic names (1996). Moreover, the adjacent areas of North-East England and South-East Scotland have rarely been juxtaposed in one study. Since the area between the Forth and the Tees once formed a single Anglo-Saxon polity, known by the etymologically p-Celtic name of Bernicia, it makes sense to study this region as a whole. Subsequent developments, of course, may have led to differing patterns of toponymic survival on each side of the border: the region to the south of the modern border seems, for example, to have been relatively monolingual in the early Middle Ages, whereas the northern area saw serious competition between English-, Brittonic-, Gaelic- and Norse-speaking power-groups, and so its inhabitants were perhaps more accustomed to encountering and maintaining place-names in unfamiliar languages. Likewise, the modern study of the different regions has been different: perhaps most importantly, work on south-eastern Scotland has been more intensive than on Northumbria, meaning that more, and more minor, place-names have been identified north of the border than in the region surveyed by Mawer. While it does not seem likely that Mawer missed many plausible p-Celtic etymologies for the names which he studied (cf. Coates 2000b, 296, 323–24), a more intensive survey would doubtless produce more material, and possibly more p-Celtic material. A case in point is Milfield, which did not appear in Mawer’s study and is ostensibly English in derivation: it is almost certainly to be identified with Bede’s Maelmin, a p-Celtic name.

The precise areas surveyed here are the pre-1974 counties of Northumbria, Durham, Selkirkshire, Roxburghshire, Berwickshire, Peeblesshire and the Lothians. From the main studies of the place-names of these counties (cited above), I have collected all the names for which a p-Celtic etymology is suggested. This survey was supplemented by reference to numerous other works on place-names in the area, but due to their less reliable nature, I have included names and etymologies drawn from them only when I consider them to be reasonably plausible (e.g. Hull 1939; Johnston 1934; Dixon 1947; Watson 1995). One kind of name is excluded here: river names and place-names whose only p-Celtic element is a river-name. River-names tend to be more stable than other kinds of names, and so are less useful for the historical study of linguistic and cultural change which is the object of the present study. Moreover, this study focuses on the distributions of place-names, but rivers are hard to plot meaningfully in these terms in a region of Bernicia’s size. It must be emphasised that a complete reassessment of all the etymologies gathered is beyond the scope of the present paper: my aim is rather to establish the corpus of place-names in South-East Scotland and North-East England which probably contain p-Celtic elements, and to comment on their distribution. Needless to say, it would be revealing to extend the survey, incorporating for example the long-lasting p-Celtic speaking regions to the west of the survey area (including, to various extents at various times, the ethnically British kingdom of Strathclyde), and perhaps indeed more southerly regions such as the polity of Elmet, which seems to have existed in the vicinity of Leeds, and its even more shadowy companion Reget.

Previous discussions of each name are summarised in the appendix. Names are identified there first by their latitude and longitude, and second by their Ordnance Survey grid references. It is admittedly conventional to locate place-names by their co-ordinates on Ordnance Survey Landranger maps, but the usefulness of these is now diminishing as paper maps are superseded by electronic ones (such as the free Ordnance Survey maps at www.streetmap.co.uk). The current county in which each name falls is also given. It is regrettable that the pre-1974 counties could not be listed, but it was not always possible to identify in which a place-name would fall, and the consistent application of current names was thought preferable. I have made a point of citing the most important (usually the earliest) attestations given in the secondary sources, hyperlinking from each name's entry in the appendix to a separate list of attestations. It is a familiar complaint not only concerning the more popular surveys, but also those of Watson and Nicolaisen, that early attestations of names are not always given, and when they are, are not always referenced. Fortunately, the works of Macdonald (1941), Dixon (1947) and Mawer (1920) afford more comprehensive data. Of these, only Mawer’s is readily accessible, making the inclusion of this material particularly important. Each attestation is given two references: first, the primary source from which it came (full references to these sources are listed at the end of the attestations), and second, the secondary source in which I found it.[3] Finally, the appendix gives each place-name a designation according to how likely it seems that the name is indeed p-Celtic. It should be noted that a designation of ‘unlikely’ may show that I have considered all the evidence available to me and decided it to be insufficient to give the name a p-Celtic derivation, not necessarily that the name itself is not p-Celtic. A ‘possible’ designation, meanwhile, is given for a name where evidence points plausibly to a Brittonic derivation, but where another derivation which is given also seems possible, or where it is plausible, but there is insufficient evidence to be sure. A ‘probable’ designation is given to names, the p-Celtic derivations of which are agreed by several scholars, or where different derivations are given by different scholars but all are p-Celtic in origin, or where the name has characteristically British form in early attestations.

