British and European River-Names

zyxwvu KITSON - RIVER-NAMES 93 and Erme (Devon), he agrees about the first, offers various possibilities of differentiation for the second, and gives the third a different derivation a l t ~ g e t h e rFor . ~ ~Rye there is agreement about meaning but disagreement about roots.34He disagrees about Amble (Corn.), ascribing it to the same root they agree on for Amber (Derbys) and Gamber (Herefs), the same as foam-born A p h r ~ d i t e . ~ ~ For all of Yealm (Devon) (one of the names in Al-), Hayle (Cornwall) (one of names in Sub also mapped by Tovar 1977 map 4), Newerne (Gloucs), and names the hydronymists put with them, he both disagrees and invokes a larger number of roots.36So he does zyx zyxw zyxwv zyxw zyxwv zyxw zyx zyxwvu the long-vowelled name(s), though preferring the alternative of irregular phonetic development within Welsh; Rivet and Smith (1981:377) would make that within late British. Either seems preferable to Pokorny’s fishy etymology elaborated by Hamp (1984:130-132 etc.); the Usk is not famed as a trout river! 33 Nicolaisen’s 3 *er-, Mann’s *or- ‘to start up’ (transitive or intransitive); Oare Water (Som) and Ayre are agreed to belong with the Arrow; Nicolaisen would add Armet Water (Midlothian) Arnot Burn (Kinross), Ernan Water (Aberdeen), and following Forster (19415346) Ore (Suffolk), held by Ekwall a back-formation from Orford. Ekwall thought Yarrow (Lancs) possibly identical with this Arrow or possibly with Ganv (Glam), = Welsh garw ‘rough. For Arrow he cites cognates including Skr. a‘rvun(t)-‘running, swift’ and h q a - ‘surging’, leaving open which of the two is the sense. For Earn he cites a largely different set of cognates including ‘Germanic *arnia- “lively, energetic” (in Gothic arniba %c.),’ English run, and Skr. arjuti ‘runs, flows’, whose extended root * e m - he thinks likely to underlie the particular form, may though ‘a derivative with an n-suffix from the root *(s)per-,on which see FROME, also be thought of. The possibilities being many, a definite etymology cannot be suggested.’ Nicolaisen’sdefinite equation with urng seems preferable; its sense would suit the Arrow well enough, and presumably also the Earn in the days before draining of the Levels, though typically of hydronymists Nicolaisen holds back from such an exact suggestion. The Erme Ekwall held most likely to be English, therefore from eormen ‘vast’, therefore a back-formation from the place-name Ermington. ‘I have no suggestion to offer, if the name is pre-English.’That of the hydronymists is very much preferable. 34 19 *reg- ‘to water’ Rye (WRYorks), Rye Water (Ayr) added by Nicolaisen, possibly Ryburn (WRYorks) and Rye (Sussex); Ekwall makes the root *rei- ‘to flow’ in Lat. rivus, O E rib ‘small stream’, Skr. rifj- ‘stream’, and the name possibly identical to a Welsh stream Rhiw flowing into the Severn, if that is not named from rhiw ‘hill’. 35 2 *am- ‘channel’ for Amble, with which Nicolaisen puts OE Amalburna (Suffolk) more confidently than Ekwall, who is even less enthusiastic about alternatives; Amber and Gamber represent the ‘wbhros of Latin imber ‘shower’, Greek dgpds ‘foam’. 36 1 *el- ‘to flow, to stream’ Yealm (Devon) for which Ekwall, finding nothing in Celtic, guesses an adjective parallel to Latin almus ‘kindly, bounteous’, Allow (Cornwall), Alaw (Anglesey), which he sees as ‘sounding’ streams, comparing Welsh alaw ‘music’, or failing that ‘beautiful’ after a sixteenth-century Welsh glossary, and the large group in Alaun-, Aln (Nhb), Alham (Suff.), Ayle Burn (Cu./Nhb), Ellen (Cu.), 94 zy zyxwv TRANSACTIONS OF T H E PHILOLOGICAL SOCIETY 94,1996 for Carrant, discussed in detail below, and Nidd; as for the root to which they ascribe that, he denied that it exists?’ The Norfolk river Ant Ekwall did not think really has a river-name at all.38As we have seen, he did not anticipate the hydronymists’ view of the Thames group nor my bringing Derwent into their Dravunt-group?’ zyxwvuts zyx zyxwvu zyxwvuts one Allen (Corn.), four Welsh rivers Alun (Gwent, Glam., Flint, Pemb.), Allan (Stirling), Ale Water (Roxb.), to which Nicolaisen adds another Ale Water (Berwick), Allander Water (Dumbarton), possibly Allan Water (Roxb.), and Ekwall citing divine and personal names as well as place-names similarly formed concludes ‘Some meaning such as “holy” or ”mighty” would give a suitable starting-point’; 7 *sul- ‘current, stream’ Hail (Gloucs), Hail (Hunts), derived by Ekwall from *sul’dirty, sallow’, Shiel (Inverness), Shiel (Ross), Nant Heli (Montgomery), Halai (Mont.), Halen (Carm.) added by Nicolaisen, and a number of tidal rivers explained by Ekwall as from *suf- ‘salt’ which Nicolaisen admits cannot be excluded, Hayle (Cornwall), Hayle (- Camel estuary, Corn.), Haill (= Helford, Corn.)- one would say they are too close together all to have had the same name, and ‘salt river’ for the large Hayle is the likeliest of the three; 15 *nebh- ’mist, spray’ Nevern (Pemb.), mentioned but not explained by Ekwall, Naver (Sutherland) added by Nicolaisen together with possibly Newerne (Gloucs.), thought by Ekwall a back-formation from a place-name Newerne and not to be brought into connection with the Pemb. name. I’ 17 *n(e)id- ‘to flow’, whence *nidi ‘flowing water’, Nicolaisen’s etymon for Neth (Cornwall), Welsh Nedd 5 Engl. Neath (Glam.), and the second element of Glasney (Corn.), that or something similar for Nidd. Ekwall allows IEur *net(Mann’s *noris) as a possibility for Neth, but would prefer ‘to derive the name from a British word corresponding to OIr necht “clean”’ like the Scottish names Nethan (Lanark) and two Nethy, though admitting phonetic difficulties which o n the standard view, as he admits, would exclude that; yet Jackson’s detailed account of the crucial sound-change (1953:407-4 1 1) perhaps leaves the possibility open. He proposes connecting Glasney with Neth. but denies identity of Neth and Nedd irrespective of the etymology. For Nidd he considers several alternatives, bringing in Continental names Nid(d)a. ‘Sturmfels, Ortsnamen Hessens, derives Niddu from a root nid “to roar, to flow” in Skr nadh, nadi .‘river”. No such root exists. Very likely we may derive the name from the root nei- “to be brilliant”, found with various determinants in Lat niteo, nideo, MIr niumde “brilliant”, W nwyf “vivacity”. The meaning would be “brilliant river” or the like.’ zx 12 *untji ‘end’ (f. adj.), not an important river-naming element. Krahe (1 955: 15) found Continental instances mainly in the Lithuanian hydronym An&, but its reportedly great frequency does not fit such semantics. Krahe’s following discussion overstates the extent to which words meaning ‘boundary’ occur as or in river-names, although there are some. It seems possible that this as a river-namingelement arose by folk-etymological misdivision of some of the many names containing sub-participial -untia, and that Krahe’s example 3 shows the process in action in the Alpine region in the Middle Ages. Ekwall judged Ant just a back-formation from the place-name Antingham, which seems at least as likely; the OE personal name Antu is well attested. 3y I am surprised that it seems to have been left to me to do this in print (the link was made in lectures at Cambridge in the 1970s by Dr. G. P. Cubbin). Probably that is because of the element of analogy, which the mind-set of hydronymists almost as much as that of laryngealists seems to lead them to avoid. zyxwvutsrqpon zyxwv zyxw zyxwv KITSON - RIVER-NAMES 95 The Indo-Europeanness of alteuropuisch names was obvious to Krahe and his colleagues from the beginning. Occasional attempts to prove otherwise depend on ignoring a lot of the evidence presented above and falsifying some of it. A recent such exercise, that of Vennemann (1994), parades a technical linguistic (specifically morphological) virtuosity that may mislead the unwary but lacks proper control in several directions. The collapse of diachrony already mentioned in the argument about suffix-combinations resurfaces in the guise of addition without documentation of many new items to Krahe’s lists of alteuropuisch names, ignoring early forms that are documented, and unsignalled use of examples that are not river-names.4O He thence admits to his stock of roots many that are not ancient substantive elements but German qualifiers, including apparently personal names (e.g. p. 248 and n. 50). I suspect this is commonly, as in one of the most glaring instances it certainly is$*a case of closely followingBahlow (19 6 9 , who, among 4” E.g. Segustero(n) and probably others in note 15 above. When, rarely, he does document a name, Ammer, earlier Ambra, in Bavaria, from late antique and early mediaeval sources (n. 40), consistently attesting the consonant-cluster that shows it to belong to the root mentioned at the end of note 35 above, he arbitrarily rejects their evidence in order to fit a preconceived pattern of vowel-consonant alternation. Yet with names only attested in recent times he insists on projecting the quality of unstressed vowels back onto remote antiquity, not allowing the possibility that some might be phonetically conditioned in particular languages. 