A SECOND OLD EUROPE
[Namn och Bygd 59, 1971, S. 52—66; 
print version of a lecture, made at the VI. Nordiske navneforskerkongres 
(Helsingør, 23.—26. 8.1971)]

The title of my lecture presupposes that I recognize that there already exists one Old Europe for the place name researcher. It's the legacy of Hans Krahe known as the Old European river name system, the significance of which to the Nordic countries is one of the main themes of this congress. But my recognition of this Old Europe is characterized by many restrictions and modifications. Of these, three are important here, viz. that Krahe's system does not the merit the position he gave it, that the borders he drew for it, constantly modifying them, can hardly be correct, and that it does not make up the oldest layer of river names recoverable by us (cf. ref. 1, 320 ff.). There exist many more other river names from prehistoric times and part of them form a rather homogenous group which just as much merits being called a system as the core of the great name group which Krahe has collected and ordered. This is the "second Old Europe" we will discuss.

We must in the Nordic no less than in the other Germanic countries take account of prehistoric names which are not only non-Germanic, but also non-Indo-European. For also in those countries, as is well-known, people have lived and of course given names too in millenia before the invasion of the Indo-Europeans. But the long source-less time, and in particular the radical transformation of the Nordic languages before the beginning of the literate period would have erased most traces of them, so that it would seem almost hopeless to reach some tangible results. In spite of that, when I appr. 15 years ago began to turn my attention to these questions, I became aware of a slowly increasing number of names also in Scandinavia, most of them in Norway, which by their phonology, morphology or the distribution of their stem appeared to be pre-Germanic. Separating whole groups of them in which one case supports the other, will only succeed if we compare the oldest Nordic onomastic material systematically with that of other countries, namely those which have shown themselves particularly rich in very old onomastic material. This I have attempted on a clearly circumscribed material.

In this endeavour I took, somewhat at random and at first without having the Nordic area in mind, as a starting point Dur-, a river name stem widely distributed in Europe, to which belong among others Durius (Duero) in Spain, Duranius (Dordogne) in Southern France and two Duria (Dora) in Upper Italy. As the phomeme sequence -ur- in old names is rare in general, it puzzled me that the uppermost Po area beside the two Duria also has several rivers named Stura, also Nure, Curona and even more. They are joined, with stem vowel -i- or -a-, by Taro, Ira and Stirone. The -ir of these last ones is even much more rare than -ur (and -ar). A particularly impressive group of such names I found more easterly around the Lake Como. This was called lacus Larius in antiquity, and into it a Maria (Mera) streams, and into that a Liro. Shortly west of Como however is a river Lura. Thus Lur-, Lar- and Lir- close to each other. This seemed to permit me to consider also the mish-mash of the previously mentioned Stura and Stirone to be relevant.

But also Dur- is attested near Lake Como. On its western side is a mountain named Monte Duria. Thus not only lakes and rivers were given names of the observed type. This is true btw also in Krahe's system, so that his opinion, that most of his name stems were 'water words', alone for that reason is untenable. Landscapes too and settlement and camp sites had of course names early on which could be formed by the same means, and also of those a lot may have persisted through the millenia.

This sketchily described state of affairs in the higher, more mountaneous parts of the Po area stands, as far as my knowledge goes, in contrast to a complete absence of river names of the described morphology in the lower, more open parts. In Southern France a similar contrast seems to exist between the Rhone depression and the adjacent mountaneous country to the west. A third landscape rich in -ur- and -ar- names is the mountaneous country of northern Swizerland, in particular the area of the river Aare. There one finds a Dura (die Thur, -> Rhine) and then Aare (-> Rhine) and Suhre (-> Aare), as well as der Jura (mountain massif), further a group of related formations with a consonantal suffixation or derivation. There is a river Orbe (-> Neuenburger See) and a river Surb (-> Aare), die Urnäsch (-> Sitten -> Thur), die Uerke (-> Suhre) and the town Murten (on the Murtensee), then die Sorne, (-> die Birs -> Rhine), die Sarine (Fr. name of the Saane) and the town Sarnen (with Sarner See) plus Uri (with Urnersee) and finally 5 times the river name Murg, which is reckoned to be Celtic but can't be so everywhere. Most of these derivations are geographically so clearly associated with the basic forms Dur-, Sur-, Ur- etc. that we may count them as belonging to this group. Also the upper Po country has its share of such fornmations, among others with the river name Urbis (Orbe), which should be identical with Orbe in Switzerland as well as an Orb in southern France and one in Hesse plus the town name Urb in the Rhineland. Other than that, Switzerland seems only to have Dur- (Thur-) in common with the Po area. However, most of its formations are also attested in other related areas.