I present p-Celtic place-name elements in their equivalent Modern Welsh forms, e.g. tref for neo-Brittonic *trebā. This approach is admittedly losing favour as the number of resources available to the p-Celtic toponymist grows (e.g. the Geiriadur Prifysgol Cymru; Padel 1985; and the emerging Vocabulary of English Place-Names, Parsons–Styles 1997–; cf. Matley 1965); and it should be emphasised that although the definite article is generally given as y (as this is its form in Modern Welsh) its most likely form in Cumbric is yr (Nicolaisen 1976, 168). Modern Welsh, however, retains the attractions of ease of consistency and comprehension, and is the form used by most of my secondary sources, and so has been preferred here.

The electronic mode of presentation used here has made it possible to provide a distribution map which gives the user a relatively large degree of control over what information is presented, and on which the identity of any given plot can be established by moving the mouse pointer over it. Though still primitive in terms of the possibilities offered by technology (and restricted in part by the need to work efficiently and reliably on a variety of web-browsers), this is still unusual in the field of place-name studies, and its advantages are hopefully apparent. The topographical data for the map was derived from open-access sources: rivers and national boundaries from the GeoCommunity's GIS data depot (specifically from http://data.geocomm.com/catalog/UK/datalist.html), and relief from the CGIAR Consortium for Spatial Information's SRTM 90m Digital Elevation Data.

General linguistic issues

Several names show features which are worth discussing: a lack of lenition where it would be expected grammatically; a mixture of p-Celtic elements with those from other language groups; and a high frequency of certain elements. A further issue is the possibility of giving a relative dating to the names on the basis of the order of generic and determining elements which they show.

The fact that several names discussed in this study do not show lenition where they would be expected to in Welsh is easily dealt with, since Jackson has shown that such phenomena can be explained by sound-substitution between Brittonic and Old English (1953, 195); a further cause is likely to be irregular sound-changes and folk-etymologisations following the loss of lexical meaning from the place-names caused principally by the death of Brittonic in the region studied.

Several names mentioned in this study have been said by scholars to be hybrids, containing elements from two different languages. By hybrid, however, it should not be understood that the name was coined as such. In names such as Traprain Law the first word is Brittonic and must have existed independently before the coining of the modern name. An English-speaker then must have borrowed the settlement-name Traprain, presumably only as a naming-element rather than a meaningful phrase, and used it to refer to an adjacent law (hill) (previously called Dunpelder). Thus, the name Traprain Law means, in its English form, ‘Traprain hill’, rather than ‘hill of the farmstead of the tree’ (cf. Nicolaisen 1976, 4–7). The problem of hybrid names is more acute when it comes to Gaelic-Brittonic hybrids. In Gaelic, as in Brittonic, names can be formed in two ways: as a compound (with both elements in the nominative) and as a name-phrase (with the generic first and the second element in the genitive case, as in Auchenfedrick, i.e. Achadh Phadruig ‘Field of Patrick’). In both languages, the second type is much the most common, and the name-elements often similar in form. In some cases, certainty is impossible (as with Ecclesmachan); in others, factors of semantics or stress can aid interpretation (e.g. Carriber).

There are several groups of names which begin with the same sequence of letters in this study. For example, many names begin with cVr- (where V is any vowel). Most of these can be derived from Brittonic caer, ‘fort’ (on which see Padel 1985, s.v. *ker), although some may be from Brittonic carn, ‘cairn’, and one or two may be from Northumbrian Old English carr, ‘rock’, itself borrowed from Brittonic. At times, the modern form of such names appears as Caer-, through analogical remodelling on the basis of similar Welsh names. Several names begin with trV- and these can mostly be derived from Brittonic tref, ‘farmstead’ (on which see Padel 1985, s.v. tre). The names beginning with tVr- can either be derived from tref or from Brittonic tor(r), ‘hill’ (on which see Padel 1985, s.v. tor; metathesis of /rVC/, particularly /rVn/, > /VrC/ is very common in Old English: Campbell 1959, §459).