4 1 The pair Nagold and Singold, from which he hypostatizes an ancient suffix -ald(pp. 239, 258), even alleging it ‘frequent’ (p. 239), though the consonant-sequence does not occur at all in Krahe (1962:305-341). It is patent from theg in both that they do not contain any such thing but are much more recently-formed compounds. Nagold, eighth-century Nagalta, is in Baden-Wurttemberg, Singold according to Bahlow (1965449) is a tributary of the Wertach west of Munich, attested as Singalta. They are to be taken with the Sinkel in Bavaria, attested in the eleventh century as Sinckalta (Forstemann 19 13-1 6:II.ii 736). That compound has kalt(a) ‘cold (stream)’ as its substantive element. Forstemann (I 1628) cites ninth-century Smalecalta from Thuringia as a clear parallel, Nagalta and eleventh-century Langalta as more doubtful parallels. Langalta seems clear enough; simplification of velar conis fairly routine. Nagalta is odd at first sight, but Nagalsonants in a cluster like -&‘nail’ as first element in stream-name compounds is paralleled in German Nagalbach (Forstemann 1913-16:II.ii 364-365) and English Nailsbourne, so it is best taken as *Nagalkalta with haplology. Vennemann’s basic error is to ascribe ‘Old European’ date to a formation only found on the territory of one later linguistic group. His deplorable lack of references masks the extent to which, despite his rude words about Bahlow (1 965), he freely follows him in taking ‘some seemingly recent settlementnames [to] encapsulate old hydronyms’ (pp. 224-225). The kind of excess that leads to (apparently his own excess, not apparently in Bahlow) is glaringly visible in Isamanninga near Munich cited as from a ninth-century source (p. 224). It is a German -ing zy zyxw zyxwv 96 zyxwvu TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 94, I996 zy other flaws for which Vennemann (1994:224) castigates his ‘poor philological scholarship’, does not use asterisks. All this follows naturally from Vennemann’s rejection in principle (pp. 225-235 passim) of all consideration of meaning before carrying out morphological analysis. That would at best render the methodology dubious, since (as wiser heads like Szemerenyi have not tired of reminding us) semantic links are what justify positing etymological links in the first place; yet granting it for the sake of argument, it is then vital to have control of what is norm and what variation provided by observable frequencies in a large sample. But Vennemaim ignores that. The undocumented additions to Krahe’s names include ones radically at variance with the phonetic patterns in Krahe’s, falsifying the relative frequencies of vowels4* and the phonology of the suffixes which is central to his argument. Krahe’s segmentation Al-ma-nu and the like is established by a great number of examples. Instances with a preceding suffixal vowel like Med-amu-nu are rare and require to be explained as rarities; they do not at all justify Vennemann’ssystematic segmentation Medurn-an-a and the like.“3On standard accounts of Indo-European, zyx zy ~ derivative much likelier to be from a German dithematic personal name in - M u m than from an old hydronym, and the first a in his hypothetical *Isarnunu makes it strongly unlike the phonetic sequences presented by Krahe. Such augmentation of what Krahe approved as alteuroptiisch creates a hybrid Aunt Sally much easier to knock down than Krahe’s own version, and is obviously not valid methodology. None of the names in this note is in Krahe’s own long lists (1 962, 1964), which must be the starting-point of any fair linguistic appraisal. At least whether that is the reason or not, his statement that ‘as Krahe’s tables show’ e and o ‘are relatively infrequent’ vowels and the three-vowel system of ‘major suffixal vowels’ di/u he derives from it are flatly false to the large amount of material in Krahe (1 962:305-342), where u is clearly the least frequent vowel. It only occurs with some suffixes, and in general different suffiwes have different vocalic possibilities, some open to explanation from general phonetics and many from Indo-European etymology. This constitutes the evidence which Vennemann (p. 231) denies exists ‘to decide whether, e.g., -un-/-in-/-un- are ablaut alternants of the same suffix, three different suffixes, or something else’. It also shows as the complications with nasals do that Vennemann’s segmentational analysis is at too high a level of abstraction to be valid anyway. Nor is his u pion‘ assumption about uniformity of VC or CV segmentation across all suftixes justifiable anyway. It is really another Aunt Sally, to aid discarding ‘representations such as “-n% (-no-)” and “ti- (-to-)”’ which ‘are precisely what IndoEuropeanists are accustomed to‘. Custom seems to have dulled his awareness that that presentation is simply shorthand for saying that these -n- and -t- suffixes are productive in the i- and o-declensions and not other declensions. Vowelsegmentation is accidental not essential. Vennemann’s argumentation here sits ill with ’. zyxwvu zyxw zyxwvuts zyxwvu zyxwvu zyxw zyxw KITSON - RIVER-NAMES 97 e.g. Brugmann’s (1886-92 vol. 11), they are secondary phenomena, deriving from variants with sonant qz,Q, etc., which all the liquid and nasal suffixes had beside their consonantal versions. Vennemann’s insistence on segmentation of the vowels as a first step rigs the probabilities against Indo-European in a way not acceptable when testing affinities of linguistic material for which Indo-European is known to be a candidate. Even discounting sonants, epenthetic vowels between voiced stops and following non-homorganic nasals are fairly common in languages! Vennemann should have chosen non-nasal suffixes, but then he could hardly have made his case. The purported account (pp. 235-242) of other aspects of ‘the language of the Old European toponymy’ is largely an elaboration of the invalid initial analysis, with some curious explaining away of similarities to Indo-European and a priori rejection of vowelquantity (pp. 241-242) and of the possibility of historical change in the accentual ~ y t e mVennemann .~~ is unsure too on the relations of name-systems to languages at large. He acknowledges only in passing, in a subordinate clause, room for doubt ‘if the toponyms are indeed characteristic of nouns in general’ (p. 235); he argues about accentual systems and root vowel frequencies entirely on the assumption that they are, which in this material, as I shall show, is highly unlikely. Yet he imports (pp. 233-234) strangely restrictive a priori notions of what it was possible for ancient people to notice in place-names!5 Still Vennemann deserves thanks for supplying what zyxw zyxwv zyxw zyxwvut a rather distastefully repeated insistence that his is a ‘linguisticanalysis . . .which is in accordance with the methods of general morphology’ (pp. 228-229), an accolade he denies Krahe’s. 44 His argument that the hydronymic language had initial accent through its (kpre)history (pp. 245-246) is circular. The a priori assumption of constancy of accentual type over millennia is on the evidence of known languages implausible anyway. Making it again rigs the probabilities against Indo-European, in which the choice is between believing in a minimum of two or three distinct accentual r6gimes in the long period of the reconstructed language (Szemerenyi 1972:141 may insert a third before the two of 138;Lubotsky 1988:181 definitely posits three, followinga work of Beekes 1985 which I have not seen). Likewise the ‘speculation about word order’ (pp. 25225 3) betrays ignorance of actual changes in the history of e.g. insular Celtic. Compare particular points addressed in notes 68,74, and 75 below. 45 That early peoples made more distinctions than their successors, and more than earlier generations of scholars attributed to them, has been the message of English place-name studies in the last thirty years, epitomized perhaps best by Gelling (1 978, 1984). It is all the more extraordinary, and quite unacceptable, for Vennemann (pp. 233-234) to reject a prion‘, as ‘modern-feeling’and too specific, etymologies of 98 zyxwv zyxwv zyxwvu zyxw zyxwv TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 94,1996 had been a gap in the literature and showing us what a seriously worked up attempt to analyse the alteuropiiisch linguistic material as non-Indo-European would look like. It is reassuringly much less coherent than the traditional Indo-European General Indo-Europeanness does not exclude the possibility that some particular formations may have been wrongly so ascribed. This is most likely to be true of rare ones. But attempts in the literature to prove even minor alteuropiiiisch elements really non-IndoEuropean have been conspicuously unsuccessful. Tovar’s (1977:18-21) attempts to do so for ones in the Baltic area in his map 8 wererefuted bySchmid(1977:315-316),whoshowed thatthough the particular etymologies of Krahe attacked by Tovar were not right, the first and third have lexical cognates within the Baltic languages - Tovar’s names from Basque etc. are therefore simply not equivalent - and the second does in other Indo-European languages, including Welsh cadr ‘strong’,Celtic catu- > W. cad ‘battle’.Tovar’s forms here, or some of them, may just possiby want adding to Continental material used in Rivet and Smith’s (1979:302-304) discussion of the name Catterick, earliest attested (in Ptolemy) as