Appendix: Discussion of Individual Names

This is from a preprint of an article whose definitive and final version is available free-access as The Heroic Age, 10 (2007),

Abercorn (West Lothian: N55:59:27 W3:28:06; 308500 678500). Probable.
Watson gave the etymology of this name as aber + corniog, ‘horned confluence’ (1926, 461), followed by Macdonald (1941, 105). Breeze, however, noting that Abercorn lies on a stream called the Cornie, more convincingly derived the name from aber + curnig, ‘mouth of the Cornie stream’ (1999, 41–43). Meanwhile, he gives curn as meaning ‘cone-shaped hill’, noting that the settlement lies on a hill of this type (cf. Padel’s reconstruction of a Cornish *kernan ‘mound, hill’, corresponding to the Welsh place-name element curnen, 1985, s.v. *kernan), taking curnig as an adjectival derivative of this. Thus the stream would take its name from the hill and the settlement from the stream: ultimately, ‘the mouth of the river of the cone-shaped hill’.

Aberlady (East Lothian: N56:00:20 W2:1:34; 346500 679500). Probable.
Watson thought that the form Aberlessic in the earliest attestation did not refer to Aberlady (first attested in that form c. 1221 as Aberlauedy; 1926, 460); subsequent commentators, however, have accepted the identification. Mackenzie thought that aber here had the meaning of ‘port’ (1931, 105), while Johnston said that aber here is the Gaelic word abar, meaning ‘marsh’ (1934, 77). However, given the location of the site on an estuary, a p-Celtic derivation seems more likely. Nicolaisen quotes this as a p-Celtic name with first element aber (1976, 164). As Mackenzie argued, the form of the second element of Aberlady must owe something to Old English hlæfdig (‘lady’). But it is most unlikely that this—which for Old English-speakers would have been the generic element—originates ultimately in hlæfdig: it is more likely a folk-etymologisation of the earlier second element represented by -lessic, which as Watson argued seems most likely to be equivalent to Welsh llusog ‘bushy’ (cf. Padel 1985, s.v. les).

Amble (Northumberland: N55:19:29 W1:34:05; 427500 603500). Unlikely.
Mawer argued that this is probably Celtic, citing similar names elsewhere (1920, 5). Padel discussed a similar river-name in Cornwall, where Amble is the Cornish element *amal, equivalent to Welsh ymyl, ‘edge’ (Padel 1985, s.v. *amal). He found that the second vowel might have been expected to be e rather than a, thus *amel, and a plural *emla, which could thus give *ambla through a process of dissimilation. However, early spellings for one of the Cornish names insist on *emla or *emle, whereas the early spellings for the Northumbrian name demand a-. Moreover, Ekwall and Watts prefer Old English Anna (personal name) + bile ‘promontory’ (Ekwall 1960, 9; Watts 2002, s.v. AMBLE-BY-THE-SEA). The Old English derivation seems therefore more likely.

Ancrum (Scottish Borders: N55:30:46 W2:35:43; 362500 624500). Possible.
Watson, followed by later commentators, gave this as the river-name Alne + p-Celtic crwm or Gaelic crom, the whole meaning ‘bend of the river Alne’ (Watson, 1926, 467; Nicolaisen 1976, 172; Mackenzie 1931, 114; Hull 1939, 18). The early attestations already show variation between -crum- and -crom-.

Auckland, Bishop (Durham: N54:39:18 W1:40:44; 420815 528930). Probable.
Both Mawer (1920, 7) and Ekwall (1960, 18–19) agreed that this name was Celtic, on the basis of its early form Alclit; it was then Scandinavianised on the basis of auk-land ‘additional land’ (Breeze 2002a, 124). It may have been transferred from Dumbarton, whose p-Celtic name is also Alclut, or be a parallel formation, ‘rock of the Clyde’, the Clyde in question being now the river Gaunless (Breeze 2002a, 125; Watts 2004, s.v. Bishop AUCKLAND). The name seems also to be attested in an annal for 844 by Roger of Wendover, mentioning Alutthèlia, interpreted by Breeze as Alclut + Old English þelu ‘planks, plank bridge’ (2002a, 125). Either way, a p-Celtic etymology is clear.

†Aunchester, Anterchesters (Northumberland: unlocated). Possible.
Mawer said that this was a Celtic name, linking it with various continental names such as Antros, Antrum and Antricinum (1920, 7; cf. Hull 1939, 19). The name would in that case follow the general pattern of English borrowing of Romano-British names, with the first element consisting of the first syllable of the original name, and the second an English element, often ceaster, as here. However, in the absence of a clear etymon, it impossible to be sure of a p-Celtic origin.

Barnbougle (City of Edinburgh: grif ref. N55:59:32 W3:20:25; 316500 678500). Probable.
Watson gave this as pren + bugail, ‘herdsman’s tree’ (1926, 351), followed by Nicholaisen (1976, 166). Macdonald (1941, 4–5) and Padel (1985, s.v. bugel), preferred bryn + bugail, ‘herdsman’s hill’, noting that bryn is masculine and therefore we would not need to explain the lack of lenition of /b/ in this case. Mackenzie, adducing a wide range of early forms, suggested a hybrid Gaelic-Cumbric name barr + bugail (1931, 74). But given that consonants tend to be more stable than vowels, the persistent n in attestations seems more noteworthy than the a in the first syllable. Although bugail seems not otherwise to be attested in English place-names, no more likely candidate is forthcoming.

Bathgate (West Lothian: N55:54:28 W3:38:28; 297500 669500). Probable.
Watson (1926, 381–82), followed by Macdonald (1941, 80–81) and Nicholaisen (1976,172), gave this as possibly deriving from baedd + coed, ‘boar wood’. Mackenzie considered it Gaelic (1931, 77), which would account for the non-lenited /c/ in the early attestations (the earliest being Bathchet c. 1160 and Batket 1153-65)—but this is a normal sound-substitution for names passing from Brittonic to Old English (Jackson 1953, 195). The name may also be attested in line 24 of the putatively late sixth-century poem in praise of Gwallog, attributed to Taliesin and surviving in the thirteenth-century Book of Taliesin, as coed beit, but this is uncertain (Breeze 2002c, 169).

Binchester (Durham: N54:40:41 W1:41:01; 420500 531500). Possible.
Mawer gave this as Romano-Celtic Vinovia + ceaster (1920, 22); Ekwall preferred to etymologise the first element as Old English binn, ‘manger’, the old fort possibly having been later used to shelter cattle (1960, 43). Jackson, combining their readings, argued that [bin] cannot be directly from Romano-British [vin], and derived it ultimately from Vinovia, positing influence from Old English binn (itself a loan from p-Celtic; 1953, 89 and 260). This is certainly plausible, but it may be that a phonological identification did take place. Old English had no [v-] (Campbell 1959, §50.1). Anglo-Saxons were later to substitute [f-], in loan-words (Campbell 1959, §539), but an earlier sound-substitution of b is not unlikely, because in early Old English, /b/ still had the medial and final allophone [β] (Campbell 1959, §§55–57), which would have encouraged the identification of /b/ with [v] (cf. Smith 1980, 31–32). At any rate, given the structure of the name, corresponding with other Romano-British names partially borrowed into Old English with the addition of an Old English word (cf. Aunchester, above) I am inclined to accept a p-Celtic derivation.

Blenkinsopp (Northumberland: N54:58:27 W2:29:37; 368500 564500). Probable.
This name has recently been studied by Breeze, who dismissed the suggestion that it may contain a personal name (*)Blenkin to argue instead for blaen ‘top’ + kein ‘back, ridge’ + Old English hop ‘valley’ (2002d, 292).

Caerketton Hill (Midlothian: N55:53:08 W3:13:28; 323500 666500). Probable.
Watson derived this from caer + the personal name Catell (cf. Carkettyll, below; 1926, 369). Given the form of the early attestation—Karynketil, 1317—it may instead contain p-Celtic carn, ‘cairn’. Either can be supported archaeologically—according to the hill exhibits a prehistoric fort and cairn—but either way it seems likely to be a Brittonic name.

Caerlanrig/Carlenrig (Roxburghshire: N55:19:51 W2:57:19; 339500 604500). Probable.
Watson, followed by Nicholaisen, gave this as caer + llanerch, ‘clearing’ (with scribal confusion between n and u~v in early attestations, including the first, Carlaverik, 1511), adding that there seems to have been a fort at this site (Watson, 1926, 368; Nicolaisen 1976, 162; for the second element cf. Padel 1985, s.v. lanherch). Macdonald agrees that there was a fort, but argues that in fact we have caer + Old English lang ‘long’ and hrycg ‘ridge’, since the site exhibits a long ridge and a poor situation for growing trees (1991, 6, 37). If her interpretation is correct, then the name is an interesting hybrid form; either way, however, the initial caer- seems beyond doubt.