Names beginning with pen-—particularly distinctive because of the dearth of ps in Old English—are more problematic. Most of them are likely to be from Brittonic pen, ‘summit, promontory’ (on which see Padel 1985, s.v.), but those in penny- may be derived from pen y, ‘hill of the-’ or from Modern English penny ‘a penny’ or its ancestors. In these cases the guiding factor must be the second element. There is also a group of names beginning with prVn-, prVm- or pVrn-. These have been derived from the Brittonic element pren, ‘tree’ by past scholars.[4] However, if all these names are to be derived from this element we must posit a substantial lexical difference between Cumbric and other Brittonic languages, for this element is very sparsely attested in Welsh place-names, and appears only three times in Cornwall (Padel 1985, s.v. pren). Alternatively, these names may be derived from Brittonic bryn, ‘hill’ (for which see Padel 1985, s.v. *bren; Parsons–Styles 1997, s.v. *brunnjo-). However, we must then posit an unusual sound-substitution of /p-/ for /b-/ at some point in each name’s history. The fact that bryn, while common elsewhere in p-Celtic toponymy, occurs only rarely as a first element in etymologies collected by this study supports the idea that some or all names previously thought to contain pren in fact contain bryn. Whichever element one chooses, however, the derivation is still Brittonic.

Hough has recently emphasised that such evidence as we have suggests that up to about the sixth century, Brittonic compound place-names (like English ones) placed their determiner first and their generic second (as in Moridunum, ‘sea-fort’); from about the sixth century, the order was reversed (giving us modern Welsh place-names such as Llanfihangel, ‘the church of Michael’; Hough 2001). Discussing tref-names, Hough assumed that tref was necessarily the generic in a given name, which may not be wholly reliable. However, in the names collected here, it is possible to establish which elements have which function in a reasonable number of cases: adjectives when paired with nouns are almost certainly determiners; forms X y(r) Y (‘the X of the Y’) must for syntactic reasons also be cases of generic + determiner; and sometimes the semantics of the words are otherwise informative, as when one element of a place-name is unlikely to denote a topographic feature and so must be a determiner (as in Mosspeeble, which must be ‘field of the tents’ and not ‘the tents of the field’). (In theory, if the second element of a name is both masculine and shows lenition then it must also be the generic, but there is no clear case of this in the dataset; lack of lenition, meanwhile, cannot be relied on.) As the following table shows, both earlier and later names are attested:

ANALYSIS OF NAME DISTRIBUTION

Overview

An overview of the distribution of all the names accepted as p-Celtic in this study shows that the number and density of these names increases the further north one looks. This is probably partly an artefact of the more intensive collecting of names north of the border, but nonetheless seems likely also to reflect linguistic patterns during the early medieval period. Generally speaking, south of the Tweed valley, all the names are further inland (and correspondingly often on ground higher than about 200 metres). The Tweed valley itself is more complex: there are clusters of p-Celtic names on the higher ground bordering it, on both the north and south sides, but there are also a few names in the valley itself. North of the Tweed lie the two hill-ranges of Moorfoot and Lammermuir. Among the Moorfoots, higher and steeper than the Lammermuirs, and further west, lies the larger group of p-Celtic names, situated along narrow river valleys. The Lammermuirs also appear to have been significant, however: the largest concentration of p-Celtic names occurs to the north of the two hill-ranges. These survive in large numbers on the narrow lowland strip between the hill-ranges and the Forth, and are most obviously (but not exclusively) compatible with the idea of a British polity in this area, surviving the Anglian linguistic and cultural expansion longer than its neighbours to the South. There could be many reasons for this survival—the defence offered by the hill-ranges, for example, or the abundance of fertile land in the Tweed flood-plain, which would sustain a large number of Anglian settlers who would have no cause to look elsewhere for land (for the medieval landscape, see Higham 1993, 83; cf. also 1–9, 34–38). Whatever the reason, our data suggests that p-Celtic thrived longer in this area than in any other region of this study.

Names Containing the Brittonic Elements Tref, ‘Farmstead’, and Caer,‘Fort’.