The situation of the -ur-/-ar- names in Switzerland is related to the one in Upper Italy. Also here there are adjacent more open areas where such names are rare. O. Springer, Die Flußnamen Württembergs und Badens (1930), contains besides Murg only one Murr and one Morre, further perhaps Würm (from Wirm). On the other side the names we discussed seem to stretch only a short distance into the high mountains. If we add the Italian (and the Southern French) group there emerges an impression of a relict or refuge area in mountain areas which had become suitable for settlement relatively early.

I found -ur-, -ar- and -ir- names in all the European countries in which I searched, mostly, it seems, scattered thinly, but here and there concentrated as in the areas mentioned. This is in particular the case in a large area in Northern France, Belgium, Luxemburg, the Netherlands and North West Germany on both sides of the Rhine, with a center of gravity around the mountain and forest landscapes of the Ardennes and the Eifel (the "Fünfländerraum", five country area). It contains hundreds of names of the types we've mentioned, in some places conspicuously concentrated. Thus the Moselle above Trèves is joined, shortly interspersed, by Saar, Su:r (Sauer) and Sir (Syr), and this Su:r is joined by Ur (Our, with the town of Urb), and this again joined by die Irsen and Iren (Ihrenbach). We have in these areas, mostly several times or even often, the stems Ur-, Ar- and Ir-, Sur-, Sar- and Sir-, Dur-, Fur-, Mur-, and Rur- plus further, mostly in suffixed forms of the types mentioned above. It seems as if there has existed a widespread tendency in the west to extend the short stems of the basic layer, and above all in four kinds of ways, with -s-, -k- and -n- suffixes as well as (probably late only) with -apa. The former happened in particular around the Rhine. There we find names with the stem forms Urs-, Burs-, Durs-, Murs-, Nurs-, Surs- and Wurs-, further Ars-, Bars-, Fars-, Kars-, Mars- and Nars- plus Irs-. Here I leave outside consideration the fact that a perhps large part of these stems could be based on a form of type of *Duris-.

The group with the -k- suffix is distributed further and also more important. It is strongest in the Netherlands, but extends over almost the entire area of the names mentioned and is probably its most impressive characteristic. The Fünfländerraum can be circumscribed in general with formations of the stem Urk-: from river Ourcq (-> Marne) in the west to river Orke (->die Eder -> Weser) in the east and from river Ource (-> upper Seine) in the south to the island Urk (in the Zuidersee) in the north. The name stems which form this group are Urk-, Burk-, Kurk-, Lurk- and Murk-, Ark-, Bark-, Kark-, Mark-, Sark- and Wark-. Of the round dozen old island names on the southern North Sea coasts three, Borkum, Marken and Urk, belong to this group of formations, while three more, Marne, Voorne and the no longer existant Voorn, are formed with an -n- suffix. This type of formations is distributed widely, but is not so prominent as the type with the -k-.

The suffixed forms mentioned greatly contribute, especially by their abundance, to giving the onomastic material of the Fünfländerraum its particular appearance, but are not limited to it. Also the fourth and probably youngest form of derivation that I mentioned, the one in -apa - Or-pe/Ur-f, Dor-pe, Sor-pe and Lor-fe, Ar-pe, Mar-pe and Sar-pe - has, although it is almost completely limited to the mountaneous area to the right of the Rhine, parallels in the far distance (Lith. Dumpis, Nurupis, Surupis, Urkupis usw.). Almost completely limited to our area are till now, apart from isolate formations, only the stem forms Fur- and Rur-. The first belongs to coastal areas, the second non-coastal ones.