Katoura ktdnion ?7 More interesting to my mind is the possibility that some originally alien words were absorbed as major elements into the IndoEuropean naming-system. Tovar (1977 map 7) adduced one zyxwv Krahe involving adjectives of colour and position -which are among the commonest qualifiers in Old English place-names (Kitson 1993:34-41). and may well have been so in much earlier times. Compare note 79 below. The fragments of etymology from Basque produced out of a hat as it were by Vennemann at a late stage of his argument (pp. 260-262) are not a convincing substitute, not least because of collapse of chronology again. He does not properly meet the objections of Dr. R. L. Trask in his n. 82 on his ‘determiner’ suffix, and the derivative suffixes he mainly invokes are usually seen as being at least to some extent borrowed from Celtic in which they do have a convincing internal history (Russell 1988, esp. 165-169, and refs.). Vennemann‘s phrasing ‘always’ ( 1 994:26 1) ignores scholarly debate to which Russell (1 988: 166) gives references. This goes too in my opinion for the palaeontological theories which Vennemann reveals (pp. 2 15,263,etc.) effectively underlie his linguistic ones - in effect a new kind of autochthony since time out of mind. He propounds an axiom ‘Toponyms are rarely changed, they are merely adapted‘ (p. 264j, which known history ancient and modern of migrant conquerors shows to be untrue. ‘Since they all spoke closely related languages‘ (p. 263) begs large questions, made more explicit, if not necessarily answered, in Kitson (forthcoming). 47 An attractive new etymological suggestion by Hamp (1 993: 1 19) would however dissociate these from Catterick. zyxwvu zyxw zy zyxwv KITSON - RIVER-NAMES 99 plausible candidate. This one has plenty of examples in Britain, in rivers Carron, Carrant, and the like, though one must say Tovar’s symbols are not very accurately placed. He offered it as non-IndoEuropean altogether, but that won’t do, since Schmid (1977:316) pointed out that there are a whole row of Baltic examples Tovar missed; anyway what would speakers of a Mediterranean language be doing naming places in Norway? Yet it remains true as Tovar (1977:19-20) says that *kar(r)a ‘rock, crag, tor’ is ‘a standard example’ of Mediterranean vocabulary; and it is not necessarily true that the root to which Nicolaisen (1971:95) after Krahe assigns it:* that yielding English ‘hard‘which Mann’s dictionary gives us IEur. *kordhos, -us, ever existed in that sense without the dental consonant. How you evaluate this depends on your idea of IndoEuropean. If you operate with the loose notion of roots and extensions in the standard dictionary, Pokorny’s, the hydronymists’ etymology looks acceptable; if you prefer the discipline of exact hypothesis as in Mann’s dictionary, which will never be standard partly because he died before he could index it does not.50The two dictionaries tend to opposite kinds of error; either is a salutary corrective to unthinking belief in the editorial choices in the other. Pokorny’s approach on the whole fits better the early stage of IndoEuropean from which much in the river-naming system must originate, Mann’s, the late stage directly reconstructable from descendant languages; but Mann’s handling of the evidence is the 4x 14 kar- ‘hard, stony’. Rivers listed by Nicolaisen more or less identical with the Gloucestershire Carrant are three Carron (Stirling, 2 Ross), two Carron Water (Stirling, Kincardine), Burn of Carron (Banff), to which he adds Carrot Burn (Renfrew), White and Black Cart Water (Renfrew). Carey (Devon), Cary (Somerset), Ceri (Cardigan), Ceri (Radnor) Nicolaisen derives from the same root via aformation *karisu;Ekwall was less sure whether to posit for them adjectival derivatives of the British etymon of Welsh caru ‘to love’ or of the root *ker- ‘to move, run’ found in Latin currere, Welsh cerdded. “l Though one must warm to a man whose life’s work was ‘to avoid the German fault of overloading etymologies with irrelevancies, the French fault of trimming the evidence to achieve a neat pattern, and the Anglo-Saxon fault of being content with approximations’ (Mann 1987:viii). 5o The only serious evidence for it would be if Greek and Indian words in karkarmeaning ‘harsh, rugged‘ are as Pokorny thought reduplications of this root; but they and words like Skr. karkaiah Crow’, which he nevertheless would connect with them, seem much likelier to belong as Mann thinks with a separate onomatopoeic root kark-, to which some ‘normal-grade’ words in kerk- seem also related. The balance of probability thus is that words for ‘nut’etc. also beginning kar- are as Tovar (1977:20) and Mann take them likewise separate from ‘hard. zy zyxwv 1 00 zyxwvu zyx zyxwvu TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 94,1996 more convincing in this case. Personally I am inclined to agree with Ekwall that, for the majority of these river-names which are -ntformations, any nominal root is unlikely. As we have seen, he explained the Gloucestershire Carrant as ‘friendly, pleasant stream’ from the etymon of Welsh cum ‘to love’ (1928:70). That base is available for Indo-European, but I doubt if it would be used in names so widely. I prefer to see this river-name as being, as so many are, a reflex in u of the o-grade of a root in e , namely here *k& ‘to cut’. The reference in the case of the Gloucestershire stream will be to cutting new channels after flooding. It runs through flat muddy country for which Nicolaisen’s etymology is wildly inappropriate, to a confluence at Tewkesbury, the flood capital of England; substantial parts of its length show two or three parallel courses even on the one-inch Ordnance Survey map. The other rivers in Cur- do seem mostly to be in the kind of rocky country Nicolaisen’s etymology envisages; maybe there something to do with cutting gorges or glens would be in order. IV ‘Old European’ river-names have furnished evidence of a newly direct kind in the long-running dispute about the so-called Urheimut or ‘original homeland’ of the Indo-European-speakers. As is well known, a case to be taken seriously has been made for three, albeit argued often less on evidence than on romantic preference for what one’s ancestors should be. An origin in roughly Poland and Germany is preferred by those who like autochthony since time out of mind, in roughly Rumania and Bulgaria by those who like precocious urbanism, in the steppes north of the Black Sea by those who like vigorous migrating tribes brandishing battle-axes at effete sedentary folk. The three candidates are conveniently mapped by McEvedy (1 967, map for ‘4500B.C.’). The Pontic steppe hypothesis has been most believed in the English-speaking world because of energetic advocacy by two ideologically motivated archaeologists, the Marxist Australian V. Gordon Childe between the wars and the Lithuanian-American feminist Marija Gimbutas in recent decade^;^' but on a dispassionate view of the archaeology it is the zy zyxwvu Childe emphatically acknowledged himself an ‘exponent of Marxist prehistory’ (1 040[ 19791:93).A late paper of his outlines the part played by Marxist theory in his early reasoning on the ‘Aryan cradle’; regrettably I have not been able to verify the ’I’ zyxwv zyxw KITSON - RIVER-NAMES 101 weakest of the three. Evidence of loan-words into proto-Finnish from the very earliest stages of proto-Indo-Iranian tells also strongly against it: were the Pontic steppes the homeland of all IndoEuropean proto-Finnish should not have been in contact with the easternmost dialects until after however long it took for the rest to complete their migration into Europe.52On the traditionally used linguistic evidence,words for flora and fauna reconstructablefor the parent language, of which there are surprisingly many (e.g. Thieme 1958:72, 1964:596), the north-central European hypothesis is the strongest and the Balkan hypothesis weakest. Variants of the latter propounded in recent years, in Anatolia by Lord Renfrew who likes precocious farming, and in the Caucasus by Thomas V. Gamkrelidze who likes glottalic consonants as used in Caucasian languages,are not serious contenders. The contribution of river-names to this argument is that in Europe south of the Baltic and north of the Alps and Carpathians, between roughly the Rhine in the west and perhaps the Don in the east, all ancient river-names are etymologically alteuropukch. At least so say the hydronymists, and river-names in the area have been so intensively studied, and attempts to overturn the assertion have been so conspicuously unsuccessful, that I think we must take it as established.Further south and east, as well as west, some names are non-Indo-European. Those who believe in place-names as an index of Viking settlements or Jackson’sriver-nameareas as a guide to that of the Anglo-Sa~ons~~ will think it a fairly obvious inference that the homogeneously alteuropuisch area was inhabited by people of alteuropaisch, that is Indo-European, linguistic stock before the surrounding areas were. It agrees pretty well with maps that get drawn from distributions of flora and f a ~ n a . This 5 ~ aspect of rivernames has yet to receive the general acceptance which is its due, zy zy zyxwv zyxw documentationwhile preparing this article. Marija Gimbutas as far as I know did not thus openly admit modem ideology as a factor in her modelling of the (*re-)IndoEuropean past. All the same, critics such as Sergent (1982), though not impartial, seem right to have identified it as one. 5 2 See discussion of Harmatta (1978) in Kitson (forthcoming). 