Cairndinnis (East Lothian: N55:57:42 W2:40:56; 357500 674500). Possible.
Watson translated this as ‘cairn of the fortress’, implying carn + dinas, and pointed out that the dinis form is common in Cornwall as well as appearing in several other Scottish names (Watson, 1926, 372; cf. Padel 1985, s.vv. carn, *dynas). This name is not mentioned in any of my other sources, and Watson gave no early attestations for it. According to, it, and two nearby prehistoric enclosures, lie less than a kilometre to the east of Traprain Law.

Cambois (Northumberland: N55:08:42 W1:31:23; 430500 583500). Probable.
Mawer said that Cambois, attested first around 1150 as Kambus, may be Celtic, connecting it with Gaelic and Irish camus (meaning cambas), ‘bay, creek’ (1920, 38) and arguing that the final syllable has been influenced by Anglo-Norman bois, ‘wood’. Certainly no English or Norse etymon seems likely. A Gaelic derivation, however, seems unlikely given the southerly location of the site. It seems more likely that it would be derived from the Brittonic cognate of the Gaelic word, camas, cemais, ‘bend in river, bay’ (for which see Padel 1985, s.v. *camas; cf. Coates 2000b, 323). This suits the location of the site at the confluence of Sleek Burn and the River Blyth, and accordingly I have reckoned the name as probably p-Celtic.

Cammo (City of Edinburgh: N55:57:23 W3:19:22; 317500 674500). Possible.
Watson and Mackenzie etymologized this name, attested first in 1296 both as Cambok and Cambo, as Gaelic meaning ‘crooked’ (cf. Cambois above; Watson, 1926, 143; Mackenzie 1931, 114). One might hope, however, to see clearer evidence of the expected -ach ending if this were the case. Dixon, however, gave this as Brittonic with the same meaning. The forms in -ok would then be from Brittonic *cambāco- (with a compositional suffix; Dixon 1947, 158; cf. Padel 1985, s.vv. cam, *camas).

Carcant (Scottish Borders: N55:45:42 W3:00:49; 336500 652500). Probable.
Watson gave this as caer + can, ‘white fort’, or caer + cant, explaining cant as an orb or the rim of a circle—the place being the circular opening of a cul-de-sac (1926, 369; cf. Dixon 1947, 197). Padel suggested the meanings ‘district, region’, ‘edge, border’, or possibly ‘throng, host troop’ for *cant (1985, s.v.). Either way, a p-Celtic etymology seems clear for both elements.

Cardrona (Scottish Borders: N55:38:39 W3:06:21; 330500 639500). Probable.
Watson, noting the earliest attestation Cardronow (c. 1500) gives this as caer + tronau, meaning ‘fort of the thrones’ (1926, 369). However, tronau is an English word whose earliest attestation in Welsh is in 1545 (Geiriadur Prifysgol Cymry, s.v.) and this derivation is therefore unacceptable. An alternative possibility is perhaps trwyn, ‘nose’, attested in Welsh and Cornish place-names in the sense ‘promontory’ (Padel 1985, s.v. tron), but I have found no examples using the plural form (trywnau), which is demanded here. A p-Celtic derivation for the first element at least, however, seems clear.

Carfrae (East Lothian: N55:55:00 W2:40:54; 357500 669500). Probable.
Watson gives this as caer + bre (‘hill’; 1926, 369; cf. Padel 1985, s.v. *bre). He gives no early attestations, but Nicolaisen agrees with this etymology (1976, 162), and no other interpretation seems likely.

Carfrae (Scottish Borders: N55:47:25 W2:47:27; 350500 655500). Probable.
See Carfrae, above. Watson says there is a fort here (1926, 369), possibly on the basis of the Ordnance Survey map which also marks one.

Carham On Tweed (Northumberland: N55:38:23 W2:19:38; 379500 638500). Possible.
Mawer gives this as either Old English carr, ‘rock’, + hām, ‘village’, or Old English dative plural carrum (1920, 39). Nicholaisen accepted the -hām etymology (1975, 72), while Watts preferred carrum (giving ‘(The settlement) at the rocks’, 2004, s.v. CARHAM). But Mawer cites Richard of Hexham as saying ‘Carrum, quod ab Anglis Werch dicitur’ suggesting that Carham was not the Anglian name for this settlement (1920, 39). Werch is now Wark, about five kilometres down the River Tweed and, as Mawer points out, must derive from Old English (ge)weorc ‘earthworks, fortifications’ (1920, 207). This raises the possibility that Carham in fact contains caer, and that Werch is a partial translation of the p-Celtic name. Either way, Richard of Hexham’s testimony suggests that we may here have a p-Celtic name.

Kirkettle (Midlothian: N55:50:27 W3:10:03; 326985 661475). Possible.
Watson gives this as caer + the p-Celtic personal name Cadell (1926, 369).

Carlowrie (City of Edinburgh: N55:57:21 W3:22:15; 314500 674500). Possible.
Watson thinks this may contain caer, but says that there is no proof that this is so (1926, 370). Macdonald thinks that this solution is inadequate, as the first element could be any one of three Gaelic words as well as caer (1941, 5). Neither commentator gives an etymology for the second element, which would decide the language of the first.

Carnethy (Midlothian: N55:50:25 W3:16:16; 320500 661500). Probable.
Watson gives this hill name (a summit of the Pentlands) as possibly deriving from carneddau, ‘cairns’—he notes that the stress of the modern word falls on the penultimate syllable (1926, 369). Dixon agrees (1947, 103).

Carraw (Northumberland: N55:02:16 W2:14:39; 384500 571500). Probable.
Mawer gives this as a p-Celtic root cadro- (presumably meaning it as an ancestor of caer) with an unknown suffix, possibly similar to that found in Stranraer, saying it is ‘clearly not an English name’ (1920, 39–40). Ekwall has it as carrau, the plural of OBrit carr, ‘rock’, or as carr + Old English rāw ‘row’ (1960, 88). But early attestations of rāw are few (see Smith 1956, s.v.), encouraging the p-Celtic identification, which Coates accepts (2000b, 323).

Carriber (West Lothian: N55:57:41 W3:39:34; 296500 675500). Probable.
Macdonald has this as a hybrid of p-Celtic caer + Gaelic abar, ‘fort in or by the marsh’ (1941, 58). His etymology rests to some extent on the continued presence of marshy land near the site. However, he gives the modern pronunciation to be penultimately stressed, suggesting either a p-Celtic name or a Gaelic name-phrase, since Gaelic proper compounds are, as I have discussed above, initially stressed. Accordingly, it is more likely to contain the p-Celtic element aber, ‘river mouth’. Watson mentions the name (1926, 105), but gives no etymology for it. The co-ordinates given are for Easter Carriber.

Carrick (Northumberland: N55:15:45 W2:09:04; 390500 596500). Probable.
Mawer gives this as caer with Old English wīc (1920, 40; cf. discussion of hybrids, above); Nicholaisen accepts this (1976, 162), but Coates suggests gwig ‘wood’ (2000b, 324). The grid references given are for Carrick Heights, the only place with this name I have found in Northumberland.

Carriden (Falkirkshire: N56:00:59 W3:34:54; 301500 681500). Probable.
Watson (1926, 369–70), supported more recently by Dumville (1994, 296-97), thinks this is the Kair Eden mentioned in the Capitula prefixed to a thirteenth-century copy of Gildas’ De Excidio Brittaniae (ed. Williams 1899, 12); other early forms, at any rate, are similar. He takes the second element as a regional name Eidyn (cf. 1926, 340-42, again implicitly supported by Dumville). Mackenzie thinks this cannot be caer + eidyn because the stress falls on the first syllable—he gives it as caereden, ‘the wall’ (1931, 297). But this suggests that he thinks the final -en is a suffixed definite article, which is unattested in Brittonic. Macdonald thinks this reference is to Edinburgh, and claims that this name cannot be derived from caer + eden due to the stress falling on the first syllable (1941, 25–26). He notes the Old Statistical Account (1794), which gives the pronunciation as carrin, and says that the form must have been assimilated to p-Celtic cardden, 'thicket'. However, he is uncertain about this as a derivation for the name, and concludes that this derivation ‘must still remain in doubt’. Either way, it is hard to doubt that the first element is, in the early attestations, caer.

Carrycoats (Northumberland: N55:06:35 W2:07:09; 392500 579500). Unlikely. Mawer gives early forms such as Carricot and Carre Cottes and notes the possibility of a derivation from caer + y + coed. But Watts took the name to derive from the river-name Carry Burn and Old English cot ‘dwelling’, noting that although the name Carry Burn is first attested in 1866, the name is well-paralleled by older river-names and indeed apparently pre-Indo-European (2004, CARRYCOATS HALL). Watts’s interpretation is plausible, and although it would still attest to continuity of nomenclature, river-names are excluded from the present study. One way to salvage the caer y coed interpretation would be to argue that the burn was named from the settlement, but this does not seem likely.