These names both denote settlements, and their distributions are similar. Both elements appear to denote settlements of some importance (in caer’s case, the meaning implies fortification; for tref see Padel 1985, s.v. tre). Settlement-names have been argued to be less common in Brittonic than in Old English—settlements being named instead after topographical features—and moreover are thought not to survive very often south of the modern border (Nicolaisen 1976, 162; Hough 2004, esp. 27–29). My distribution map confirms this to some extent—only three or four of these names are found south of the Tweed valley, each on ground above 200 metres within about twelve kilometres of Bellingham on the North Tyne, Bellingham being the most westerly (inga)hām-name (Carraw, Carrick, possibly Troughend and, depending on whether the localisation is correct, †Powtreuet). The rest of them occur in almost equal numbers in the lowland area south of the Forth, and in the high ground of the Moorfoot and Lammermuir hill-ranges. In fact, of the possible p-Celtic names which survive in these two hill-ranges, about half contain the elements tref or caer. Moreover, the absence of tref- and caer-names from the areas in which the p-Celtic distribution is less dense, combined with their most definite presence in the ‘border’ area of the Moorfoot and Lammermuir ranges and further north, might lead us to surmise that these were names which would usually be found on lower ground, perhaps the sort of names given to more important settlements, which would naturally occupy the best land and therefore might be early targets for new settlers. That the tref-names in the ‘border’ area of hill-ranges north of the Tweed almost exclusively occupy the same area as the caer-names between the two ranges, or are found upstream of a further caer-name in a narrow valley, supports the hypothesis that these names have survived here because they had some organised protection. This militates against Nicolaisen’s statement (1976, 161–62) that the name caer in the north did not denote a fort, as it did in Wales, but nothing more impressive than a fortified farmhouse (as with Cornish ker): I am inclined to think that caer denoted something more substantial than that. This distribution of tref- and caer-names is identical to that of the nine massive early medieval silver chains found south of the Forth (included in the interactive map on the basis of Breeze 1998), partly on the basis of which a polity of ‘Gododdin’ has been posited stretching from the Forth to the Moorfoot and Lammermuir hills (Dark 2000, 215). The new evidence fits in well with this theory.

Names containing the Old English Elements -hām and -ingahām

So far we have established that p-Celtic names most often appear on the coastal plain to the north of the Lammermuir and Moorfoot hills, and on the edges of high ground or far up narrow valleys within high ground. The former fact immediately refutes the idea that Brittonic people tended to settle in such infertile, but defensible, territories simply by choice, as some commentators suggest (e.g. Russell 1992). However, the model implicit in the discussion so far—that the p-Celtic names which we know of are likely to have survived because of the comparatively late arrival of Old English speakers in the area—has yet to be confirmed. After all, it is clear that English arrived everywhere eventually. One way to check that this chronological model is likely to be correct is to plot some of the earlier Old English place-names, and discover if the distribution patterns are mutually exclusive. If so, this would imply that Old English did indeed reach the areas which are devoid of p-Celtic names first. If not, the hypothesis must be reconsidered.

The earliest Old English place-names were long thought to be those which contain the element -ingās. However, in more recent years it has become accepted that these names are not as old as the habitative names containing the elements -hām ‘homestead’ and -ingahām ‘homestead of the group X’. Moreover, the oldest place-names in Old English are now considered to be those which contain purely topographical elements (Gelling 1978, 106—29). However, as these could also have been coined at any time during the Old English period, these names are less useful than the names with suffixes in -hām and -ingahām, which seem to be distinctively early (the former being taken to be the earlier: Hough 1997, 150). This hypothesis is founded on southern English evidence, and we must be aware of the possibility of different naming patterns in the North, but it is reasonable to accept it as a starting point. The map plots all the names containing the elements -hām and -ingahām which are listed in Nicolaisen (1976, 72, 76) and Mawer (1920, 231).

It will be apparent that to a large extent the distributions of the earliest Old English names and the p-Celtic names are mutually exclusive, particularly in the southern part of the area studied. The lowland between the hills and the east coast is relatively crowded with these -hām and -ingahām names, which lie particularly along rivers. Moreover, the distribution of these names correlates well with the two heartlands of Bernician Anglo-Saxon power established by Rollason on other (archaeological and historical) evidence (2003, 48–52). Seven such names lie along the Tyne alone, while the single p-Celtic name that also lies on this river below the confluence of the South and North Tyne is Corbridge, which was a Roman encampment which may well have been known to the Anglians before they arrived in the area and thus have been preserved by them, in the usual way that Roman names have been preserved, that is, by the affixation of the first syllable of the Roman name Corstopitum to an Old English element, in this case brycg (cf. Aunchester, Binchester and Ebchester). In fact, there are -hām and -ingahām names (prominently Hexham, Ealingham, and Bellingham) lying along this river well to the west of several p-Celtic names (such as Cocklaw, Kirkley and Corbridge), implying that the first route for the settlers was along river-valleys. This has previously been shown to be the case for southern England (e.g. Myres 1986, 101–03). One river in our area, the Wear, seems to have been less of a route for cultural influence however, for it shows only one of the Old English names (Wolsingham) and four p-Celtic ones (Bishop Aukland, Binchester, Chester-le-Street and Penshaw/Painshaw). Moreover, Wolsingham is far upstream of the p-Celtic names. This appears to be an anomalous distribution, as it is not repeated elsewhere in the study.