53 Or comparable things in most European countries. 5 4 Compare Kilian (1988)’s schematic pl. 5, fleshed out in the shading on pl. 66 (repr. Mallory 1989:255) with Mann (1943:76) or the version of that by Kilian (1988 pl. 4), or more schematically Thieme (1958:70). zyxwv 102 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 94,1996 zy zy partly because Krahe himself did not recognize that his etymologies entailed Common Indo-European antiquity for them; partly because W. P. Schmid, who did, has combined that recognition with theories making his own speciality, Baltic, the centre of everything; partly because hydronymists don't deign to produce anything so vulgar as a clear map of the area or a clear list of which names around it they grant as non-Indo-Eur~pean;~~ partly because of reluctance by general linguists and archaeologists dealing with something so emotive so remote in time to attach as much weight to the evidence of piace-names as they do to the kinds with which they are most familiar. I weigh the details of the different kinds of evidence, and their interactions, at length in a companion paper (Kitson, forthcoming); suffice it to say that on available evidence the north-central European Urheimat is in my opinion the only reasonable working hypothesis, and that those who would still argue for another need to find convincing explanations why the prima facie implications of borh river-names and flora and fauna should be wrong. They are not convincing so far in their attempts to explain away either, much less both. Granted at least approximately an Urheimat, and knowing approximateIy from linguistic reconstruction the early metal-using, mixed-farming, horse-drawn-wagon-riding,village-dwelling level of culture, then, on the null hypothesis usually adopted of rough correspondence between large archaeological and linguistic groups in the absence of evidence to the contrary, we should be able to identify with fair probability the material cuIture(s) whose bearers were speakers of Common Indo-European. The only material culture of the required level with a requisite spread across both western and eastern Europe is the one called either Danubian or zy zyx ' 5 Vennemann (1 994:266) quotes gossip to the effect that Krahe was actively antipathetic to mapping. This quirk is of a piece with the incongruity between Krahe's inclusion of Sanskrit river-names in his detailed analysis (1925a:1-3, cf. page 89 above) and his theoretical construct of alteuropuisch as a product only of western stocks, not Common Indo-European. Of course his followers have produced many maps, including such fascinating constructs as that of Udolph (1981 60);yet the finest cartography, that of Schramm (1973), serves linguistic categories not fully viable, and that symbolizes a curiously negative correlation between good maps and whole vision which stiil seems to persist in the hydronymic literature. One feels too the lack of any introductory monograph for non-specialists. Krahe (1964) is not fully enough focused on alreuropuisch or (despite its title!) even on river-names, and is aimed too exclusively at a German public. zyxwv zy KITSON - RIVER-NAMES 103 after its ceramic decoration Linear Pottery (or often German Linearbandkeramik), lasting for most of the sixth and fifth millennia B.C., and I echo archaeologists who have taken the linguistic evidence seriously56in thinking that the material culture essentially correspondingto Common Indo-European. It is possible using their maps of it5’to construct a proto-dialect map for Indo-European of a higher level of internal consistency linguistically than the constructs of scholars without particular archaeological evidence such as Schwarz (195 1:206 map 3), though it should be stressed that only at the south-eastern end is there positive reason to believe in the essential continuity of the sub-groups concerned as individual entities from these prehistoric times to their historical appearance. The main successor cultures, Funnel Beaker and Globular Amphora in the fourth millennium and Corded Ware in the early third:* would also belong linguisticallyto Common Indo-European rather than to any particular descendent language-groups, but it seems clear that they do not take in all the Indo-European-speakers of their periods. Funnel Beaker and Globular Amphora do not spread as far as Linear Pottery did east, south, or west; Corded Ware and associated groups spread east and south but not west. The first successor culture centred in, or at least overlapping with, the Urheirnat that does spread significantly west is that of the Beaker People already mentioned, of the late third millennium.Bell-beakers are in fact the on& archaeological phenomenon of any period of prehistory with a comparably wide spread to that of river-names in zyxwv + zy zy zyxwvut 56 I would mention here Jhos Makkay (1987, 1992), although he goes for a grandiose scheme involving all three Urheimat candidates.From the point of view of linguistic origins that is most implausible, and merely represents, as he freely admits (1992:193-194), a regress of one historical stage. His detail (e.g. 1987174-178, 1992:207-209) implies the north-central European as the ultimate heartland. 5 7 Most usefully I think Makkay (1987:74, 1992:195, 216) and Luning (1988:156). Less interpretative maps include Piggott (19655 1) and Kilian (1988 pl. 12), more interpretative ones Piggott (196557) and Kilian (1988 pl. 13); N.B. contradictions between the latter at the eastern end! The map of house types (Clark and Piggott 1970231) is of striking relevance, as is the very interesting one of early archaeological evidence for cattle by Benecke (1994:96), knowledge of which I owe to Dr. B. Maier. 58 Funnel Beaker is mapped e.g. by Piggott (1 965:63) and Kilian (1 988 pl. 25 after C. J. Becker), it and Globular Amphora schematically by Mallory (1989250); cf. Clark and Piggott (1970:86). Corded Ware groups are mapped by Kilian (1988 pl. 36) and Mallory (1989:247). To these and Linear Pottery the map of early evidence for wheeled transport by Piggott (1983:59) is also relevant. zyx zyxwv 104 TRANSACTIONS OFTHE PHILOLOGICAL SOCIETY 94,1996 the western half of Europe.59The presumption must I think be that Beaker Folk were the vector of alteuropuisch river-names to most of western Europe. Rivers in the base Arg-,which we have seen there is cause to think was not already in use at the earliest stage of the rivernaming system, and which therefore should be associated with such a vector if one existed:’ fit their distribution exceptionally we11.6’ That they were a single-speech commmunity can be asserted more confidently of the Beaker Folk than of most archaeologically identified groups for the very reasons that have caused archaeologists difficulty in interpreting them. As McEvedy (1967:28) put it, ’the bell-beakerfolk march convincingly in every prehistorian’stext, but they do so from Spain to Germany in some and from Germany to Spain in others, while lately there has been a tendency to make them go from Spain to Germany and back again (primary and reflux movements).’ One ‘firm datum seems to be that the British beaker folk came from the Rhine-Elbe region.’ This confirms what the long chronology now indicated for Common Indo-European would suggest anyway, and what to me, as remarked above, the rareness of non-Indo-European names in England suggests, that the old dissenting minority of Celticists were right to see the arrival in Britain of Indo-Europeans, as evinced in river-names whether or not in ethnic proto-Celts, as early as the third millennium. McEvedy’s map of Beaker Folk identifies them zy zyxw 5y Maps include Piggott (1965:101), Replogle (1980166), Kilian (1988 pl. 38 after V. Gordon Childe),McEvedy (1967 ‘1850 B.C.’).To these should be compared an archaeologicaldistribution,that of megalithic chambered tombs (Piggott 1965.61) whose linguistic correlates must clearly be non-Indo-European, and for ‘directional trading networks’ (p. 77 above) the one comparably ancient archaeological phenomenon that is certain to do with trade if any does, amber (e.g. F’iggott 1965:138), with its converse Mycenaean finds and influences in northern Europe (e.g. Piggott 1965:136). 6o Not necessarily exclusively: cf. ’Apyavn) in north-west India. If that comparison is valid despite the phonetic difficulty acknowledgedin note 17 above, it confirms that the theme Argant- became productive in a sub-Indo-European dialect continuum of which proto-Indo-Iranian speakers were still effectively part, i.e. beginning presumably not later than the early third millennium - which would fit the Beaker expansion very nicely. The gap in the Globular Amphora-Funnel Beaker heartland will then be either because alteuropaisch river-naming was already complete there or because rivers there that were once called Argant- came to be called by the competing theme Alb- which remained productive longer as discussed above. The latter looks likelier. 