Chester-le-Street (Durham: N54:51:27 W1:34:23; 427500 551500). Probable.
Mawer gives this as Romano-British Congavata + chester (< Old English ceaster; 1920, 43–44). However, Congavata is, according to Rivet and Smith, entirely Latin (‘hollows’), and is now identified with a place in Cumbria (1979, 315). As Watts emphasises, however, the early forms of Chester-le-Street, which focus on forms such as Cuncacestre, seem to derive from the Roman name Concangis (2004, s.v. CHESTER-LE-STREET). Although this is obscure, a p-Celtic identification for these forms (also suggested be Ekwall 1960, 101) would only be undermined if we discarded this identification, and further accepted Mawer’s suggestion that the first element might be a p-Celtic personal name rather than a place-name element. This does not seem overly likely.

Clackmae (Scottish Borders: N55:38:49 W2:41:34; 356500 639500). Possible.
Macdonald etymologised this tentatively as ‘the stone field’ (1991, 27), corresponding to clegr (cf. Padel 1985, s.vv. *cleger, *ma). This does not seem unlikely; cf. Watson’s handling of Clackmannan (1926, 103).

Cockerton (Darlington: N54:32:03 W1:34:36; 427500 515500). Unlikely.
Mawer gives no etymology for this, but cites Sephton as saying that it is of Celtic origin (1920, 49). If this is correct, however, then the cocker- element is a river-name and so excluded from this survey in any case.

Cocklaw (Northumberland: N55:33:00 W1:49:09; 411500 628500). Possible.
Mawer says that if his early forms (Creklawe, Crokelawe, both 1296) do indeed pertain to the modern Cocklaw then it must be derived from Celtic crug with Old English hlæw, both meaning ‘hill’ (1920, 49–50). However, Ekwall emphasised the form Coklau (1479) and derives the name instead from Old English cocc, ‘cock’ + hlāw (1960, 115). The derivation of this name therefore depends on the identification of the correct early forms. The map references are for Cocklaw Dean, the only instance of such a name I could find in this area (Cockle Park, also in the county and included in Ekwall’s entry, is a different pocation).

Cockpen (Midlothian: N55:51:36 W3:04:48; 332500 663500). Probable.
Watson gives this as coch + pen, ‘red head’, although he notes that the stress is on the final syllable (1926, 356; cf. Padel 1985, s.v. cough). Dixon agrees with the etymology and notes that the local name Redheugh, ‘red ridge or spur’ supports this derivation (1947, 143). This supporting evidence is also important as an example of variation between English and Brittonic place-name forms.

Corbridge on Tyne (Northumberland: N54:58:30 W2:00:34; 399500 564500). Probable.
The early attestations of this name show variation between Cor- and Col-, as in the earliest two forms, Corebricg and Colebruge, and there has been extensive debate about what its etymology may be (cf. Mawer 1920, 52–53; Ekwall 1960, 12; Watts 2004, s.v. CORBRIDGE). Some relationship with the Roman name Corstopitum seems clear, however; Rivet and Smith reconstruct the original name as Coriosopitum (1979, 322). Hind, meanwhile, reconstructs it as Brittonic *corio, ‘army’ + *ritum, ‘ford’ (1980, 169–71). A p-Celtic origin, at any rate, seems clear.

Corsick (East Lothian: N55:57:36 W2:26:37; 372400 674200). Possible.
Macdonald etymologised Corsick as Welsh cors, ‘bog’, and Old English wīc, ‘dwelling’ (1991, 35). A more enticing comparison is Welsh corsiog ‘marshy’ and its Brittonic cognates, either as a substantive noun or perhaps as a shortening of a name like the Cornish names Gersick-an-Awn and Pengersick (cf. Padel 1985, s.v. *cors). The attestations are too late for use to be sure of either interpretation, however. The co-ordinates are those of Corsick Hill, based on the Landranger reference given at

Craik (Scottish Borders: N55:21:58 W3:02:06; 334500 608500). Possible.
Macdonald etymologised this name as Old Welsh *creic ‘rock’, which is plausible (cf. Smith 1956, s.v. *creic; Padel 1985, s.v. crak1; Macdonald 1991, 33). With only late attestations, it could also be Gaelic creag, or the Scots reflex of these Celtic words, but the consistent -k militates somewhat in favour of the Welsh etymon.

Cramond (City of Edinburgh: N55:58:28 W3:18:27; 318500 676500). Probable.
On the basis of early forms such as Caramonth, Watson gives this as caer + the river-name almond (1926, 369). The modern form arises from the stress falling on the penultimate syllable. Later commentators have accepted the etymology (Mackenzie 1931, 71; Dixon 1947, 156; Nicolaisen 1976, 162).

Cribbielaw (Midlothian: unlocated). Possible.
Dixon gives the first element of this as Brittonic crib ‘crest’ (1947, 283). The later syllables of the word are hard to be sure of (crib + Old English bell ‘hill’ + Old English hlāw? crib + y + lle ‘place’?), but the Brittonic character of the first element is plausible.

Crichton (Midlothian: N55:51:06 W2:59:02; 338500 662500). Possible.
Dixon gives this as Brittonic craig + Old English tūn, ‘rock farm’ (1947, 169).

†Crudderland(s) (West Lothian: unlocated). Unlikely.
Macdonald gives this obsolete name as deriving from crwth + Old English land, ‘harper’s land(s)’ (1941, 20–21). However, crwth was borrowed into Old English, giving ME crowder, ‘player on a harp’ (Middle English Dictionary, s.v.), which seems the more likely derivation in this case.

Dalkeith (Mid Lothian: N55:53:46 W3:02:56; 334500 667500). Probable.
Watson gives this name as deriving from dal + coed, ‘wooded meadow or plateau’ (1926, 382). Mackenzie, Dixon and Nicolaisen agree, though noting an alternative translation ‘wooded valley’ (Mackenzie 1931, 221; Macdonald 1941, 162; Nicolaisen 1976, 172). For the sound change /-d/ > /-θ/ see Cubbin (1981–82).

Din Fell (Scottish Borders: N55:15:02 W2:51:32; 345500 595500). Possible.
Watson gives the first element of this as Brittonic din, ‘fort’ (1926, 372), but it seems to have no early attestations.

Dinley (Scottish Borders: N55:15:35 W2:49:39; 347500 596500). Possible.
Watson gives the first element of this as Brittonic din, ‘fort’ (1926, 372), but it seems to have no early attestations.

Dinmontlaire Knowe (Scottish Borders: N55:11:19 W2:42:58; 354500 588500). Possible.
Watson gives the first element as din and the second as (presumably Old English) mont, ‘hill’ (1926, 372), but it seems to have no early attestations. The -laire element is presumably one of several Older Scots words of this form (cf. Dictionary of the Older Scottish Tongue, s.v. Lair).

Dinsol (Midlothian: N55:49:32 W2:52:17; 345500 659500). Possible.
See discussion of

Dinwiddie (Scottish Borders: N55:10:06 W2:49:48; 347225 586315). Possible.
Watson gives the first element as din and places it in Ettleton Parish (1926, 372). The co-ordinates are for the site of Ettleton’s medieval parish church.

†Dunpelder (East Lothian: N55:57:42, W2:39:59; 358500 674500). Probable.
Watson interprets the second element as pelydr ‘spear shafts’, implicitly taking dun as a Gaelicisation of din. This seems likely. The name has since been replaced by Traprain Law.

Dreva on Tweed (Scottish Borders: N55:36:20 W3:21:31; 314500 635500). Possible.
This is first attested only in 1649, as Draway. Watson says this may be for (y) + tref + ma, ‘the tref place’ (1926, 363). It is hard to be certain of this, but tref may indeed be the first element (see Padel 1985, s.v. tre for discussion of lenition of initial /t/ in tref; cf. s.v. *ma).

Dunbar (East Lothian: N55:59:54 W2:31:22; 367500 678500). Probable.
Watson interprets Dunbar as Gaelic, meaning ‘summit-fort’, but adds that it is ‘probably taken over from British din-bar with the same meaning’ (1926, 141; for bar see Padel 1985, s.v.). He gives no early attestations, but seems to be thinking of the form Dynbaer in Stephanus’s Vita Sancti Wilfredi, which indeed appears to be p-Celtic.

Ebchester (Durham: N54:53:39 W1:50:16; 410500 555500). Unlikely.
Mawer gives this as meaning ‘fort of Ebbi’ (1920, 71), Ekwall offering the superior ‘Roman fort of Ebba’ (1960, 158; cf. Watts 2004, s.v. EBCHESTER). Hull gives it as ebach < Romano-British Epiacum + ceaster (1939, 25), but Rivet and Smith identify Epiacum instead with Whitley Castle, near Kirkhaugh, Northumberland (1979, 360), while Ebchester is near the Roman fort of Vindomora in Co. Durham. Thus the Old English derivation seems more likely.

