The enormous impact of Roman civilization on the populations of Europe is well-documented. Roman products were carried also to remote areas, such as Scandinavia and India. In most of Europe, the STYLE of manufactured products was powerfully influenced by the Roman patterns. The PROVINCIAL ROMAN STYLE, developed in the western provinces, spread soon over large areas extra provinciam, particularly among the Germanic peoples. In the Gothic style, which originated among the Goths north of the Black Sea and spread westwards, Germanic patterns were mixed with classical (particularly Hellenistic) and Oriental forms.

Influenced by the Roman Empire, many European peoples introduced the use of coins. After the period of Emperor Trajan (98B117 AD), silver coins were spread over the entire territory inhabited by Germanic peoples and towards the end of the 2nd century, such coins reached Denmark, Gotland, and the region of the lake Mälaren in Sweden. They were used in commerce (not only as jewels), and were found at thousands of places beyond the Empire. There are also finds showing that the Germanic peoples borrowed the Roman custom of putting gold coins in the mouth of the dead.

In the 4th century, a Christian congregation existed among the Goths living in the territory of present day Muntenia. For 7 years, their bishop was Ulfila, who translated the Bible to the Gothic language (cf. the record of Auxentius Durostorensis, above, pp. 210B211).

An interesting group of imported objects is that of small FIGURINES OF ROMAN GODS AND GODESSES, of which about 15 are known from Denmark and some also from Sweden. These were widespread in the Germanic world of those days. In Scandinavia, they were never found in tombs or in settlements but constituted single finds in the woods and fields. They were mass-produced in the Roman Empire, for the needs of the general populace. Their use may be compared to that of pictures of saints in the Catholic world today. These figurines were copied at least in one place (Öland), and it is not excluded that they were used also by the population of Scandinavia as cult objects.

Also LATIN INSCRIPTIONS were found in Scandinavia. Two swords with inscriptions are known, one from southeastern Sweden and one from Gotland, as well as two bronze vessels, all dated to the 2nd century AD. One is the so called Apollo Grannus vase, found north of the lake Mälaren in Sweden, with the inscription AAPOLLONI GRANNO DONUM AMMILIVS CONSTANS PRAEF TEMPLI IPSIVS VSLLM@. (´To Apollo Grannus was this gift given by his chief of temple Ammilius Constans.´) Another bronze vessel with a similar inscription was found in Norway, east of Mjösen. These objects were most probably taken from temples in Gallia or Raetia by plundering soldiers.

Objects of the kind discussed in this section may easily be transported to far away areas. They may give information regarding technology: the people of Öland were capable of producing figurines of Roman gods; economics B indicating commerce with the Empire, or suggesting that soldiers from the place in question may have served in the Empire. Regarding questions of the social circumstances in the area in question, not to mention the language or ethnicity of the people living there, they have no value.

Archaeological remains of an assumed ADaco-Roman@ population were described from a number of sites in Oltenia and Transylvania. Protase presented more than thirty such sites (cf. above, pp. 159B173). The assumption of a Roman population is based on the Roman provincial style shown by the remains. However, at least at eight of these sites, (cf. table 9, p. 170), the existence of non-Roman peoples is shown. Among these are Chilia and St|neÕti in Oltenia, and Cip|u, Comol|u, MediaÕ, Bezid, and Reci, in eastern Transylvania, which date from the middle or the 2nd half of the 3rd century, when the Romans left the limes trans-Alutanus and eastern Transylvania. To the abandoned areas, free Dacians, Carps, Iazyges, and Goths migrated. Instead of the continuation of the population from the Roman period, we see here the influx of non-Roman elements already in the mid-third century. The classification of these sites as ADaco-Roman@ is simply wrong B en example of the ambiguity of this notion, cf. above, pp. 199B200.