61 But for the gap mentioned in the preceding note where it overlaps the Funnel Beaker-Globular Amphora heartland. zyxwvut zyxwv zyxwvu zyxw KITSON - RIVER-NAMES 105 zyx zy zyxw linguistically with Celto-Ligurians, but in that his admirably tidy mind was, typically, a degree too tidy. Considerations of phonology indicate that more than one linguistic group was involved. It is normal in reconstructed Indo-European for groups of related words not all to have the same vowel in the root syllable. The commonest vowel gradation is between e, o, and zero; e is the commonest vowel in root syllables,by so much that the o and zero of ablaut-related words are supposed to have arisen out of e in most instances, by what mixture of accentual patterns and combinative sound-changes is disputed. It will not have escaped readers’ notice that the commonest vowel in root syllables of these river-names is a. It is routinely explained by the hydronymists as deriving from an Indo-European o-grade. That is in principle reasonable, since in many of the descendant languages original o and a have fallen together, usually as a. Another source of ci less generally recognized is the zero-grade of original long vowels (Burrow 1979); that is relevant to some of these roots6*- it accounts for the Thames group for example - but will not do for most. Language-groups that level short a and o include Germanic and Baltic, Slavonic,Illyrian, Hittite and Indo-Iranian; but Celtic and Italic like Greek and Armenian preserve the original distin~tion.6~ It follows that Celts speaking normal Celtic sounds cannot have been wholly responsible for bringing alteuropaisch river-names to any area. It would seem to follow, as Professor Nicolaisen has consistently urged, that in Spain, Gaul, Britain, and Italy, where the only historically known early Indo-Europeans were speakers of non-levelling languages, they were preceded by speakers of levelling languages not historically known. This hypothesis, pretty well required by the linguistic evidence, finds so good an archaeological correlate in the Beaker People that I think it would now be flying in the face of the evidence not to accept those as bearers of the river-names to these countries. Not the first in all parts of Gaul, and not necessarily the first in all parts of Britain, depending on how you relate such archaeological zyx zyxwv And notably to more basic words for water, since as Pokorny (1959-693 23) points out OE 2 p e a r d ‘ocean-watch’,eagor ‘river-bore,ocean flood’ and ON A g i r a sea-god imply *Ek@s and the like beside *ak@ Meillet (1937:98-100) is fuller on this than Szemerenyi (1989 0IV.l).All the levelling is to a except in Slavonic to 0:hence river-names like Oder < *Adam,with the occasional early form in A- (Krahe 1962:305). 106 zyxwvu zy zy zyx zy TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 94,1996 phenomena as the Windmill Hill culture of the late fourth and third millennium64to the western end of Linear Pottery earlier; but to the fairly large extent that these river-names over western Europe are a single system implying a single speech-community, the Beaker Folk must be that community and the people essentially responsible for the system. What the late Professor C.F.C. Hawkes called in British archaeology ‘cumulative Celticity’, built up by successions of comparatively small tribal migrations, will then have operated on the linguistic side as well. That the predecessors of the Celts proper for so long had in most of Britain been people of similar Indo-European speech explains why there is not a significant survival of recognizably non-Indo-European river-names,and why the fewserious candidates for non-Indo-European among recorded place-names all seem to be in Scotland. That the river-names kept their north European nonCeltic phonology will be because the Celts proper took them over as names, with denotative not fully lexical meaning. The contrast with a stratumin thenamesof historical Pictish kings,suchas Usconbutsand Bliesblituth which no-one would dream of trying to etymologize as Ind0-European,6~deserves emphasis. Royal personal names were presumably felt to be bound up with ancestral identity in a way which river-names were not. Island-names and tribal names are likely to have fallen somewhere between the two. Again the candidates for non-Indo-European cluster in Scotland;66 but their apparent zyx zyxw zyxwvuts zyxw 64 Mapped e.g. by Piggott (196558) together with Chassey-Cortaillod. What these cultures were linguistically is very debatable. McEvedy’s (1967) map for ‘2250 B.C.’ marks Windmill Hill as Indo-European, Chassey-Cortaillod not. The yardstick for judgement there is relation to the material culture of Linear Pottery; not all archaeologists would put the two on opposite sides of the fence. Various kinds of linguistic hybrid might well also be applicable. This is a convenient place to remark that the dates on McEvedy’s maps are not homogeneous. He used historical dates where available, otherwise radiocarbon, but that involves unevennesses not appreciated in his day. ‘2250 B.C.’ means if I have got the calibration right (Pearson 1987) c.2900/2800 in most of the map outside Egypt; ‘1850 B.C. = c.2300/2200 in the operative parts, ‘4500 B.C.’ between 5500 and 5000. 65 Commented on by Jackson (1955:145). The names, with manuscript variants some gross (this similarityto the data of note 4 above may be significant),are conveniently set forth by Chadwick (1949:7). The tribal names Tuexali in Aberdeenshire and Creones in north Argyll, the island-names Ebudae the Inner Hebrides and (H)Aernodae or Acmodae likeliest the Shetlands if not just scribal corruption of Ebudae. Of these Creones looks IndoEuropean in form even if Rivet and Smith cannot etymologizeit. ‘’ zyxwvu zy zy KITSON - RIVER-NAMES 107 phonology is sufficiently close to Celtic to raise serious doubt whether what is in question is not just exotic non-Celtic Indo-European. Rivet andSmith’s(l979:463-464)commentson Taexaligiveanideaofthe possibilities for several. Nothing in the recorded river-names is of as exotic phonology even as Tuexali. These examples make the point that the correspondence between language and ethnicity is not only not the same in names as in ordinary vocabulary, it is not even the same in different nameregisters. It would be possible to apply that principle a stage further back and construct hypotheses whereby the bringers of alteuropuisch river-names to Britain were after all in their ordinary speech proto-Celts. For this to be true one or both of two things would be necessary. One is that this river-naming was in use earlier among the northern and eastern Indo-European sub-groups that levelled o and a than it was among the ancestors of Greek-, Celtic-, and Italicspeakers. The relatively high degree of survival of these roots at the Indo-Iranian end would then only partly connote earliness of origin, in addition area of origin within Common Indo-European. The phonetic feature of a in o-grades in river-names would then have spread dialectally pari passu with the river-name types themselves. The other possibility is the phonaesthetic one that the vowel a was felt especially suitable for water-words. This is not as silly as it sounds, because it would grow out of what Meillet (1937:99)has said of Indo-European *a in general ‘la voyelle *a se trouve notamment dans des mots de caractkre populaire, technique, ou affectif.’ Popular, technical, and affective: river-names might simultaneously be all three. There is probably a thesis to be written on phonaesthesia in Indo-European water-words. It is notable how not only most of the river-namesbut also most of the lexical items on relevant roots have vowels other than e. Whether your point of comparison in the ordinary vocabulary of Common Indo-European is the word for ‘water’ in its more flowing manifestations *ak@ or the more general substance * wod6r, it is rather full of back vowels!67 The possibilities of non-Celts, dialectal mix, and phonaesthesia are not mutually exclusive:you can have various combinations of the zy zyx zyxw h7 An original nominative *wedGr is assumed in most of the theories canvassed by Szemerenyi(1972:159) to explain the variation in this heterocliticword; but if real its lack of visible reflex in descendant languages is likely to mean it was levelled out of the paradigm early. 