Eccles (Scottish Borders: N55:39:59 W2:22:30; 376500 641500). Probable.
Watson says this may represent Brittonic eglwys or Gaelic eaglais (1926, 153). Nicolaisen gives it as eglwys, which seems much the more likely (1976, 172).

Ecclesmachan (N55:56:43 W3:30:53; 305500 673500). Possible.
Watson says that the Saint commemorated in this place name is St. Machan, who is said to have been a disciple of St. Cadoc of Llancarvan (1926, 151). Mackenzie derives it from Gaelic eaglais, as does Macdonald (Mackenzie 1931, 232; Macdonald 1941, 47–48). Nicolaisen gives the first element as eglwys, and notes that it cannot be older than the 7th century if Watson’s hypothesis about the second element is correct (1976, 172). The saint’s name, however, appears to be a Gaelic form of p-Celtic Mawgan, so demanding caution in accepting a p-Celtic etymology.

Castle Eden (Durham: N54:43:50 W1:20:30; 442500 537500). Unlikely.
Mawer says this may be derived from the river-name Eden, which is found elsewhere in Britain, or from the Gaelic word eudann or aodann, ‘forehead, hill brow’ (1920, 71). Ekwall says it is to be derived from Eden Burn, on which it is situated (1936, 153). He gives this latter name as from *ituna, ‘to gush forth’, and compares it with the river Eden in Cumbria. His 1960 edition does not mention the name, but gives the same etymology for Eden (1960, 160); Watts, however, supports the 1936 interpretation (2004, s.v. River EDEN §2).

Edinburgh (City of Edinburgh: N55:57:12 W3:11:18; 325900 673991). Probable.
Watson discusses this name at length, but comes to no conclusions about the edin- element. He does, however, give the first as din, ‘fort’, following the earlier attestions (e.g. 9th century Dineidyn; 1926, 340–41), and this is assuredly p-Celtic (though Mackenzie improbably derives it from Gaelic dun + aodann, ‘fort on the hill brow’ 1931, 38). Dixon gives this as ‘fort of Eden’, Eden probably being a district name (No ref—does she mean MacDonald? He’s cited, 1941, 120–21, in the app.).

Egglescliffe (Stockton on Tees: N54:30:54 W1:21:38; 441500 513500). Possible.
Mawer gives this as some Celtic form of Latin ecclesia with the later addtion of Old English clif, and notes that the voicing may have been under the influence of ON personal name Egill (1920, 72). Ekwall concurs with this derivation for the first element (1960, 162). Whether ultimately from Latin or from an equivalent of eglywys, this name would qualify for inclusion here. But Watts was cautious, emphasising the possibility of Old English personal names like Ecgi, Ecgel (2004, s.v. EGGLESCLIFFE)

Eggleston (Durham: N54:36:24 W1:59:38; 400500 523500). Possible.
Mawer gives this as ON personal name Egill + Old English tūn ‘enclosure, settlement, estate’ (1920, 72). Ekwall gives it as Old English personal name Ecgwulf or Ecgel + tūn (1960, 162), while Watts offers the further possibiliy of Ecgi (2004, s.v. EGGLESTONE ABBEY). But as Watts recognises, eglywys is also a possible first element.

Gateshead on Tyne (Gateshead: N54:56:19 W1:37:09; 424500 560500). Unlikely
Mawer interprets Bede’s form ad caput caprae as a folk-etymology attempting to make sense of an unfamiliar Celtic name, linking the form further with Gabrosenti in the Notitia Dignitatum (1920, 92). If this is correct, the first element of this name may have been gabro-, ModW gafr ‘goat’. However, Gabrosenti is now identified instead with Moresby in Cumberland (Rivet–Smith 1979, 364). Ekwall says this is a ‘headland or hill frequented by (wild) goats’, implying an English etymlogy (1960, 193; cf. Watts 2004, s.v. GATESHEAD). Even if the name was originally translated from Celtic into English, this would not make it eligible for inclusion as this study seeks surviving p-Celtic names.

Glendhu Hill (Northumberland: N55:10:15 W2:41:04; 356500 586500). Possible.
Etymologised by Watts in the same way as Glendue below (showing gaelicised spelling). Glendhu is first attested, however, only on the 1868 Ordnance Survey map, and might in any case have been transferred from Glendue or from Glendhu Hill in Cumbria (on which see Watts 2004, s.v.). The modern map shows a Glendhu to the south of the hill, but the antiquity of this name is uncertain.

Glendue (Northumberland: N54:59:02 W2:10:52; 388500 565500). Possible.
Suggested by Ekwall to be glyn ‘valley’ + du ‘black’ (1960, 198) and accepted by Coates either to be this or its Gaelic cognate (2000b, 323). Brittonic seems much more likely this far south than Gaelic, and it is accepted as Brittonic by Watts, who adds that ‘Glendue is one of the narrowest and darkest valleys in S Tyndale’ (2004, s.v. GLENDUE FELL). Watts suggests that ‘the treatment of PrW ü shows that the name cannot have been borrowed into English before c. 1000’; if he is right then the case for a Gaelic etymon might be strengthened, but his reasoning here is unclear.

Gloster Hill (Northumberland: N55:20:02 W1:35:58; 425500 604500). Unlikely.
Mawer thinks this is Romano-Celtic (1920, 94). However, Ekwall says it is ‘no doubt’ named from Gloucester in Gloucestershire—compare the early attestations Gloucestre (1178) and Gloucester-hill (1637)—and would therefore be an English name (1960, 199)—but a parallel formation on a Romano-British place-name Glevum (as with Gloucester) is possible. Watts adds a further possibility of Old English glēow ‘sport’ + ceaster ‘Roman fort used for sport’ (2004, s.v. GLOSTER HILL).

Gogar (City of Edinburgh: N55:56:24 W3:19:32; 317297 672680). Unlikely.
Watson, followed by Dixon, gives this as probably p-Celtic go + cor, ‘a small setting’ (Watson, 1926, 210; Dixon 1947, 275–76). I have not been able to find any attestations of this meaning for cor elsewhere. However, it may also have to do with one of the word’s other meanings, that is, ‘boundary, limit’ (noted by Padel 1985, s.v. *cor2). It seems likely, however, that Watson is referring to a nominal form of the Irish verb -cuirethar, ‘to put, set’. On the other hand, the second element may be carr, ‘rock’. Mackenzie thinks it may be based on a p-Celtic root cognate with gog, ‘velocity’ (1931, 109). However, I have found no other attestation of this word. I find none of these etymologies very convincing.

Gromyre (West Lothian: unlocated). Unlikely.
Watson thinks this is from p-Celtic gronna, ‘swamp’, + Old English myre (1926, 379). The former element is not attested in the Geiriadur Prifysgol Cymry, however, and so this seems unlikely. Mackenzie thinks the first element is derived from ON grómr, ‘a dirty spot’ (1931, 187).

Halltree (Scottish Borders: N55:45:44 W2:56:02; 341500 652500). Possible.
Dixon gives this as Old English heald, ‘sloping, inclined, bent’ + p-Celtic tref, or ON hallr, ‘slope’ + p-Celtic tref (1947, 284). Nicolaisen thinks this would be an extraordinary form due to the order of the elements, which would suggest a true hybrid name (1976, 169). I concur with this judgment. One option would be to infer Old English heald + trēow ‘tree’; another p-Celtic hâl ‘moor, marsh’ + tref (cf. Padel 1985, s.v. hal).

Heddon, East and West (Northumberland: N55:00:39 W1:47:26; 413500 568500). Unlikely.
Mawer says this is ‘probably pre-English’ (1920, 109), but his association of the second element with the ‘common Welsh place name suffix -wen’, on the basis of early forms such as Hidewine (1177) is unconvincing: I am unaware of such a suffix. Ekwall gives it more plausibly as derived from the Old English personal name Hidda + Old English *winn, ‘pasture’ (1960, 231).

Hepburn (Northumberland: N55:30:51 W1:53:55; 406500 624500). Unlikely.
Mawer thinks the earliest forms—the earliest being Montem Hibberdune (c. 1050)—suggest that this is a Celtic name, later corrupted (1920, 111). However, he gives no etymology. Ekwall gives it as Old English heah + byrgen, which seems more likely (1960, 235; cf. Watts 2004, s.v. HEPBURN).

Hexham on Tyne (Northumberland: N54:57:58 W2:06:11; 393500 563500). Unlikely.
Mawer thinks this name—first attested in Bede’s Historia ecclesiastica as Hagustaldensis ecclesia—is derived from a Celtic river-name which has been folk-etymologized to give what appears to be Old English hagusteald, ‘bachelor, warrior, virgin’, as a first element (1920, 114). He argues that the forms on the pattern of heagostealdes ea, ‘bachelor stream’, comprise a very unusual name. However, even if this is Celtic, or pre-Celtic, if it is named from a river then it is not relevant for this study. Ekwall, however, is happy with the ‘bachelor’ derivation (1960, 237), while Watts would accept either this or a homophonous personal name Hagustald (2004, s.v. HEXHAM). For a fuller consideration see Bullough 1999, with refs.