It is, as far as I can see, on most sides hardly possible to indicate the borders of the large area under discussion with any clarity or certainty. But towards the Northwest a clear border can be discerned. It may be characterized with six of the otherwise very rare Burk- names: from the island Borkum at the mouth of the river Ems via Borken on the lower Ems but also west of Münster (in Westph.), the Borken-berge at Haltern (on the Lippe) and Bork north of Dortmund to Borken SSW of Kassel. North and east of this approximate line formations of the type we followed become extremely rare. I shall came back to this "Borken border". But a thin chain of such names do stretch along the North Sea coast to the Jutland peninsula. There is a river Stuhr at Bremen and river Sturia (Stör) -> river Elbe in Holstein, river Egi-dora (Eider) with river Sorge (older Sorke) and several Marne or Marren. Also the island name Föhr presumably belongs here. This chain leads to a remarkable stronger group of such formations in the south, west and north of Jutland, the most of which have close relatives or perfect cognates in the more southerly occurrence areas of these name types, but also in Norway and also Sweden, whereas the Danish Islands are almost empty. On this point compare the map drawing below with its name list.

With that I face the question of the Scandinavian part of the mentioned name groups. But first some remarks on methodology. I'm aware that I will have categorized many names wrongly, and that many, perhaps the majority, also could be explained satisfactorily otherwise - and especially in the Nordic countries usually are -. This is inevitable in such old name layers, and that goes for Krahe's name sequences too. I am however convinced that I may assume that names or name stems which are distributed over various different (old) language territories in similar groups and applications and which lack specific markers of later or local origin, in general will belong together even if they in a few of the relevant countries can be explained from the (present) language of those. One may interpret our river name Su:r bei as "the sour one" easily, but not the island names Norw. Surn-øy and Greek Syrie:, Sy:ros and Syrnos. Apart from that, with the name types I presented it is the peculiar layering and grouping which would remain an enigma with that type of explanation. Furthermore, such a procedure as yet usually pays too little attention to the morphology of the names. In spite of that it is obvious, looking at the above example, that some Germanic rivers might have got the Su:r name the other way. I accept this willingly in the knowlegde that the peculiar overall picture which we see before us, could never and in no way have been there if most of it had resulted from rather random coincidences and similarities of very diverse origins. The same is true of one other serious source of errors, that we of no prehistoric name are certain that we know its oldest form and of many may be certain that we do n o t know it. Even if a third or even half of the names I've collected and approved should have to be excluded, there would remain enough of them to justify my conclusions. Also the incompleteness of the material will not disturb the overall picture by much.

One of the most difficut questions is that of the phoneme correspondences. This is the case as well in general — to what extent may we eg. treat Murg- and Murk- (plus Morg- and Mork-) or Dur- and Tur- as one? — as especially in particular with respect to the Germanic Sound Shift. That this, compared to more southerly parallels, in the names mentioned in Switzerland and also the north west area apparently applied only sporadically should not disturb us, since these areas only very late became Germanic geworden. But in the geatest part of the nordic countries such an explanation fails. That most of those names of that area which have seemimgly unshifted consonants only should have arrived there after the soundshift had taken place is, apart from a few cases, completely improbable. Or did this language change leave them untouched? This is actually possible. However, thew explanation for that would take up so much space, that I here abstain from it, at least, at least for the time being, and treat the non-application of our sound shift in most of the Nordic names, which I deal with here, as a fact which we can't very well deny.

I turn now to the situation in Norway, where the most important part of the onomastic material is collected and ordered in a particularly good and transparent form. Here it's the names of the islands no less than those of the rivers which contain parallels to the examined formations in the more southerly countries. This is so because of the nature of the country and the particularly high age of settlement along its coasts. From O. Rygh's great onomastic works come hundreds if formations, which have either exact cognates or at least relatives in those -ur- and -ar- names, which turned out to be so typical for the southern areas we outlined. There in the South there are not many characteristic stems and types of suffixed forms, of which Norway doesn't have its share (or seems to have). It is not possible here to list many of the parallels, so I'll restrict myself to a few cases in which Jutland forms the missing bridge