The Tweed valley is also liberally scattered with -hām and -ingahām names, in this case mainly -hām names. There is some overlap here: from Melrose westwards, the area is characterised only by p-Celtic names, but the p-Celtic names Mindrum and Primside lie near the Old English names Downham and Yetholm, all along the river Glen, a tributary of the Till (itself a tributary of the Tweed). Primside lies further up the valley than the other settlements, beyond what is now a loch and a marsh, and on the edge of the slopes. Mindrum, meanwhile, seems to mean ‘mountain back’, a name that is singularly unsuited to its situation in a river valley. It may have been moved from the nearby cliff-tops 130 metres above the valley bottom. Printonan is a relatively easterly p-Celtic name, so it may be of interest that its second element seems to mean ‘marsh’, perhaps denoting less attractive ground. Another group of names can be seen as a cluster: Peniel Heugh, which can be hypothesised to have been a centre of power (Proudfoot–Aliaga-Kelly 1997, 34–36), Ancrum about five kilometres to the south-west, and Pirnie a few kilometres to the north—though the -hām name Smailholm lies only a few kilometres north again.

Strikingly, apart from a single name at the mouth of the Forth (Tyninghame), no -hām and -ingahām names appear north of the Moorfoot and Lammermuir hills, already noted as an important dividing line in this study. Names containing tref and caer are therefore almost entirely mutually exclusive of -hām and -ingahām names. This supports the idea that tref and caer names were likely to belong to settlements with the best land, and therefore rarely survive where the Old English first spread.

The valleys that run deep into the southern Scottish uplands seem to be prominent areas for Britonnic place-names, while the Cheviots, whose valleys are narrower, seem to have been less so. In Northumberland, where there was plenty of low lying coastal land, land above 200 metres seems to have remained Brittonic-speaking to some extent, but in the Scottish borders, where the average height of the land is higher, the Anglian cultural influence seems to have followed river valleys further into the uplands to settlements like Yetholm. Bellingham and Ealingham (both of the -ingahām type, which is considered to be slightly later than the -hām type) illustrate the river-following model to the best extent—they are further west than most of the Northumbrian -hām and -ingahām names, along the broad valley of the North Tyne. I have not considered any later strata of Old English names, although such a study would probably yield some interesting results.

CONCLUSION

Place-name evidence has long been seen to support a model of the spread of Anglo-Saxon culture which involved the Brittonic-speaking population being replaced violently by migrating Germanic-speakers; the dearth of p-Celtic place-names in England has been considered a major obstacle to inferring that Anglo-Saxon culture spread through other mechanisms. The evidence adduced here shows that in Bernicia at least there was more survival of p-Celtic toponymy than has hitherto been assumed. However, it also suggests that in some areas within Bernicia, p-Celtic names were more likely to be erased. We have, then, reasonably clear evidence in the region for different mechanisms of settlement and cultural change. Several names in this study show a gradual Anglicisation (or, in the case of Bishop Aukland, Norsification), partly by folk-etymology (such as Aberlady, Bathgate, Binchester and Milfield), which is consistent with this. Another possibility is that discontinuity of settlement meant that in some areas new names were coined in Old English, disguising the adoption of local nomenclature when settlements remained stable. A third possibility, argued for by Hough, is the systematic translation of place-names from Brittonic into Old English—a process which could in theory involve extensive cultural contact and communication, but result ultimately largely in place-names in one language (2004). Hough herself offered few concrete examples of this process, but the present study has identified plausible examples of Brittonic place-names competing with English place-names of identical meaning (Cockpen, vs. Redheugh; possibly Carrum, vs. Wark), suggesting that to some extent Hough must be correct.