Inchkeith Hill (Scottish Borders: N55:43:38 W2:49:18; 348500 648500). Unlikely.
Watson thinks this name may be derived from ynys + coed (1926, 382). However, given its inland location, it seems likely that the name was transferred from Inchkeith Island, in the Forth, and therefore could have been coined at any time by any speech community.

Keith, Upper and Lower (East Lothian: N55:51:09 W2:52:20; 345500 662500). Probable.
Watson derives this and the other keith names in the area from coed (1926, 382).

Kielder (Northumberland: N55:14:03 W2:35:28; 362500 593500). Unlikely
Mawer in 1920 states that this is pre-English, and in 1929 agrees with Ekwall to derive it from caled ‘hard’ + dwfr ‘water’. But they and Watts think it is probably a river-name in origin (Mawer 1920, 127; 1929 180; Ekwall 1928, 61–2; Watts 2004, s.v. KIELDER).

Kilpunt (West Lothian: N55:55:41 W3:26:59; 309500 671500). Possible.
Macdonald, on the basis of numerous early attestations along the lines of Kenpunt (c. 1200) derives this from Gaelic ceann + p-Celtic pont, and says it is a half Gaelicized British name, with the original being *Penpont, ‘summit of the bridge’ (1941, 43). Forms in Kil- are first attested in 1467 (with Kilpont).

Kirkley (Northumberland: N55:04:58 W1:44:35; 415000 576500). Probable.
Mawer gives this as cruc + Old English hlæw (1920, 128). Ekwall, seeking to explain the medial syllable in early forms like Crikelawe (1175), gives it as cruc + Old English hyll + hlæw, and cites it as an example of name-elements identical in meaning accumulating as each one becomes opaque (1960, 280). But the hyll element seems unnecessary to explain the e, which could amongst other things be an inorganic composition vowel, and thus I concur with Mawer in this case, as does Coates (2000b, 324).

Kittyflat (Scottish Borders: N55:44:09 W2:52:10; 345500 649500). Unlikely.
This is first attested late, in 1567, as Ketteflat and Kittoflat(t). Dixon gives this as possibly deriving from coed + ME flat (1947, 291). This does not explain the medial vowel, and thus seems fairly unlikely.

Lampart (Northumberland: N55:03:50 W2:29:41; 368500 574500). Unlikely.
Mawer says this is ‘clearly not of English origin’ (1920, 131); its form has changed little since its first attestation, in 1291, as Lythel Lampard. Mawer’s conclusion is not in itself convincing, especially because Old English long + port ‘trading place’ is possible.

Lauder (Scottish Borders: N55:43:07 W2:45:28; 352500 647500). Probable.
Breeze has recently discussed this name, arguing it to be cognate with Middle Welsh llawedrog 'heaped, loaded', and with other Welsh toponymic parallels, therefore meaning 'mound' and denoting the rise of which Lauder stands (2000). This seems plausible.

Lindisfarne (Northumberland: N55:41:05 W1:48:10; 412500 643500). Possible.
This has recently been discussed at length by Coates, who considers it to be an Archaic Irish name-expression *Lindis-ferann, ‘domain at/of Lindis’ (2000a). Watts, however, seems to give credence to an older association with llyn (2004, s.v. LINDISFARNE).

Linlithgow (West Lothian: N55:58:16 W3:35:45; 300500 676500). Probable.
Watson gives this as llyn + llaith/llwyd + cau, ‘lake of the damp/grey hollow’ (1926, 384). Mackenzie gives the second element as OW llyth, ‘soft’, and says the whole thing means ‘lake of the (soft) marshy field’, taking as the last element the late cae rather than earlier cau (1931, 44). He thinks this interpretation of the last element is more fitting to the early attestations, in -cu (e.g. Linlidcu 1138), than cau. Macdonald, however, agrees with Watson’s interpretation, noting that it fits the geographical situation (1941, 53). He also notes that the pronunciation of the name current at the time of his writing lacks the element llyn, and takes this to be the older form. However, given that the earlier attestations contain the llyn element and the later do not, it seems more likely that this was lost than gained. Nicolaisen agrees with cau as a final element for this name, as do I (1976, 172).

Linsheeles (Northumberland: N55:20:55 W2:10:17; 389230 606070). Unlikely.
Mawer gives this as meaning ‘shiels by the linn or pool’ (1920, 135). The first element seems likely ultimately to derive from p-Celtic llyn, ‘lake’. However, this need not reflect an old place-name, but rather the dialectal borrowing of the llyn-word into English from p-Celtic (for which see Oxford English Dictionary, s.v. linn1).

Liston (City of Edinburgh: N55:57:20 W3:24:11; 312500 674500). Possible.
Macdonald, followed by Dixon, has two etymologies for this—the first, Brittonic llys, ‘court’ + Old English tūn, ‘farmstead’, a hybrid name, the second, Old English personal name Lissa + tūn, entirely Old English (Macdonald 1941, 39–40; Dixon 1947, 213). He cites Ekwall on this personal name but says that there is no real evidence for it, and no trace of inflexion in the early attestations, the earliest being Listona (1163–78). Given the Old English second element, an Old English first element seems more likely. However, I shall not entirely dismiss the Brittonic derivation.

Longniddry (East Lothian: N55:58:42 W2:53:27; 344500 676500). Probable.
Watson gives this as Old English lang + p-Celtic newydd, ‘new’ + tref (1926, 363; cf. Padel 1985, s.v. nowyth). Observing early attestations such as Langnudre (1424), he thinks the development of newydd in this case may have been influenced by Gaelic nuadh/nodha. Mackenzie gives the second element as noeth, ‘naked’ (1931, 221; cf. Padel 1985, s.v. noth). Nicolaisen agrees with Watson (1976, 169), and I am inclined to agree, although I suggest an alternative nawdd, ‘protection’ as the second element, in which case the whole would mean ‘sanctuary farm, estate’.

Manor (Scottish Borders: N55:35:06 W2:31:59; 366500 632500). Unlikely.
Watson says this is from maenor, ‘the stone built residence of the chief of the district’ (1926, 383). This is a later meaning, however, influenced by French and English ‘manor’—the original meaning in Welsh is territorial, along the lines of ‘group of townships’ (Geiriadur Prifysgol Cymry, s.v.). The name could, of course, also proceed from the aforementioned French and English word, which seems altogether likelier. The identification is uncertain—the grid reference is for Manorhill House.

Melrose (Scottish Borders: N55:36:07 W2:43:25; 354500 634500). Probable.
Watson says this is probably Brittonic, moel + rhos ‘bare moor, blunt promontory’, noting that the site of Old Melrose is more suited to the latter meaning then the former (1926, 496)—which fits, in turn, with Padel’s reckoning that ‘promontory’ is the earlier meaning of rhos (1985, s.vv. *moyl, *ros). Mackenzie is happy with ‘bare moor’ (1931, 137), and Nicolaisen agrees (Nicolaisen 1976, 172).

Milfield (Northumberland: N55:35:42 W2:06:17; 393500 633500). Probable.
Milfield is not ostensibly a p-Celtic name, but is conventionally identified with Bede’s Maelmin, equivalent to moel ‘bare’ + mynedd ‘hill’ (Watts 2004, s.v. MILFIELD). Alternatively, Coates has suggested mael ‘prince’ + min ‘edge’ (2000b, 323). Either way, the form is p-Celtic.

Mindrum (Northumberland: N55:35:09 W2:14:51; 384500 632500). Probable.
Mawer gives this as mynydd + druim, a hybrid p-Celtic/Gaelic name which would mean ‘back of the mountain’ (1920, 143). However, this would be a compound, which seems unlikely (see the discussion of hybrid names, above). Ekwall gives the same meaning, but has the second element as p-Celtic trum or drum, ‘ridge, slope, hill crest’, which seems more likely (Ekwall 1960, 327); Watts agrees (2004, s.v. MINDRUM) as does Coates (2000b, 323). Hence the name would mean ‘mountain ridge’ or ‘mountain of the ridge’.

Mossfennan (Scottish Borders: N55:34:09 W3:24:18; 311500 631500). Probable.
Watson spells this Mossfennon, situating it in Glenholm parish, and gives it as maes + fynnon, ‘field of the fountain’. The earliest attestation is in fact Mosspennoc (1250), but variation between /f/ and /p/ has been noted and discussed by Mackenzie (1931, 75). Mackenzie himself implausibly gives the first element as English meaning ‘moss’ and the second as p-Celtic pennoc/pennan, which he says may mean ‘hill water’ (1931, 182, 189). I have found no attestations of this word, and Watson’s derivation seems the most likely.

Mosspeeble (Dumfries and Galloway: N55:13:55 W2:58:07; 338500 593500). Probable.
Watson gives this as maes + pebyll, ‘field of the tents’, and says it is in Ewesdale (1926, 378). Mackenzie gives the second element as pibell, ‘a duct’. He has the first as English ‘moss’, but thinks this has superseded a Celtic original (1931, 185). Cf. also Peebles, below.