108 TRANSACTIONS OF T H E PHILOLOGICAL SOCIETY 94,1996 three. But I think non-Celtic Indo-European-speakers are likely to have been involved in fact, whether or not they are the whole story, both because that it is the hypothesis which makes best sense of the archaeologicalevidence (the greater population density of a farming culture than a hunting-gathering one is likely to have led to the linguistic strain of the westmost group of Linear Pottery predominating in the mixed cultures which sprang from that and its mesolithic neighbours) and because it is widely acepted that placenames in the Low Countries imply the existence of at least one group of not historically attested Indo-European-speakers,68not the same as the ones we are concerned ~ i t h . So 6 ~do names in Spain, another country where the only historically attested early Indo-Europeans were Celtic. Comparing Spanish alteuropuisch names with British ones gives a glimpse of the dialectal range that must have characterized the Beaker phenomenon. Either group shares one feature with historical Celtic that the other lacks. The Spanish names like Celtic proper mostly keep Indo-European 0.There the diagnostic feature is initial p (Schmoll 1959:93,78-80; Rodriguez 1980), lost from Celtic and the alteuropuisch of Britain. Whether the first alteuropuisch river-namers in this country were non-Celtic Indo-European-speakers or proto-Celts with a north European cultural package, one would like to be able to discover other names of theirs. Telling them apart from Celtic formations proper may never be a very exact science, except for the names revealed phonetically by o-grades in a; but there are some plausible pointers for some items. One might well expect the first immigrants to name the biggest southern rivers. The root Sab- in Severn forms some river-names in Gaul and Ireland as well as what is now a forestname Savernake;but Rivet and Smith (19795 1)say ‘No clear Celtic etymon is identifiable’, so that is a candidate.’O The alteuropiiisch zyxwv zyxw zyx zyxw zyxw zyx zyxwvu zyx Kuhn (1962); cf. Meid (1984). One or other group of such people would be enough to explain the voiceless fricatives that surprise Vennemann (1994:243) in some names, which he does not exemplify, if any are as he asserts inexplicable from the known history of Germanic (and Gaulish) and their speakers. He is either naughty or ignorant to call his explanation ‘the only one in existence’ (p. 244). 6y The common ‘Nordwestblock’stream-namingelement *apa does not seem to be found in Britain, though the *ab- of Welsh afon,Latin arnnis is related to it as mentioned above, a relation not shared by neighbouring Germanic. The -st- suffix productive in ‘Nordwestblock‘ names (at a later period than *apa to judge by their distribution) is also not conspicuous in British ones. “I The new suggestion of Hamp (1988:7, 1990:139) is phonetically unconvincing. zyxwv zy zy zyx zyxwv KITSON - RIVER-NAMES 109 etymon of Thames we have seen. That it was a non-Celtic rather than proto-Celtic name may be indicated by the unusual suffix-formation Tumesis paralleled in the Spey Tuesis, with which goes the Tweed whose name was something like * Tgesetk7’ The s in these last two names may well be, as that in Thames may be, a suffix already present in the lexical base, but if so that strengthens the impression of connections with north European languages and Sanskrit rather than Celtic.’* The geography of surviving names may give useful hints too. Nicolaisen (1976:177-178) lists fourteen different formations in the suffix -no-l-ni in Scottish river-names, some in more than one river. That is one of the more productive suffixes in Krahe’s (1962:296+) table, but it is remarkably rare in English river-names. Most that look as if they might contain it turn out either to have a Germanic etymology, or documentably different suffixes like Aluunos which we saw earlier, or to be formed on Celtic lexical items already containing n , like Leadon (Gloucs), Lidden and Lodden (Dorset) corresponding to Welsh llydun ‘broad’, or Teign derived by Ekwall from *tugnli etymon of Welsh tuen ‘a sprinkling’. When those categories are excluded the likely residue are only Devon (LeicdNotts), one of whose two possible etymologies would make it the same as the Scottish Devons; Lemon (Devon), Leam (Northants), Lymm (Lincs), Lympne (Kent) identical with the Scottish Levens; Olchon (Herefs); Roden (Shropshire); Seven (NRYorks); and S e ~ e r nThat . ~ ~ there should be less than half as many in England as in Scotland is remarkable. This is not a function of Anglo-Saxon replacement of older names; the rareness is as striking within a purely English context. One would expect that for some considerable time in prehistory pre-Celts relative to Celts in Britain would have had a fringe position roughly like the Celtic lands relative to England now, or perhaps better the king’s territory to Parliament’s territory during the Civil War. Replacement at that zyxwvu zyxwv Adapting the quite adequate etymology given by Ekwall(1928:423) slightly in an alteuropaisch direction. 72 The base given by Ekwall’s is *teui-, *tu- ‘to swell, be powerful’ as in Sanskrit t u v h ‘powerful’, tavisi‘power’,ON )kktr ‘violence’,Gothic bris (in )&shundi) etc.; cf. the Sanskrit and Lithuanian words cited in connection with Thames above. 73 Also possible would be Duddon (Cu/La), whose etymology is not known, but it looks suspiciously like many Old English formations with qualifying elements Duddun and Dudding. Correspondence of Leen (Nt), Leon (He) with Welsh lliant ‘flood(s)’, and of Loddon (HaBrk), Loddon (Nf)?, Lodon (He) with Gaelic ldn ‘marsh’, suggests that n was in their etymons. zyxwvu zyxwv 110 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 94,1996 stage might produce the observed distribution. There may be other explanations that would work here, some perhaps better, but 1draw attention to this as a phenomenon that at any rate seems to need explanation, and to the mode of argument as one that may uncover chronological distinctions in some cases. Most of the formations in question have adequate Celtic roots, but most would have been available in neighbouring dialects as well, and some of the formations are paralleled on an alteuropuisch scale. The Lymne group, commonly etymologized from the Celtic word for elm, are one of them. Some may individually be so formed, but I think the Celtic tree-meaning is likely to be as secondary with that group as it is with the Denvents. V Some typologically minded linguists (e.g. Vennemann 1994:244252 and refs., Kuiper 1995:72-76) have used the great frequency of a as compared to e in the river-names as a pretext for holding them not Indo-European at all. That without examination of possible causes is not a reasonable objection, especially given the similar frequency in relevant lexical items, and in linguistics as in other subjects people who operate on that level of abstraction, whether motivated as structuralists or deconstructionists, tend not to be convinced by anything so mundane as detailed evidence; yet I think the hydronymists’ failure to address the objection even if the objection is unreasonable is one reason why they have made so little impact in English-language scholarship. The prevalence of o-grades over ‘full’or ‘normal’e-grades has in fact a perfectly rational explanation. One is accustomed to think of names as nouns; and perhaps this has been assumed for alteuropuisch river-names by the hydronymists as well as by their opponents. But from an Indo-European morphological point of view the most economical explanation of them is as not nouns but adjectives. The simple consonantal suffixes are all and only those productive in Indo-European adjectives;the complex ones are those used to form participles (which of course are a kind of verbal adjective) and superlatives. This is a decisive point against Vennemann’s view (1994:33) that ‘the suffixes themselves do not look particularly Indo-European’, even on his premiss admitting to zyxwv zyxw zy zyx zyxw zy KITSON - RIVER-NAMES 111 the argument only ‘their phonetic substance - the meanings are unknown’. Indo-European explains which suffixes do not occur as well as accounting for those that do.74 Stress in adjectives as in other parts of speech was levelled variously according to the accentual types of the descendant languages (Lubotsky 1988:15 etc.), but it is likely that all the main kinds of adjectives, including all those represented in river-names, wereoriginally accented on thesuffix(Hirt 1921-37:V§$173-177). So in an adjectival subsystem operative since early in the history of Indo-European the vowel-grades appropriate to reduced stress would have been n0rmal.7~Zero-grade is less common than ograde, but it is represented, e.g. in the Is- of names like Aire, Ure, and Isere. Suffixal stress in adjectives may well have been a function of the suffixes not of the adjective category as such (Lubotsky 1988:l-2), so that explanation for the vocalism could still apply, though less tidily, if the river-names were after all primarily nouns, with suffixes like -nt- which become exclusively adjectival only in a secondary stage in their development. Adjectival origin is still indicated by other aspects of them for which it offers rational explanationswhere the previously existing hypotheses as far as I can see do not. Adjectival nature explains why variants with -i- occur freely, indeed preponderantly, with the two suffixes that are solely participial, and only those: they are not nouns directly formed from participles but derivative adjectives analogous to Greek ~ ~ O O J U L‘being OS of one substance’ besides feminine participial aha, just ‘being’. It explains76why despite the well-known fact that Indo-European used zyxwvu 74 Either Vennemann’s Finnish inventory (p. 233) or what he more seriously argues for, the Basque one (pp. 255-262) should generate suffixes that do not occur, e.g. surely a voiced -g-(p. 260) if the hydronymic language were seriously related to Basque on his account of it. Additional weaknesses of his argument are that he does not present (or point the reader to) a whole picture of the suffixal inventory of any one language, nor of the group he argues for as a whole, nor does he discuss what in his favoured language-group the suffixes actually observed in the river-names would mean. Nor of course does his hypothesis yield any explanations for observed frequencies such as those furnished by Indo-European participles (p. 82 above). 