First the stems already mentioned Urk- and Burk-. Jutland has the latter in Bork (on the Ringkøbing Fjord), the former probably in Vork (W of Vejle) and Orke-krog (on the Solkær Å, SE of Kolding). In Norway I'll mention with this stem only Ork, the old name of the river in Orkdal, and the Yrkefjord (E Haugesund). Sweden contributes here at least wth the lake name Örken (Småland). Burk- is found i.a. in the Norwegian river name Borkn (Bortna, Gauldal) and the lake name Borken in Östergötland. Then Urt-. France has the river name Orthe (-> Sarthe) and Belgium Ourthe (-> Maas), the Netherlands the old town names Orten and Lower Saxony Ohrte, Denmark Orten (NW of Varde) and Orte (western Fyn) and further an Orte-krog (on the Uge Å, SW of Åbenrå) and Yrt (in Sundeved), Norway i.a. Orten and Urter as island names and twice Ørteren as names of a lake, and likewise Sweden has Örten. To the stem Murs- belongs a Mursa plus Mursila and Mursella in the old Pannonis, on the Lower Rhine the old hydronym Mörs, in Jutland the island name Mors and in Norway Mors (Moss on the Oslofjord), even this the old name of a river.

The Netherlands Fur- (and Fu:r) briefly mentioned above have many relatives in the Nordic countries, on the Jutland peninsula in the island name Fur (and probably also in Föhr), in Norway in the frequent river names For/Fora, Fura and Forn, in Sweden at least in the lake name Furen and the well-known river Fyres-å. The island names Voorn and Voorne in the West seem to have an exact cognate in the west Norwegian Fonn. Also Dur-, the name stem, from which I started out with, is not quite infrequent in the Norwegian river names nicht ganz selten (s. NE 33 and 35 f.). In Jutland it's different. There I found this stem, aprt from Egidora , only in the derivations Dørken (at Give) and Dorf (Vendsyssel). Dorf may be compared to the old river name Durbia (France) and Durbis (Britain) as well as probably Dyrfa in Western Norway (NG 11, 599). Dørken seems only in the West to have a few relatives (in the Netherlands de Durk, in Western England a river Dork).

One of the most intersting among the name stems I have traced is the Sir-, which is coupled with Sur- and Sar- in the Moselle area. Of the few old name stems which contain -ir-, this is the most widely distributed. Norway has it in the island name Si:ri (Utsira), outside of Haugesund) and the rivers Sira (Agder) and Sire-å (Hallingdal), further two *Si:r-vin, Sweden apparently in the island name Sirk-ön (in the Åsnen, Småland), Denmark in the place name Sir or Sire (N of Holstebro) and in the island name Sejer-ø (old Syr-ø). There follows then in the Netherlands a deserted Sier on the island of Ameland, itself probably an old island name, the name of the brook Sire-beke (Zierbeek) at Brussels, the already mentioned Sir (Syr) in Luxemburg and a bit more to the south (in Lorraine) the place name Sierck, which probably also was the name of a brook, in Switzerland (Thurgau) a Sirn-ach and futher in Southern Italy the island Sirenes (outside of Campania) and der river Siris (Lucania) as well as the Sirnides islands outside of Crete. And there are still many more names. Sir- (and Sīr-) is also at least 6 times, from Norway and Sweden to Italy and Greece used in connection with islands and even more often and similarly layered with water courses. This ensures that the names are related and forbids us, as was done in Norway, to explain the stem from the historical individual languages focussing on just one of its uses. With the closely related Sur- the situation is most likely similar to that of Sir- (see above).

Their considerable share of the stem Fur- as well as of the -k- and -s- derivations - in Jutland alone in Bork, Vork, Orke-krog and Dørken, Bars-ø, Mors, Nors and Ørs - testifies to the close relationship between the Nordic -ur-/-ar- names with those of Western Germany and its neighboring countries, whereas other things, as shown, suggest much wider connections. The examples I have adduced will at the same time suffice to make it clear that beside Krahe's system, also distributed over large parts of Europe, have existed and exist many old names, above all river names, which in the phonological form of their stems as well as in their methods of derivation are related to each other, and also by their distribution on landscapes show that they belong together, so that we with the same right (or not) may organize them as a system as Krahe's name sequences. On both sides much work remains to be done, much more than hitherto performed, before we get close to the insight attainable. Especially in the Nordic countries most such work still lies ahead of us. As long as not all (seemingly) pertinent names have been collected and ordered - and also filed - we may not hope to solve the most important questions with any reliability. One thing however stands clear in my conviction, that there exist two systems in the oldest onomastic material in Europe, which are related in many respects, but fundamentally different in others, and that also the Nordic countries have their share of them, and at least of the latter even a large share.