Niddrie (City of Edinburgh: N55:56:09 W3:07:33; 329758 671981). Probable.
On the basis of early forms such as Nodereyf (1266) and Nudref (1290), Watson gives this as newydd + tref and thinks the development of newydd may have been influenced by Irish nuadh/Sc.Gaelic nodha (1926, 363). Mackenzie thinks the second element is p-Celtic noeth, ‘exposed’ (1931, 221), but other commentators have followed Watson (Nicolaisen 1976, 169; Dixon 1947, 236).

Niddry (West Lothian: N55:56:46 W3:27:02; 309500 673500). Probable.
See Niddrie above.

Ochiltree (West Lothian: N55:56:41 W3:32:48; 303500 673500). Probable.
Watson gives this as uchel + tref, ‘highstead’ (1926, 209). Note the lack of lenition on the /t/, evident since the earliest attestations, Okeltre and Ockiltre (1211-14). Others agree (Mackenzie 1931, 221; Macdonald 1941, 61; Nicolaisen 1976, 169).

Ogilface (West Lothian: N55:53:18 W3:45:26; 290186 667538). Probable.
Watson gives this as uchel + maes, ‘high field’ (1926, 378). Macdonald agrees, and notes the corresponding geography of the site (1941, 97). The co-ordinates given are marked on the OS Landranger map as Craigs but are also known as Ogilface Craigs.

Painshaw/Penshaw (Sunderland: N54:52:31 W1:29:42; 432500 553500). Probable.
Mawer gives this as pen + Old English sceaga, ‘wood’, but this does not fit the early attestations (the earliest of which are Pencher, c. 1190; Pencher 1305). Ekwall agrees with the derivation for the first element and posits an i-mutated plural of p-Celtic *carr as the second element (1960, 356).

†Panbart Hill (East Lothian: Unidentified). Possible. Watson identifies this obsolete name, attested only in 1573, as Panbart Hill, as ‘part of the lands of Hetherwick’ and derives it from pant + pert ‘hollow wood’ or ‘wood in the hollow’ (1926, 374). The first element might also be pen; Watson’s second element in Modern Welsh is perth, ‘copse’, and is unattested in the non-fricativized form pert, but t for θ is not unlikely.

†Panlaurig (Scottish Borders: N55:46:28 W2:20:39; 378500 653500). Possible.
Watson gives this obsolete name, attested only in 1509, as Panlaurig, as deriving from pant + llanerch, ‘hollow in the glade’, and says it is ‘in the territory of Duns’ (1926, 374). Another possibility for the first element is pen. The co-ordinates given here are those of Duns itself.

Pappert Law (Scottish Borders: N55:32:52, W2:50:01; 347500 628500). Possible.
Watson gives the second element as pert, ‘wood’ (Nicolaisen giving the Modern Welsh form perth and its modern meaning ‘bush, brake or copse’; 1926, 357; 1976, 164). As in Panbart Hill, the final -t is no cause for surprise. Watson situates the hill ‘in Selkirk’, and it is Selkirk’s co-ordinates that are given here.

Pardivan (East Lothian: unlocated). Probable.
First attested in 1144 as Pardauarneburne, this is etymologised by Watson as par + dwfn, ‘deep field’ (1926, 372). Dixon, adding two later attestations including Pardivan (1773), agrees (1947, 168). However, I have been unable to find other attestations of par as a p-Celtic element. More likely is parth + dwfn, ‘deep area’ which would also account for the lack of lenition shown on the second element, as parth is masculine. However, the element parth is not attested in Welsh place-names (Geiriadur Prifysgol Cymry, s.v.). The etymology of the second element is, at any rate, convincing. The same form seems to be attested three times within a small area (see Pardovan and Parduvine below), and it seems unlikely that these names all arose independently. Pardovan is much the best-attested, but Pardivan is also attested early and so has been accorded the status of ‘probable’.

Pardovan (West Lothian: N55:58:51 W3:31:56; 304500 677500). Probable.
Well-attested, first as Pardufin in 1124, this has been etymologised in the same way as Pardivan, above (Watson 1926, 372; Macdonald 1941, 62).

Parduvine (Midlothian: N55:50:30 W3:06:41; 330500 661500). Possible.
Cf. Pardivan, above, though Parduvine is first attested only in 1773 (as Pardivin; Dixon 1947, 119; cf. Watson 1926, 372). The likelihood that the name was transferred from Pardivan or Pardovan is sufficient that this has been listed only as a ‘possible’.

Peebles (Scottish Borders: N55:39:08 W3:11:08; 325500 640500). Probable.
Watson gives this as pebyll, ‘tents’, and speculates that the spellings with o—principally in the earliest attestation, Pobles (c. 1124)—may be influenced by Gaelic pobull (1926, 383). Mackenzie agrees (1931, 214). Cf. also Mosspeeble, above.

Peffer Mill (City of Edinburgh: N55:55:52 W3:08:45; 328500 671500). Unlikely.
Dixon gives this as Brittonic pefr, ‘bright’, and says it probably derives from the river Braid Burn, which may once have been called Peffer Burn (1947, 102). Nicolaisen concurs with this etymology (1976, 164). The name has accordingly been excluded from the survey.

Pen of Ettrick (Dumfries and Galloway: N55:21:23, W3:15:51; 319954 607662). Probable.
Watson gives the etymology of the first element as pen, but gives no early attestations (1926, 354). This appears on the OS Landranger map as Ettrick Pen.

Pencaitland (East Lothian: N55:54:23 W2:53:21; 344500 668500). Probable.
Watson gives this, first attested in 1296 as Penketland, Penkatlond and Penkatelen, as pen + coed + llan, ‘head of a clearing or enclosure in or near a wood’ (1926, 355). Mackenzie agrees (1931, 177–78), as does Nicolaisen at least with the first two elements (1976, pp 165, 172). However, a more likely interpretation strikes me as pen + coedlan, ‘copse’.

Penchrise (Scottish Borders: N55:20:59 W2:47:53; 349500 606500). Probable.
Watson gives the first element as pen (1926, 354).

Pencraig (East Lothian: N55:58:47 W2:40:58; 357500 676500). Probable.
Watson gives this as pen + carreg, ‘rock’ (1926, 354).

†Pendourik (Midlothian: N55:51:36 W3:04:48; 332500 663500). Probable.
Watson gives the first element of this obsolete name as pen, and says that its only attestation, in Blaeu’s Atlas novus, places it NE of Dalhousie Castle (1926, 355–56). From the appearance of the second element, it seems likely to me to be dŵr, ‘water’, while the final -yk may be an adjectival or diminutive suffix similar to that noted by Padel in Cornwall (Padel 1985, s.v. -yk; cf. s.v. dour). The co-ordinates given here are for Dalhousie Castle.

†Pen(te)iacob (Scottish Borders: N55:42:54 W3:12:12; 324500 647500). Probable.
Watson gives this as a pen-name, with the meaning ‘James’ head’ (in the shorter form) or ‘head of James’ house’ (implicitly pen + ty ‘house’ + Iacob; 1926, 354). Breeze has instead suggested an origin in Cumbric *penty, 'lean-to, hut, shed, outhouse, cottage' (cf. Welsh XXXXX; Breeze 2000, 72). Mackenzie says this name is obsolete and has been replaced by Gillemorestun (literally ‘estate of the servant of Mary’; 1931, 217), which has itself been replaced by Eddleston (‘Edulf’s estate’).

Penicuik (Midlothian: N55:49:54 W3:13:22; 323500 660500). Probable.
Watson gives this as pen + y + cog (‘cuckoo’), adducing parallels for place-names mentioning cuckoos (1926, 355; cf. Padel 1985, s.v. *cok). Dixon agrees (1947, 264), and Nicolaisen agrees with the first element at least (1976, 165).

Peniel Heugh (Scottish Borders: N55:31:52 W2:32:53; 365500 626500). Probable.
Watson gives this as pen + y + gwal, ‘end/summit of the wall’, and says that it is near a fort (1926, 354). As Macdonald pointed out, the derivation of the second part is speculation, though she noted early medieval dry stone structures which might account for it (1991, 17). Although the name lacks early attestations, I am inclined by the name’s location (on a hill-top) to agree with the derivation of the first part.

Penistone Knowe (Scottish Borders: N55:26:11 W3:12:38; 323500 616500). Possible.
Watson gives this as pen + ystum, ‘head of the bend’ (1926, p 354). This suggests that Cumbric had prosthetic y-, like Welsh, but unlike Cornish and Breton. There is no evidence to suggest this state of affairs. However, pen + y + *stum is possible. Derivation from English penny + stone is possible, but penny in place-names tends to be associated with points for tax-collection or with a rental value of a penny. In neither case does Peniestone Knowe stand forward as a candidate, so the p-Celtic etymology is certainly possible.