7s It is again naughty of Vennemann (1994244) to call his laryngealist explanation ‘the only kind of explanation that I know’. At least he does not quite go so far in his laryngealism as to posit a proto-Indo-European in which the vowel a never existed, as Kuiper does. 76 If explanation is needed. The preponderance of dithematics in early recorded personal names is well known. I do not know a frame of reference of early enough zyxwvu zy zyxwvu zyxw 112 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 94,1996 dithematic forms in proper names from the beginning, as well of course as in lexical items including adjectives, the river-names are always monothematic with derivational morphology confined to extensions.The absent second theme is a noun always implied, never expressed. The implied noun meaning ‘running water’ would be normally *akz@:that even explains the great preponderance in the hydronymy of feminine -a over masculine -0s. Lines of explanation startingfrom theories of the origin of the &declension as such fail to carry conviction;” nor is help to hand from ancient mythologies, in which masculine river-gods as well as feminine nymphs are prominent. Adjectival derivation using these suffixes explains why the rivernames are restricted to these two declensions to an extent that would be surprising in a random sample of Indo-European nouns this large; the only exceptions acknowledged by Krahe (1962:296)seem to be the rule-proving ones of consonant-stems in -nt-, though there are i-stems in Britain.78All this tends incidentally to strengthen the view expressed in connection with names like Thames above that the great number of roots translated more or less interchangeably ‘flow’ by the hydronymists originally meant distinguishable kinds of fl~wing.’~ zyxw zy recorded geographical names other than river-names to make it quite certain that the restriction to extended monothematic form is an aspect of river-names as such not shared by geographical names in general. That the class of names with which, in the absence of other early enough recorded geographical names, river-names are most often compared, divine names, did from earliest times include dithematic formations (whether or not close compounds) is however demonstrated, by Jupiter; and the balance of probability must be that geographical names did as well. ” Of those canvassed by Szemerenyi (1 972:153-154; cf. 19859-20) the only one the river-names would fit is that of Gagnepain (1 960) that nouns in -6 were originally substantives formed from adjectives. ’x Tamesis Thames, Tuesis Spey, perhaps Tuerobis (?) Teifi and Toesobis (?)-see note 4 above. Krahe’s (1962:329) MQpts alias M ~ ~ I OinOIllyria S looks a good candidate to be put with these. That pair of forms and the relation of Tgesis to *Tgeseta Tweed might raise possibilities of consonant-stems being reinterpreted as i-stems; nstems, if there ever were any, might like -nf- stems have had their stem-consonants reinterpreted as suffixes, and there might be rare u-stems not possible to tell from o-stems with the extant material. But reinterpretation of -nt-stems if early and of the others if real presupposes a sense that river-names ought to be restricted to the o- and &declensions; and however much one explains away the other declensions the preponderance of u-stems can do with explanation. As one would anyway suppose on general grounds. ‘People of a more primitive period have a richer and more varied vocabulary for the natural objects with which zyxwv zyxw KITSON - RIVER-NAMES 113 zyx Names that aren’t nouns but adjectives are not an unparalleled concept. Krahe (1964:24-25) actually pointed out some German monothematic river-names that are formally adjectives.He thought them a more recent stratum than ones that are nouns, but no compelling reason for belief in such a chronological distinction is apparent. One at least of his names, Low German Lude, dates back as a type to before the Anglo-Saxon settlement of England, since it corresponds to OE Hljde, name of some dozen known streams in England. Ekwall(1928:273) following earlier writers thought fit to ‘assumean OE common noun hljde with the meaning “torrent,swift stream”.’ Were it really a common noun attestation only as a rivername would be unlikely. This is another place where Krahe’s analysis improves on Ekwall’s.soConversely i- mutation consistent in the Old English modifies Krahe’s for this name, showing it to be not identical with the lexical adjective hlud ‘loud’but an i-derivative of it in the manner of the ancient names. I should prefer to see Krahe’s adjectival names as just the most recent of a stratum continuously present in the onomasticon since ancient times. Be that as it may, their mere existence clinches the illegitimacy of ignoring overwhelming evidence for Indo-Europeanness of rivernames because they do not fit preconceptions about phonetics of Indo-European nouns. The linguistic material of the alteuropakch river-names is Indo-European, and they must be analysed rationally on that basis. I trust that I have somewhat advanced that analysis in the above. zyxwv zyxwvuts School of English, University of Birmingham, P.0. Box 363, Birmingham B15 2TT they come into contact, than people of a later era. The various words probably denoted streams of different size, swiftness &c.’ as Ekwall(1928:l) put it. Compare note 45 above. In Ekwall’s favour it should be said that several charters use phrasing’the Hljde’ with definite article, unusually for Old English river-names. But they are concentrated in the south-west midlands, and a more likely analysis than Ekwall’s would be that Hwiccean dialect was reinterpreting the inherited river-name as a noun. A real possibility too is phonaesthetic fashion: a stream-name Hyle in Gloucs S414(i) also has the definite article. zy zyxwvu 114 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 94,1996 zyxwv zyxwvuts zyxwvutsr zyxw zyxwvu zyxw REFERENCES BAHLOW, H. 1965. DeutschlandsgeographischeNamenwelt: Erymologisches Lexikon der Flu& itnd Ortsnamen alteuropaischer Herkunft, Frankfurt am Main: Klostermann. N. 1994. Der Mensch und seine Haustiere. Die Geschichte einer BENECKE, jahrtausendealten Beziehung, Stuttgart. BUCK,C. D. 1949. A Dictionary of Selected Synonyms in the Principal IndoEuropean Languages, Chicago: University Press. BURROW, T. 1979. The Problem ofShwa in Sanskrit, Oxford: University Press. BRUGMANN, K. 1886-92. Grundriss der vergleichenden Grammatik der indogermanischen Sprachen, Strasbourg: Triibner. CHADWICK, H. M. 1913.‘SomeGermanriver-names’,in Essaysandstudies Presented to William Ridgeway on his Sixtieth Birthday 6 August 1913, Cambridge: University Press, 1-8. CHADWICK, H. M. 1949. Early Scotland, Cambridge: University Press. CHILDE, V. G. 1949.’Prehistory and Marxism‘, Antiquity 52,93-95 (19791. G. and PicGoTT, S. 1970. Prehistoric Societies, Harmondsworth: Penguin. CLARK, COATES, R. 1984. ‘Remarks on “pre-British in England with special reference to *uenti, *&Ira and *cunico-’, JEPN 16, 1-24. COATES,R. 1987. ‘Pragmatic sources of analogical reformation’, Journal of Linguistics 23,319-340. CRAWFORD, 0. G. S. 1928. Review of Ekwall (1928). Archaeological Journal 85, 197-200. N. K., 1967. The Celtic Realms, London: Weidenfeld DILLON, M. and CHADWICK, and Nicholson. EKWALL, E. 1928. English River-names, Oxford: University Press. W. 0. 1963. Y Geiriadur Mawr, 3rd edn., Llandybie: EVANS, H. M. and THOMAS, Llyfrau’r Dryw & Gwasg Aberystwyth. FLEURIOT, L. 1991. ‘Celtoromanica in the light of the newly discovered Celtic inscriptions’, Zeitschriftfir celtische Philologie 44, 1-35. FORSTEMANN, E. 1913-16. Altdeutsches Namenbuch. I1 Ortsnamen, ed. H. Jellinghaus, 2 vols., Bonn; Repr. Munich: Olms, and Hildesheim: Fink, 1967. FORSTER, M. 1924. ‘Der Name der Donau’, ZeirschriFfur shvische Philologie 1 , 125; ‘Zum Donau-Namen’, ibid. 4 18. FORSTER, M. 1941. Der Flussname Themse und seine Sippe, Sitzungsberichte der Bayerischen Akademie der Wissenschaften, philosophisch-historisch Abteilung 1941.1. J. 1960. Les Nonlsgrecs en -0s et en -a, Paris: Klincksieck. GAGNEPAIN, GELLING, M. 1978. Signposts to the Pust: Place-names and the History of England, London: Dent. G E L L I N G , M. 1984. Place-names in the Landscape, London: Dent. GPC 1950-. Geiriadur Prifygol Cymru, Cardiff: University of Wales Press. HAMEL, A. G. V A N 1937. ‘La racine yen- en celtique et germanique’, in L. Hjelmslev et al. (eds), Mdanges linguistiques offerts a H. Pedersen, = Actu Jurlandica: Aarsskrifr for Aarhus Universitet 9.1,103-109. HAMP, E. P. 1984. ‘MiscellaneaCeltica’, Studia Celtica I8/ 19, 128-1 34. E.P. 1988. ‘Morphological interpretations’, Studio Celtica 22-23, 7-9. HAMP, HAMP,E. P. 1990. ‘Addenda to SC 22-23, 7 - 9 , Studia Celtica 24-25, 139. E.P. 1991. ‘On the paradigm of Oh. dii’, Zetschrifrfitrceltische Philologie 44, HAMP, 76-78. zyxw zyxwv zy zyxw zyxwvu zyxwvuts KITSON - RIVER-NAMES 115 HAMP,E. P. 1992. ‘Some toponyms of Roman Britain’, Studia Celtica 26-27,15-20. HAMP,E. P. 1993. ‘Nodiadau amrywiol - Miscellaneousnotes’, Bulletin of the Board of Celtic Studies 40,118- 119. HARMATTA, J. 1978. ‘Migrations of the Indo-European tribes’, Acta Antiqua Academiae Scientiarum Hungaricae 26,185-1 94. HIRT,H. 1921-37. Indogermanische Grammatik,Heidelberg: Winter. JACKSON, K. H . 