I will now try to say something about the relationship of these two name systems to each other. Both must have been fashion phenomena which were active over long time and might have have had long afterlife. They differ from our later name fashions especially by the fact that these latter are consistently centered only on the identity of the suffixes or second name elements, and then perhaps on agreement in semantic category of the stems or first elements (eg. personal names), the former ones on the other hand evidently above all on the phonetic similarity of the name stems and a limited selection of means of derivation. This may have been connected with understood a consistent lack of insistence that the name should mean something. But names should designate, not mean something, so that also such systems could fulfill its purpose. This to us strange way of naming would probably have matched the development stage of its time.

In the methods the two systems the two systems used the difference is great, but not absolut. The most striking one is probably that both limit themselves almost exclusively to the three vowels -a-, -i- and -u-, whereas -e- and -o-, the most important basic vowels of Inde-European, are almost absent. In this respect in Krahes sequences -a- is by far the most frequent and -u- the most infrequent, in my groups however -u- is in front and -i- in third place. Krahe in his groups has non with the phoneme sequence -ur-, which forms the core of the other system. On the other hand the phoneme sequence -ar- enjoys great importance on both sides. Krahe's sequences contain - in praxi, not in theory - the stems ar- (with arg-), kar-, mar-, nar-, sar-, tar- and war-, which I for the most part have claimed for my system. this is obviously justified with ar- and sar-, and probably also with the more infrequent nar- and tar-. For they are almost completely limited to those lands to which the -ur- forms are indigenous, and have partly a clear relation to them. The confusion of Sar and Sur on the Moselle, which I have mentioned several times, has many parallels. In contrast kar- and in particular war- seems for the overwhelmingly largest part to belong on Krahe's side. There are beside them only a very few kur- and wur- with typical characteristics of the second system and their distribution is not bounded as eg. the way eg. ar- and sar- are. The last -ar-stem remaining, mar-, was, it seems, from early on at home in both systems. Still, much remains to be adjusted here, and details could be mentioned which seem to testify to a mix of the two systems. It wouldn't surprise me, if a more meticulous investigation focusing on the distribution of the particular constructive elements on landscapes would suggest that the systems initially were more clearly separated and most of its common pool of stems and formation elements were the late product of an interchange.

For understanding how the two compared systems are connected their distribution area is of course of great relevance. Krahe has circumscribed that of his names at least roughly. It covers most parts of Europe, and almost only marginal areas in the South and the East remain outside, so that he with full right called his system European. Still, also here there things to amend and probably more to complement. Krahe paid in particular very little attention to the great fluctuation in the denity of his names. It's my impression, that these usually decrease appreciably, where the other system is at its strongest. This then is also distributed very widely, but much more irregularly and, as already stressed, with large lacunae. His names are situated most densely, as far as I can see, in most of the extreme margins of Europe which are reached only weakly or hardly at all by Krahe's names. They also reach into Asia Minor and probably even further. On their distribution in the central parts of Europe the most important things I can say are already mentioned. In the East at least Lithuania has a large share, and similarly according to old sources the whole periphery of the Black Sea. Apart from that I know only very little of the situation in Eastern Europe.

The thesis that this fragmentary overall picture of the distribution of the -ur-/-ar- names has led me to, is the following: These names were once in use in almost all parts of Europe and also beyond its borders. But then Krahe's system penetrated into its center, replaced there the majority of the formations of the other system, stopped its further development and expansion and pushed it far to the west and south and especially into the more inaccessible mountain landscapes. However, it took hold later in many or most of the countries and landscapes in which the names of the second system till then had been able to stand their ground and establish themselves. Most of what supports this theory has been said or hinted at above. But some amendments are necessary. The area where I had been struck by a confusion of landscapes rich and poor in -ur- names - in upper Italy, Southern France, Switzerland, North West Germany and its neighbor countries and also Denmark -, borders on those poor in such names in the East and North East. From this direction the destruction must have come, at least in the western half of Europe. That my name system was once distributed more widely is indicated by scattered occurrences outside of the circumscribed areas, mostly in more remote areas. Also the Danish Islands have their share. On those I know three pertinent formationes, all lying at their edges, the already mentioned place Orte (on Fyn) and Sejer-ø and the the lake name Fure-sø near Copenhagen, to which J. Kousgård Sørensen drew my attention. Most occurrence of this kind, it seems to me, relics of an older distribution area.