The most obvious interpretation of the evidence in this study is, however, a synthesis of mass-migration and elite-takeover models. Large-scale Anglian cultural influence, and therefore implicitly settlement, seems likely in the river valleys. Following this, an Anglian power base would have been established, which would enable the political domination of the uplands without further wholesale resettlement of Anglians. This synthesis would help to explain the distribution patterns seen in this study. Either way, the combined distributions of place-names in the interactive map may be considered to provide a snapshot of the advance of Anglian cultural influence in the North. It suggests that the most likely model for this advance is first spreading along the river valleys of the major rivers of the Tyne, Tees, Alne and Tweed, and over the plain of the East coast. It seems plausible that p-Celtic speech survived for longer in the upland areas of the Pennines, Cheviots, Moorfoots and Lammermuirs, surviving perhaps longest of all in the area to the north of these latter two upland areas, where archaeological evidence supports the idea of an independent British kingdom surviving for a time after the Anglicisation of the population further south. In the past, scholars have tended to posit a ‘fan-shaped’ model for the spread of Anglian cultural and linguistic dominance in the North, whereby influence spreads outwards from a single point (for a classic example, see Higham 1986, 254). However, the distribution maps in this study would seem to suggest a more diffuse model, whereby influence travels mainly by water, along the coast and up major river valleys. This mirrors developments in southern England.

P-Celtic place-names in north-east England and south-east Scotland have been assumed to have been rare, and implicitly of little value to the historian. On the contrary, the dataset is rich, and when plotted against topography and early English names—and when plotted without regard for modern national borders—casts important new light on one of the most obscure periods in the area’s history. The extension of the survey area, comparison with other archaeological or onomastic distributions, or more detailed examinations of place-name distribution at a local level are all approaches invited by these findings. Whatever approaches are taken, it is hoped that this study has also emphasised the value of publishing them in electronic formats, and using these to maximise the accessibility and flexibility of the data gathered.

Notes

1 I would like to thank Dr Oliver Padel for his help and supervision, and for teaching me Middle Welsh; Dr Carole Hough and Dr Simon Taylor for their suggestions and helpfulness; and Dr Alaric Hall for his invaluable assistance in preparing this article for publication, and for his unwavering support. The web-version of the interactive map was implemented by Alaric using ArcGIS software. The original files can be obtained from him directly, at alaric@cantab.net.

2 Marking names which are merely possible examples with a question mark (?), Caerketton Hill, Caerlanrig, ?Cairndinnis, Carcant, Cardrona, Carfrae (East Lothian), Carfrae (Scottish Borders), ?Carham, ?Carkettyl, ?Carlowrie, ?Carraw, Carriber, Carrick, Carriden, Cramond. ?Halltree, Longniddry, Niddrie, Niddry, Ochiltree, Powtreuet, Soutra, Tartraven, ?Torfichen Hill, Torquhan, Trabroun, Trabrown, Trahenna, Tranent, Traprain, Traquair, Treverlen, ?Troughend.

3 For the purposes of this survey, I have not attempted to verify the data of my secondary sources, so I can give no guarantee of the accuracy of references to them. References to primary sources are reproduced exactly as found in the secondary material, save only the standardisation of references used by two different secondary sources, even when their information is deficient by the standards of the rest of this study. In some cases secondary sources do not give the primary source from which they have obtained early attestations. I have tried to avoid using these attestations where possible, but in cases where there are no other attestations available, they have been included. In other cases secondary sources give an abbreviated reference to a primary source which is not listed in their list of abbreviations. In these cases I have included the abbreviated reference.

4 ?Barnbougle, ?Pirncader, Pirnta(i)ton, ?Plenderleith, Primrose (Midlothian), Primrose (Scottish Borders), Primside, Printonan; cf. Pirn (near Fountainhall), Pirn (near Innerleithen), Pirnie, Pirny, Traprain.