Penmanscore (unlocated). Probable.
Watson gives this as pen + maen (‘stone’) + English score (1926, 354). The last element seems unusual as a place name element (it is not listed in Smith 1956); though ill-attested, Old English *scoren ‘a steep, precipitous place’ seems more plausible (Smith 1956, s.v.). Watson locates it in Selkirkshire.

Penmanshiel (Scottish Borders: N55:54:01 W2:18:48; 380500 667500). Probable.
Watson gives this as pen, ‘top’ + maen, ‘stone’ + ME schele, ‘shieling’ (1926, 354). Nicolaisen agrees, at least with the first element (1976, 165).

†Pennango (Scottish Borders: N55:23:10 W2:51:36; 345615 610580). Probable.
Watson thinks the first element of this obsolete name is pen, and notes that it ‘appears to have been near the junction of Teviot and Allan Waters’ (1926, 354; the map reference given is to this confluence); Breeze agrees and argues that the second element corresponds to angau ‘death’, adducing other British parallels for a place-name along the lines of ‘death hill’ (2002b). At any rate, no Old English or Old Norse etymon looks likelier.

Pennygant Hill (Scottish Borders: N55:17:11 W2:52:31; 344500 599500). Probable.
Watson thinks this name may be pen + y + gwant, the latter element being a Brittonic word meaning ‘rest, break, division’ (1926, 354). This derivation for the second element seems somewhat implausible, but Nicolaisen agrees with the first element (1976, 165). The second element may also be *cant (cf. Carcant, above).

Pennymure (Scottish Borders: N55:25:26 W2:23:19; 375500 614500). Possible.
Watson gives this as pen + y + mur, ‘head of the wall’, and says it is the site of a Roman camp (1926, 354); Macdonald refers further to ‘a linear earthwork here’ (1991, 31). However, the Scots word muir, ‘moor’ seems more likely for the second part and therefore the first element is likely to be English as well (cf. discussion of names in pen-, above). The map reference is for Pennymuir.

†Penratho (East Lothian: unlocated). Probable.
Watson gives this obsolete name as pen + rhathau, ‘head of the circular fortified places’ (1926, 355). The Geiriadur Prifysgol Cymry gives rhath as meaning ‘earthwork’ or ‘plain’. The word is admittedly only recorded in the twentieth century with either of these meanings, but it seems to be paralleled by Ratho in the City of Edinburgh; no Old English alterntive seems likely; and the initial pen- speaks for a p-Celtic origin.

Pentland (Scottish Borders: unlocated). Possible.
Mackenzie derives this—first attested as Pentland (c. 1150) and then as Pentland(e) (1236)—from Old English penn + land, ‘enclosed land’ (1931, 215). Dixon, however, gives this as possibly from pant + Old English land or pant + llan (1947, 227).

Penvalla (Scottish borders: N55:38:30 W3:20:39; 315500 639500). Probable.
Watson thinks the first element of this name is pen (1926, 354). Nicolaisen agrees with the first element at least (Nicolaisen 1976, 165).

Penveny (Scottish Borders: N55:38:31 W3:19:41; 316500 639500). Probable.
Watson thinks the first element of this name is pen (1926, 354).

Pinkie (East Lothian: N55:56:29 W3:01:05; 336500 672500). Probable.
Marshalling early attestations such as Pontekyn, Pontekyn, Pontekin (from 1198), Dixon gives this as pant + cŷn, ‘valley of the wedge’ and notes that this suits the topography of the area (1947). Nicolaisen agrees with this (1976, 18).

Pirn (Scottish Borders: N55:44:07 W2:55:20; 342190 649490). Probable.
This is attested first in 1440 as Pyrn(e), and shortly after as Pryn (1463), Dixon (1947, 286), Watson (1926, 351) and Nicolaisen (1976, 165) all giving it as pren. I have not been able to identify Pirn’s location; the co-ordinates given are of Stow, in whose parish Watson situates it.

Pirn (Scottish Borders: N55:37:36 W3:02:30; 334500 637500). Probable.
Watson gives this as pren (1926, 351). Nicolaisen agrees (1976, 165).

†Pirncader (Scottish Borders: N55:44:07 W2:55:20; 342190 649490). Probable.
Watson gives this obsolete name as possibly deriving from pren and describes the second element as ‘doubtful’ (1926, 351), Mackenzie suggesting bryn + cader, ‘hill fort’ (1931, 74). Dixon agrees with Watson (1947, 286), as does Nicolaisen (1976, 166). The early attestations marshalled by Dixon do not lend themselves to the interpretations given, but the lateness of the evidence in this case may be the cause of this. I have not been able to identify Pirn’s location; the co-ordinates given are of Stow, in whose parish Watson situates it.

Pirnie (Scottish Borders: N55:32:56 W2:33:51; 364500 628500). Probable.
Watson thinks this name may derive from prenau (1926, 351).

Pirnta(i)ton (Scottish Borders: N55:43:35 W2:55:01; 342500 648500). Probable.
Watson thinks the first element of this name is pren and the second possibly tyddyn (‘smallholding’; 1926, 351). Dixon, adducing the earliest attestations Prentatoun(e) and Prentatto (1479) agrees (1947, 287), and Nicolaisen also gives pren as the first element (1976, 166).

Pirny Braes (East Lothian: N55:54:23 W2:53:21; 344500 668500). Probable.
Nicolaisen thinks this name derives from pren + an unknown suffix (1976, 165). I have not been able to find it, although he says it is in Pencaitland parish, so the co-ordinates given are for Pencaitland.

Plenderleith (Scottish Borders: N55:23:48 W2:25:12; 373500 611500). Probable.
Macdonald etymologised Plender- as bryn + tref. This seems plausible (as would pren- and blaen- ‘summit’); at any rate, a p-Celtic etymology is hard to doubt.

Plenmeller (Northumberland: N54:57:55 W2:26:48; 371500 563500). Probable.
This name is first attested in 1255 and 1256, when it exhibits already a variety of forms: Plenmeneure; Plenmenewre; Playnmalevere. Later forms prefer the -mel- form. Mawer identified the name as Celtic without making any clear statement of etymology (1920, 158); Ekwall plausibly offered blaen + moel + bre, ‘top of the bare hill’ (for the substitution of p for b cf. Barnbougle, above, although the reverse process may apply there; 1960, 368), which Coates accepts (2003b, 323).

Plenploth (Scottish Borders: N55:43:36 W2:54:04; 343500 648500). Probable.
Watson gives this as pen + plwyf, ‘people, community, parish’ (1926, 355), as does Dixon (1947, 288); the etymology of the second element is supported by the earliest forms, Plenploif (1593) and Plenploff (1598). The first element could also be blaen—cf. Plenmellor, above.

Posso (Scottish Borders: N55:35:19 W3:15:46; 320500 633500). Unlikely.
Watson thinks this may be derived from OW *poues, ‘rest, repose’. This seems fairly implausible (1926, 382).

†Powtreuet (Northumberland: N55:10:52 W2:26:00; 372500 587500). Probable.
Mawer thinks this is Celtic, although he gives no etymology for it (1920, 160). He does not say whether it is in Northumberland or Durham. The early attestations, among them Poltrerneth (1325) and Poltrevet (1329), certainly seem to be p-Celtic. The first element may be Brittonic pwll, ‘pool’, and the second tref or, as Coates suggests, trefred ‘abode’ (2000b, 323). Alternatively, forms like Poltrerneth may show yr as a third element. It appears to have been in the vicinity of Falstone, whose co-ordinates are given here.

Primrose (Midlothian: N55:50:33 W3:01:54; 335500 661500). Possible.
Although this name sounds very English, Watson says it may be derived from pren + rhos (1926, 351). Nicolaisen agrees (1976, 166). Neither, however, cites early attestations. The location given is tentatively based on the localisation of Primrose Cottages to ‘shortly after Stobs’ in Scott 2006 (s.v. Primrose Cottages); Stobs’s co-ordinates are given here.

Primrose (Scottish Borders: N55:48:37 W2:20:41; 378500 657500). Probable.
Cf. Primrose, above (Watson, 1926, 351; Nicolaisen 1976, 166). The identification is uncertain—the map reference given is for Primrosehill.

Primside (Scottish Borders. N55:31:55 W2:17:41; 381500 628500). Probable.
Drawing on the early form Prenwensete, Watson says this is pren + gwen ‘white’ + Old English sete (meaning setu, ‘seats, encampments, animal folds’ or perhaps sæte ‘house’; Watson 1926, 351). The reading is supported by a form from 1140, Praunwesete, adduced by Johnston (1934, 279), and is accepted by Nicolaisen (1976, 165).

Printonan (Scottish Borders: N55:43:14 W2:19:40; 379500 647500). Probable.
Watson says this is pren + tonnen, ‘tree of the bog’ (1926, 351). Nicolaisen agrees (1976, 166).