1953. Language and History in Early Britain, Edinburgh: University Press. JACKSON, K. H . 1955. ‘The Pictish language’, in F. T. Wainwright (ed.), The Problem of the Pica, Edinburgh:Nelson, 129-166. JACKSON, K. H . 1970. ‘Romano-Britishnames in the Antonine Itinerary’,Britannia 1, 68-82. JOHANSSON, K . F. 1894. ‘Uber skr. adbhyh, abdhis‘, Indogermanische Forschungen 4,134-146. KARTTUNEN, K. 1989. India in Early Greek Literature, Studia Orientalia LXV, Helsinki:Finnish Oriental Society. KILIAN, L. 1988. Zum UrsprungderIndogermanen,Bonn: Habelt. KITSON,P. R. 1993. ‘Quantifying qualifiers in Old English local names’, Folia Linguistica Histonca 14,29-82. KITSON,P. R. (forthcoming)‘Reconstruction,typology, and the “originalhomeland of the Indo-Europeans’,in J. Fisiak (ed.), Linguhtic Typology and Reconstruction, Berlin: Mouton de Gruyter. KRAHE,H . 1950a. ‘Alteuropiiische Flussnamen’, Beitrage zur Namenforschung 1, 24-5 1. KRAHE,H . 1950b. ‘AlteuropaischeFlussnamen’, Beitrage zur Namenforschung 1, 247-266. KRAHE,H. 1951a. ‘Alteuropiiische Flussnamen’, Beitrage zur Namenforschung 2, 113-131. KRAHE,H . 195lb. ‘AlteuropaischeFlussnamen’, Beitrage zur Namenforschung 2, 217-237. KRAHE,H . 1952a. ‘AlteuropiiischeFlussnamen’, Beitrage zur Namenforschung 3 , l 18. KRAHE,H . 1952b. ‘AlteuropaischeFlussnamen’, Beirruge zur Namenforschung 3, 153-170. KRAHE,H . 1952c. ‘AlteuropaischeFlussnamen’, Beitrage zur Namenforschung 3, 225-243. KRAHE,H. 1953a. ‘Alteuropiiische Flussnamen’, Beitrage zur Namenforschung 4, 37-53. KRAHE,H. 1953b. ‘AlteuropaischeFlussnamen’, Beitrage zur Namenforschung 4, 105-122. KRAHE,H. 1953c. ‘Alteuropaische Flussnamen’, Beitrage zur Namenforschung 4, 234-243. KRAHE,H . 1954. ‘AlteuropaischeFlussnamen’, Beitrage zur Namenforschung 5,97114. KRAHE,H. 1955. ‘Alteuropaische Flussnamen. Exkurs: Der Flussname Anria’, Beitrage zur Namenforschung 6,l-13. KRAHE,H. 1962. Die Struktur der alteuropaischen Hydronomie, Akademie der Wissenschaften und der Literatur in Mainz: Abhandlungen der Geistes- und SolzialwissenschaftlichenKlasse 5, Wiesbaden:Steiner,= pp. 285-342. KRAHE,H. 1964. Unsere altesten FluJnamen,Wiesbaden:Harrassowitz. KRETSCHMER, P. 1925. ‘Das nt-Suffix’, Glotta 14,84-106. KUIPER, F. B. J. 1995. ‘Gothic bagms and Old Icelandic ylgr’, NOWELE 25,63-88. zy zyxwvut zyx zyxwvutsr zyxwvu zyxwv zy zyxwvu 116 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 9 4 , 1 9 9 6 KUHN, H. 1962.‘Das Zeugnis der Namen’, in R. Hachmann, G. Kossack & H. Kuhn (eds), Volkerzwischen Germanen und Kelten, Neumiinster:Wachholz, 105-128. LUBOTSKY, A. M. 1988.The System of NorninalAccentuation in Sanskritand ProtoIndo-European,Leiden: Brill. LUNING, J. 1988.‘Zur Verbreitung und Datierung Bandkeramischer Erdwerke’, Archaologisches Korrespondenzblatt 18,155-158. MAKKAY, J. 1987.‘The Linear Pottery and the early Indo-Europeans’, in Skomal and PolomC (1987:165-190). MAKKAY, J. 1992. ‘A neolithic model of European prehistory’, Journal of IndoEuropean Studies 20,193-238. MALLORY, J . P. 1989. In Search of the Indo-Europeans, London: Thames and Hudson. MANN,S. E. 1943.‘The cradle of the “Indo-Europeans”:linguistic evidence’, Man 43,74-85 (repr. in German adaptation as ‘Die Urheimat der “Indoeuropaer”’,in Scherer 1968:224ff.). MANN, S. E. 1987.An Indo-European Comparative Dictionary, Hamburg: Buske. A. 1957-9. Die Sprache der alten Illyrier, 2 vols. Vienna: Rohrer. MAYER, C. 1967. The Penguin Atlas of Ancient History, Harmondsworth MCEVEDY, Penguin. MEID,W. 1984. ‘Hans Kuhns “Nordwestblock-Hypothese’,Anzeiger der osterreichischen Akademie der Wissenschaften, philologische-historische Klasse, 21, 2-21. MEILLET, A. 1937.Introductiona l’etude comparative des langues indo-europkennes, 8th edn., Paris: Hachette. NICOLAISEN, W. F. H. 1956.‘Die alteuropiischen Gewassernamen der Britischen Hauptinsel’,Beitrage zur Namenforschung,8,209-268. NICOLAISEN, W . F.H.1971.‘Great Britain and Old Europe, Namn och Bygd 59,85105. NICOLAISEN, W. F. H. 1976.Scottish Place-names,London: Batsford. NICOLAISEN, W. F. H. 1980.‘Onomasticdialects’, American Speech 55,36-45. 6 M ~ I L LT. E ,S. 1987.‘Venta,Gwenta,Finn, Guen’, Nomina 11,145-151. OTTEN, H. 1969.‘Die Berg- und FluRlisten im Hisuwa-Festritual’, Zeitschrift fur Assyriologie und vorderasiatische Archaologie n.f. 25,247-260. PADEL,0.J. 1985a.Cornish Place-Name Elements, English Place-Name Societyvol. 56-57. PADEL, 0. J. 1985b.‘Cornish surnames in 1327’,Nomina 9,81-87. PEARSON, G. W.1987.‘How to cope with calibration’,Antiquity 61,98-103. PEDERSEN, H. 1909-13. Keltische Grammatik, 2 vols. Gottingen: Vandenhoeck & Ruprecht. PIGGOTT, S. 1965.Ancient Europe, Edinburgh University Press. PIGGOTT, S. 1983.The Earliest Wheeled Transport,London: Thames and Hudson. PINNOW, H.-J. 1953.‘Zu den altindischen Gewassernamen’, Beitrage zur Namenforschung 4,217-234. PINNOW, H.-J. 1954.‘Zuden altindischen Gewassernamen’, Beitrage zur Namenforschung 5,l-17. POKORNY, J. 1959-69. Indogermanisches etyrnologisches Worterbuch, Berne: Francke. REPLOCLE, B. A. 1980,‘British and German bell-beaker burials’, Journal of IndoEuropean Studies 8,165-1 99. RIVET, A. L. F. and SMITH, C . C . 1979.The Place-namesofRoman Britain, London: Batsford. ROOR~GUEZ, M. S. 1980. Toponimia de origen indoeuropeo prelatino en Asturias, zyxwvutsrqp zyxwvu zyxwvuts zyxwvut zyxwvu zyxwvuts KITSON - RIVER-NAMES 117 Oviedo: Instituto de Estudios Asturianos [known to me only from the review by Leon Fleuriot, Etudes Celtiques 22 (1985), 3741. ROSENKRANZ, B. 1966. ‘FluB- und Gewassernamen in Anatolien’, Beitrage zur Narnenforschung n.f. 1,124-144. RUSSELL, P. 1988. ‘The suffix -ciko- in continental Celtic’, ktudes Celtiques 25,131173. SCHERER, A. 1968. Die Urheimat der Indogermanen, Darmstadt: Wissenschaftliche Buchgesellschaft. SCHMID, W. P. 1968. Alteuropaisch und Indogermanisch. Akademie der Wissenschaften und der Literatur in Mainz, Abhandlungen der Geistes- und Sozialwissenschaftlichen Klasse 1968,6, Wiesbaden: Steiner. SCHMID, W. P. 1970. ‘Indisch-Iranische Appellativa und alteuropiiische Gewassernamed, in M. Boyce and I. Gershevitch (eds), W. B. Henning Memorial Volume, London: Lund Humphries, 376-384. SCHMID, W. P. 1977. Review of Tovar (1977), lndogermanische Forschungen 82, 314-3 17. SCHMOLL, U . 1959. Die Sprachen der Vorkelten Hispaniens und das Keltiberische, Wiesbaden: Harrassowitz. SCHRAMM, G. 1973. Nordpontische Strome: Namenphilologische Zugange zur Friihzeit des europaischen Ostens,Gottingen: Vandenhoeck & Ruprecht. SCHWARZ, E. 1951. ‘Germanen, Italiker, Kelten, Zeitschriffur Mundartforschung 20, 193-206+. SERGENT, B. 1982. ‘Penser - et ma1 penser - les Indo-Europeens’, Annales: &onomies, Sociitis, Civilisations 37,669-68 1. S. 1988. ‘The archaeology of Indo-European: an SHERRATT, A. and SHERRATT, alternative view’, Antiquity 62,584-595. S. N. and P O L O ME. ~ ,C. (eds) 1987. Proto-Indo-European:the ArchaeSKOMAL, ology of a Linguistic Problem: Studies in Honor of Marua Gimbutas,Washington: Institute for the Study of Man. A. H. 1964. The Place-names of Gloucestershire,part I , English Place-Name SMITH, Society Vol. 38. SZEMER~N 0. YJ.I ,S. 1954. “Illyrica’, Zeitschrift fur vergleichende Sprachforschung 71,199-217. SZEMER~N 0. YJ I. ,S. 1972. ‘Comparative linguistics’, in T. A. Sebeok (ed.), Current Trends in Linguistics,vol. 9.1, The Hague: Mouton, 119- 195. SZEMERENYI, 0.J. S. 1976. ‘Problems in the formation and gradation of Latin adjectives’,in A. Morpurgo Davies and W. Meid (eds), Studies in Greek,Italic, and Indo-European Linguistics Offered to L. R. Palmer on the Occasion of his Seventieth Birthday,Innsbruck: Institut f u r Sprachwissenschaft, 401-424. 0. J. S. 1985. ‘Recent developments in Indo-European linguistics’, SZEMERENYI, Transactions of the PhilologicalSociety1831, 1-71. SZEMER~N 0.YJ.I ,S. 1989. Einfiihrungin die VergfeichendeSprachwissenschaft,3rd edn., Darmstadt: Wissenschaftliche Buchgesellschaft. THIEME, P. 1958. ‘The Indo-European language’, Scientific American October 1958, 63-73. THIEME, P. 1964. ‘The comparative method for reconstruction in linguistics’, in D. Hymes (ed.), Language in Culture and Society, New York: Harper and Row, 585598. A. 1977. Krahes alteuropaische Hydronornie und die westindogerrnanischen TOVAR, Sprachen, Sitzungsberichte der Heidelberger Akademie der Wissenschaften, Philosophisch-historischeKlasse 2, Heidelberg: Winter. TRUBAEEV, 0. N. 1985. ‘Linguisticsand ethnogenesis of the Slavs: the ancient Slavs as zyxwvut zyxw zyxwvu zyxwvu zyxwv zyxwvut 118 TRANSACTIONS OF THE PHILOLOGICAL SOCIETY 94.1996 evidenced by etymology and onomastics’. Journal of Indo-European Studies I 3, 203-256. UDOLPH, J., 198 1 . ‘Zur fruhen Gliederung des Indogermanischen’, Indogermanische Forschungen 86,30-70. VENNEMANN, T. 1994. ‘Linguistic reconstruction in the context of European prehistory’, Transactions of the PhilologicalSociety 92, 2 15-284. WATKINS, C. 1973. “‘River” in Celtic and Indo-European’,Eriu 24,80-89. ZELAZNY, R. 1967. Lord oflight, London: Panther, 1971.