References

Barrow, G. W. S. 1959. ‘Treverlen, Duddingston and Arthur’s Seat’, Book of the Old Edinburgh Club, 30: 1–9

Breeze, Andrew. 1998. ‘Pictish Chains and Welsh Forgeries’, Proceedings of the Scottish Antiquarian Society, 128: 481-84 (accessed from http://ads.ahds.ac.uk/catalogue/adsdata/PSAS_2002/pdf/vol_128/128_481_484.pdf)

—— 1999. ‘Some Celtic Place-Names of Scotland including Dalriada, Kincarden, Abercorn, Coldingham and Girvan’, Scottish Language, 18: 34–51

—— 2000. ‘The Names of Bellshill, Carmichael, Lauder and Soutra’, The Innes Review, 51 (2000), 72-79

—— 2002a. ‘The Battle of Alutthèlia in 844 and Bishop Auckland’, Northern History, 39 (2002), 124–25

—— 2002b. ‘Pennango Near Hawick and Welsh Angae "Death" ’, Northern History, 39 (2002), 126

—— 2002c. ‘The Kingdom and Name of Elmet’, Northern History, 39 (2002), 157–71

—— 2002d. ‘The Celtic Names of Blencow and Blenkinsopp’, Northern History, 39 (2002), 291–92

Bullough, Donald A. ‘The Place-Name Hexham and Its Interpretation’, Notes & Queries 46 (1999), 422-27.

Campbell, A. 1959. Old English Grammar (Oxford: Clarendon Press)

Coates, Richard. 2000a. ‘Un-English Reflections on Lindisfarne’, in Celtic Voices, English Places: Studies of the Celtic Impact on Place-Names in Britain, by Richard Coates and Andrew Breeze; with a contribution by David Horovitz (Stamford: Tyas), pp. 241–59

—— 2000b. ‘Gazeteer of Celtic Names in England (except Cornwall)’, in Celtic Voices, English Places: Studies of the Celtic Impact on Place-Names in Britain, by Richard Coates and Andrew Breeze; with a contribution by David Horovitz (Stamford: Tyas), pp. 263–345

—— 2004. ‘Invisible Britons: The View from Linguistics’, paper circulated in connection with the conference Britons and Saxons, 14–16 April 2004; accessed from http://www.sussex.ac.uk/linguistics/documents/rc_britons.pdf, 18 April 2006.

Cubbin, G. P. 1981–82. ‘A Celtic Dialect Line in North-West England’, Studia Celtica, 16: 175–82

Dark, K. R. 2000. Britain and the End of the Roman Empire (Stroud: Tempus)

Dixon, Norman. 1947. ‘The Place-Names of Midlothian’ (unpublished doctoral thesis, University of Edinburgh, 1947)

Dumville, DAvid. 1994. ‘The Eastern Terminus of the Antonine Wall: 12th- or 13th-Century Evidence’, Proceedings of the Society of Antiquaries of Scotland 124: 293-98, accessed from http://ads.ahds.ac.uk/catalogue/adsdata/PSAS_2002/pdf/vol_124/124_293_298.pdf

Ekwall, Eilert.1960. The Concise Oxford Dictionary of English Place-Names, 4th edn (Oxford: Clarendon)

Filppula, Markku, Juhani Klemola and Heli Pitkänen (eds), The Celtic Roots of English, Studies in Languages, 37 (Joensuu: University of Joensuu, Faculty of Humanities, 2002)

Geiriadur Prifysgol Cymru/A Dictionary of the Welsh Language. 1950–2003. (Cardiff: Gwasg Prifysgol Cymru)

Gelling, Margaret. 1978. Signposts to the Past: Place-names and the History of England (London: Dent)

Higham, N. J. 1986. The Northern Counties to AD 1000 (London: Longman)

——1993. The Kingdom of Northumbria AD 350-1100 (Stroud: Sutton)

Hind, J. G. F. 1980. ‘The Romano-British Name for Corbridge’, Britannia, 11: 165–71

Hough, Carole. 1997. ‘The Earliest Old English Place-Names in Scotland’, Notes and Queries, 242: 148–50

Hough, Carole. 2001. ‘P-Celtic Tref in Scottish Place-Names’, Notes and Queries, 48: 213–15

Hough, Carole. 2004. ‘The (Non?)-Survival of Romano-British Toponymy’, Neuphilologische Mitteilungen, 105: 25–32

Hull, J. E. 1939. ‘Some Notes on Place-Names of the Three Northern Counties’, in The Three Northern Counties of England, ed. by Cuthbert Headlam (Gateshead: Northumberland Press), pp. 15–33

Jackson, Kenneth 1953. Language and History in Early Britain (Edinburgh: Edinburgh University Press: Oliver & Boyd)

Johnston, James B. 1934. Place-names of Scotland, 3rd edn, (London: J. Murray)

Koch, John Thomas. 1997. The ‘Gododdin’ of Aneirin: Text and Context from Dark-Age North Britain (Cardiff: University of Wales Press)