Ross (Northumberland: N55:37:51 W1:47:14; 413500 637500). Probable.
Mawer gives this as Brittonic rhos, ‘promontory’ (1920, 169), as do Ekwall (1960, 392–93) and Coates (2000b, 323); Watts adds the possible interpretation of ‘moor’ (2004, s.v. ROSS).

Soutra (Midlothian: N55:49:32 W2:52:17; 345500 659500). Probable.
Watson says this is p-Celtic but gives no etymology. He notes that the earliest spellings are regularly Soltre (1934, 299). Nicolaisen gives it as sulw + tref, ‘steading of the wide view’ (1976, 169). Dixon agrees with this (1947, 189–90). Breeze seems to have been unaware of Nicolaisen's suggestion for the first element, but accepted the second element as tref, emphasising not only the site's commanding view, but its strategic importance as the point where Dere Street 'crosses the last upland pass before descending to the plain of Lothian', and so as a major point for the defence of Edinburgh (2000, 76-79, at 77). He suggests further that the Middle Welsh Culhwch ac Olwen contains another reference to the site in its mention of Dinsol, though I have not plotted this on the map. Early attestations are from Johnston.

Tartraven (West Lothian: N55:56:11 W3:35:43; 300450 672640). Possible.
Macdonald gives this name as Gaelic torr + p-Celtic tref. He notes its pronunciation as [tVr'treivn], with the stress characteristically on the penultimate syllable (1941, 64). I can see no reason why the Gaelic torr should be preferred in this case to p-Celtic tor(r). The second element may be trefn ‘dwelling’. The co-ordinates are for Mid Tartraven.

Tecket (Northumberland: N55:02:49 W2:12:46; 386500 572500). Unlikely.
Mawer says this is ‘a Celtic name’ but gives no etymology.

Torcraik (Midlothian: N55:49:28 W3:00:54; 336500 659500). Possible.
Dixon gives this as Brittonic tor(r) + craig (1947, 114), his earliest attestation being Torcraik (1611).

Torfichen Hill (Midlothian: N55:46:13 W3:03:42; 333500 653500). Possible.
Dixon, presenting the earliest attestation Torphichen Hill (1773), says this may be derived from Gaelic torr + p-Celtic bychan ‘small’ (1947, 104). He says this name may have been transferred from West Lothian. Mackenzie gives the same etymology (1931, 251). However, this hybrid seems unlikely, as one would expect a p-Celtic element to come before a Gaelic one in any hybrid of the two languages, due to the chronology of the languages’ domination of the area. If the second element is indeed bychan, the first would have to be a feminine noun, which Brittonic tor(r) is not—it therefore seems more likely that the first element would be Brittonic tref.

Torquhan (Scottish Borders: N55:43:05 W2:52:31; 345106 647531). Possible.
Dixon says that this—first attested as Torquhene in 1593—is obscure, but notes that two other similar names in Kirkcudbrightshire and Carrick are almost certainly derived from tref (1947, 288). Nicolaisen also thinks this may contain tref as a first element (1976, 168). The map reference is for a place in Borders county.

Torsonce (Scottish Borders: N55:40:55 W2:52:06; 345500 643500). Unlikely.
Watson says this is ‘of doubtful meaning’ but associates it with torr (implicitly the Gaelic word; 1926, 145). Dixon, whose first attestation is Torsons (1567), says that although this is obscure, if Torquhan (above) is derived from tref then this may well be too—he cites Watson who says that tref names tend to occur in clusters (1947, 288; Watson 1926, 362).

Trabroun (East Lothian: N55:57:38 W2:51:31; 346500 674500). Probable.
Watson gives this as tref + yr + bryn (1926, 359–360). Nicolaisen agrees (1976, 168). The article is not obvious in the only attestation given, but may have been adduced by analogy with Trabrown, below, possibly on the assumption that tref always occurs as a generic.

Trabrown (Scottish Borders: N55:44:11 W2:47:24; 350500 649500). Probable.
Watson gives this as tref + yr + bryn, a reading supported by the first attestation Treuerbrun (c. 1170; 1926, 359). He says it is on a hillside, and spells it Trabroun. Mackenzie spells it Trebrown and agrees with Watson’s etymology (1931, 75), as does Nicolaisen (1976, 168).

Trahenna (Scottish Borders: N55:37:24 W3:22:31; 313500 637500). Possible.
Watson gives the first element of this as tref (1926, 363).

Tranent (East Lothian: N55:56:31 W2:57:15; 340500 672500). Probable.
First attested around 1127 as Trauernent, this was reckoned by Watson to derive from tref + yr + neint, ‘homestead of the streams’ (although the original meaning of nant was ‘valley’—cf. Padel 1985, s.v. nans). Mackenzie gives it as tref + ar + nant, ‘homestead on the brook/valley’ (1931, 221). Nicolaisen agrees with Watson (1976, 168).

Traprain (East Lothian: N55:58:15 W2:39:02; 359500 675500). Probable.
Watson, following the earliest attestation Trepren (1335), gives this as tref + pren (1926, 352, 363). Nicolaisen agrees with Watson (1976, 166), though it is worth noting the lack of lenition on /p/ if Watson’s etymology is correct. Mackenzie gives it as tref + bryn, and, noting that there should be lenition on /b/ if this is the case, compares it with other instances of /v/ and /p/ being confused (1931, 75).

Traquair (Scottish Borders: N55:35:58 W3:03:25; 333500 634500). Probable.
Watson gives this—whose first attestation is Treverquyrd (c. 1124)—as tref + yr + the river-name Quair (1926, 360), and Nicolaisen agrees (1976, 168). Mackenzie agrees with this although he gives the connecting particle as ar, thus meaning ‘farm on the river Quair’ (1931, 221). This interpretation seems more likely.

Treverlen (City of Edinburgh: N55:56:44 W3:08:10; 329148 673070). Probable.
Watson suggests tref + gwr + lên, ‘homestead of the learned man’ (1926, 360–61). This seems a stretch, and it is perhaps worth noting that the Geiriadur Prifysgol Cymru notes (as erroneous) some examples of glyn as a feminine noun (s.v. glyn; if this could be so in Cumbric then we could have tref + y + glyn, with lenition following the definite article. The name was superceded during the thirteenth and fourteenth centuries by Dodinestun, ‘Dodin’s estate’, now Duddingston (Barrow 1959).

Troughend (Northumberland: N55:13:36 W2:12:50; 386500 592500). Possible.
Mawer says this is ‘a Celtic name’ (1920, 201). Ekwall says it is ‘unexplained’, but ‘possibly a British name’ (1960, 481). If it is, indeed, of p-Celtic origin, then the first element must almost certainly be tref; the second, as Watts suggests, would perhaps be gwen ‘white’ (2004, s.v. TROUGHEND COMMON). If so, Watts notes, the name was borrowed after the development of w- > gw- in Brittonic, which would be unusual for English names. But dating the change in Wales is hard (it seems not to have been complete by the seventh century), so the situation in the North is wholly uncertain (see Sims-Williams 2003, 288).

Trously (Scottish Borders: N55:41:57 W2:58:48; 338500 645500). Possible.
Watson gives this as traws or tros ‘across’ + lle (1926, 350). Dixon, contributing the earlier attestation Trouslaw (1773), agrees with this (1947, 288).

Wardrew (Northumberland: N55:00:35 W2:33:24; 364500 568500). Possible.
Mawer gives this as an unknown first element + rhiw ‘slope’, citing Ekwall’s etymology of Cumrew in Cumberland (1920, 207; for the ending cf. Padel 1985, s.v. *rew).

Wooperton (Northumberland: N55:28:42 W1:56:46; 403500 620500). Unlikely.
Mawer says this is from gwy ‘water’ + bre + Old English denu ‘valley’, citing Wepre, in Flintshire (1920, 220). However, Ekwall derives Wooperton more plausibly from Old English weoh + beorg or west + byrig (nom. sing. burh), ‘West town’ (1960, 533); Watts emphasises our uncertainty over the name (2004, s.v. WOOPERTON).

Wrekin Dike (Durham: N54:56:50 W1:31:31; 430500 561500). Unlikely.
A Roman road, appearing on the Ordance Survey map as Wrakendike. Mawer says this is ‘certainly Celtic’ (1920, 220), but Ekwall gives it as Old English wræccna + dīc (1960, 533). The co-ordinates are for the north end.

Yeavering (Northumberland: N55:34:05 W2:06:17; 393500 630500). Probable.
Mawer says this name, then pronounced [jivrin] and site of one of the most important archaeological finds of both British and Anglo-Saxon material in Northumberland, is ‘clearly a Celtic name’, but gives no etymology (1920, 221). Ekwall gives this as either deriving from p-Celtic gafr, ‘goat’ (which would have to be an i-affected plural with English palatalisation of /g/ > /j/), or a compound containing bryn, ‘hill’ (1960, 544).