The state of affairs in Denmark are remarkable also for another reason. If we ignore the geographically isolated Sejer-ø and Fure-sø , then the remaining 22 formations of the -ur-/-ar-/-ir-Systems form a concentrated distribution in the western half of the country, the boundary of which at first looks very random, since it divides Jutland as well as Fyn in two parts, which however match the western boundary the distribution area of the Late Neolithic megalith graves (jættestuer) so conspicuously well, that coincidence is quite improbable (see adjacent map). In this period, according to J. Brøndsted the first centuries of the second millenia before Christ, or at least not long after, the eastern boundary of the examined names in Denmark thus seems to have estrablished or entrenched itself. There are also other very old name types which are limitted to the thus delineated western half of the country (thus the island names on -s and the name Dover).

illstration: "Das zweite Alteuropa, The second Old Europe.jpg"

My next question is whether it there are reasons to connect one or both of the old name systems with a particular linguistic or ethnic group. There are such pointers, and they are so strong that I may try to give an answer. They point to the -ur- names having been given by non-Indo-European speaking peoples. Many of those countries which have a large share of them - thus the main part of the Pyrenaic peninsula, Liguria, Greece and the most part of Asia Minor - only late or never became Indo-European, and the Greek names of these types are at least for a large part said to be have been given by the pre-Greek population, and they were hardly Indo-European. The same goes for the name stem Dur-, which besides that betrays its foreign origin with the attested non-Indo-European nominative form Dur (in Ireland). Also the river Nar (-> Tiber) had such a name form. On top of all this, the North German "Borken border" matches the north east border of a small group of other names ot name elements which certainly or highly likely are not Indo-European (ref. 4, 270-276), and in the largest part of the Fünfländerraum old Indo-European names are very rare, but not so Celtic, Germanic and Roman ones, so that we must assume a very late Indo-Europeanization of these areas (cf. ref. 1, 324 and here ref. 3, 352 f.).

On the other hand Krahe was completely convinced that his river names were Indo-European. The strongest support of that was and is its distribution as sketched above. On top of that, most of its elements were explainable, or seemed to be, from Indo-European languages. It's another question whether they were all of that origin. Against that I have raised serious objections. (ref. 1, 326 ff.). But in spite of that we may presumably assume, that in the main it was Indo-European tribes or peoples who distributed Krahes river name system. I'll add a perhaps weak argument. The Indo-Europeans, according to the evidence of their language, were initially inland inhabitants without knowledge of the sea. This matches with the fact that the other name system seems to have persisted particularly well around the coasts of the North Sea, of the Atlantic and the Mediterranean, as if the new names and thus the new people had stopped up before them. Therefore it is probably so that the invasion of Krahe's name system and the retreat of the second reflect the Indo-Europeanization of the core areas of Europe and that in large part thus migrations stand behind this.

It should be possible, then, to place the processes we mentioned in time. Krahe placed the genesis and dominance of his system in the 2nd pre-Christian millenium. This may, at least what regards the onset, also be supported be strong arguments which he did not take into account (s. ref. 2, 219 ff. and ref. 1, 333 f.), and with this modification it seems to be true. Many elements of his formations have lived on a long time, and many of the names are certainly younger. Other than that, we will have to rely on the assistance of prehistoric research. The case of the old name border running through Denmark shows that we might expect much from having it cooperate. For the "Borken border" however it doesn't provide us with any explanation.

Even if the many hundred names, of which I have presented samples here, didn't form a coherent group, the would suffice to reduce ad absurdum Krahe's claim that his names are the basic and core layer of hydronyms in most of the participating countries. After all, it's obvious and admitted by himself that of the great rivers of Europe hardly a single one carries a name belonging to his hydronymy. The second system appears in this respect to do somewhat better, but not by much. I only need here to mention Po, Rhine and Moselle, the tributaries of which are so particularly rich in -ur- and -ar- names. Therefore I don't intend to place my system next to or instead of Krahe's in the beginnings of the European river naming. There are beside them further several hundred ancient appearing hydronyms which are formed differently and of which hardly more han a small part could be assigned to a third or even higher system, and which are of a similar age or for a large part could be even considerably older. This is in general not only the case for hydronyms. Also in this remaining part the Nordic countries have their great share.