Lucy, Sam. 1999. ‘Changing Burial Rites in Northumbria AD 500–700’, in Northumbria’s Golden Age, ed. by Jane Hawkes and Susan Mills (Stroud: Sutton), pp. 12–43

Macdonald, Angus. 1941. The Place-Names of West Lothian (Edinburgh: Oliver & Boyd)

Macdonald, J. S. M. 1991. The Place-Names of Roxburghshire (Hawick: Hawick Archaeological Society)

Mackenzie, W. C. 1931. Scottish Place-names (London: K. Paul, Trench, Trubner)

Matley, I. M. 1965. ‘Elements of Celtic Place-Names’, Names, 13: 39–54

Mawer, Allen. 1920. The Place Names of Northumberland and Durham (Cambridge: Cambridge University Press)

——1929. ‘The River-Names of Northumbria and Durham’, Archaeologia Aeliana, 4th Series 6: 175–82

Myres, J. N. L. 1986. The English Settlements, The Oxford History of England, 1b (Oxford: Clarendon Press)

Nicolaisen, W. F. H. 1976. Scottish Place-names: Their Study and Significance (London: Batsford)

Padel, O. J. 1985. Cornish Place-Name Elements, English Place-Name Society, 56/57 (Nottingham: English Place-Name Society)

Parsons, David and Tania Styles (eds), The Vocabulary of English Place-Names (Nottingham: Centre for English Name Studies, 1997)

Proudfoot, Edwina, and Christopher Aliaga-Kelly. 1997. ‘Aspects of Settlement and Territorial Arrangements in South-East Scotland in the late Prehistoric and Early Medieval Periods’, Medieval Archaeology, 41: 33–50

Rivet, A. L. F. and Colin Smith. 1979. The Place-Names of Roman Britain (London: Batsford)

Robb, John G. 1996. ‘Toponymy in Lowland Scotland: Depictions of Linguistic Heritage’, Scottish Geographical Magazine, 112: 169–76

Rollason, David. 2003. Northumbria 500–1100: Creation and Destruction of a Kingdom (Cambridge: Cambridge University Press)

Russell, Pamela B. 1992. ‘Place-Name Evidence for the Survival of British Settlement in the West Derby Hundred (Lancashire) after the Anglian Invasions’, Northern History, 28: 25–41

Scott, Douglas, A Hawick Word Book (prepublication text 2006), http://www.astro.ubc.ca/people/scott/book.pdf, accessed 18th April 2006.

Sims-Williams, Patrick. 2003. The Celtic Inscriptions of Britain: Phonology and Chronology, c. 400–1200, Publications of the Philological Society, 37 (Oxford: Blackwell)

Smith, A. H. 1956. English Place-Name Elements, English Place-Name Society, 25–26, 2 vols (Cambridge: Cambridge University Press)

Smith, C. C. 1980. ‘The Survival of British Toponymy’, Nomina, 4: 27-41

Thomas, Mark G., Michael P. H. Stumpf and Heinrich Härke. 2006. ‘Evidence for an Apartheid-like Social Structure in Early Anglo-Saxon England’, Proceedings of the Royal Society B: Biological Sciences, doi:10.1098/rspb.2006.3627

Watson, William J. 1926. The History of the Celtic Place-names of Scotland (Edinburgh: The Royal Celtic Society; repr. Birlinn, 1993)

Watson, Godfrey. 1995. Northumberland Place-Names: Goodwife Hot and Others (Morpeth: Sandhill)

Watts, Victor. 2004. The Cambridge Dictionary of English Place-Names: Based on the Collections of the English Place-Name Society (Cambridge: Cambridge University Press)

Williams, Hugh (ed.). 1899. Gildae de Excidio Brittaniae, Fragmenta, Liber de Paenitentia, Accedit et Lorica Gildae/Gildas: The Ruin of Britain, Fragments from Lost Letters, the Penitential, Together with the Lorica of Gildas, Cymmrodorion Record Series, 3 (London: Nutt)

Woolf, Alex, ‘The Britons: From Romans to Barbarians’, in Regna and Gentes: The Relationship between Late Antique and Early Medieval Peoples and Kingdoms in the Transformation of the Roman World, ed. by Hans-Werner Goetz, Jörg Jarnut and Walter Pohl, The Transformation of the Roman World, 13 (Leiden: Brill, 2003), pp. 345–80