Also at other sites, clearly non-Roman elements appear. At Archiud (Erked), in a settlement inhabited from the mid-third to the 4th century, the earthenware of Roman provincial tradition shows Aincreasing barbarization@, and there are vestiges of the free Dacians. At Mugeni (Bögöz), elements of the Cerneachov culture appear towards the end of the 3rd century. In Sarmizegetusa, the once flourishing capital of Roman Dacia, changes made on the old Roman buildings, according to a Abarbarized technique@, may have been achieved by any population. Two of the sites (Cioroiul Nou and VerbiÛa) are situated between the Danube and the Furrow of Novac in Oltenia. This area was in the 4th century re-conquered by the Empire (cf. above, p. 152 and map No 8), the existence there of Romans in that period would therefore not be impossible. Regarding the problem of the persistence of a Roman population in former Dacia Traiana, this is, however, of little relevance.

Horedt described 50 settlements in use in post-Roman Dacia Traiana. He distinguished two main groups: (a) a western group of settlements, in the valleys of the MureÕ, the lower course of the Târnava, and the SomeÕ, there are a kind of fibulae (Zwiebelknopffibeln) and combs with two rows, not found in the eastern group. The pottery is more like the Roman pottery. (b) The eastern group, found in the valleys of the middle course of the MureÕ and the upper courses of the rivers Târnava, as well as in present day Covasna and BraÕov counties, called the Sfântu Gheorghe culture, is clearly non-Roman, with influences from the Sântana de MureÕ culture, and also from the east, the Carps.

The assumption of a Daco-Roman population in 13 cemeteries (cf. above, p. 164B173) is also based mainly on material of Roman style. In several sites, as also in the case of the settlements, clearly non-Roman material remains were found. Thus, at S|r|Ûeni (Sóvárad), vessels similar to Carpic vessels and the habit of putting several vessels in a tomb is characteristic of the Cerneachov culture. Cemetery No. 1 at Bratei contains material remains characteristic of the free Dacians, pieces of earthenware made by hand, e.g., the Dacian censer (c|Ûuia), fragments of big vessels of supply decorated according to the classic period of the Dacian pottery. Also the large amounts of animal bones, not found in Roman and Illyrian cemeteries, is characteristic of the tombs of free Dacians (of the type PorolissumBSalca). Several details of the rituals, such as placing the inventary over the cavity, assembling of fragments of earthenware together with stones, animal bones, and other objects, are old Dacian customs. The ritual of burning of the cavities used at Bratei was also found in the cemetery excavated at OcniÛa, which dates from the period before the occupation of Dacia by the Romans. The funeral ritual of burned cavities is not Dacian and does not appear in significant numbers before the Roman conquest. It is found in Pannonia, Moesia, and Illyricum. Garasanin (quoted by Bârzu Cemet. 1973, p. 92), assumed that it is of Illyrian origin; it was, in any case, widedspread in Illyria. Big vessels of supply are abundant, and they are not of the Roman type. This cemetery does not continue one from the Roman period but was in use in the second half of the 4th (and, possibly, in the first quarter of the 5th) century. All these facts contradict the assumption of a Roman population having been buried here. Also the surroundings are other than Roman: only about 5 miles from Bratei, at MediaÕ (Hung. Medgyes), a Carpic tomb of cremation which dates from the end of the 3rd century was discovered (cf. above, Table 9, p. 170). Products imported from the Empire suggest that the people who used the cemetery No. 1 at Bratei had commercial contacts with Pannonia. On the other hand, nothing among the finds suggests contact with the Roman towns along the lower Danube.

About the two tombs discovered at S|r|Ûeni, Protase wrote in 1966 that they may have belonged to Daco-Romans (cf. above, p. 172). In a more detailed analysis, however, he stated that Aif the observations at S|r|Ûeni are correct@, the cremation was made on the place of the burial, a ritual widespread in Dacia Romana and not known from the extra-Carpathian regions. But the vessels found in these tombs are to a certain extent similar to those found in Carpic cemeteries in Moldavia and there are 3B4 of them in each tomb, the habit of putting several vessels in a tomb is characteristic of the CerneachovBSântana de MureÕ culture.

At Soporul de Câmpie, many of the material remains show the provincial Roman style: their fibulae are of this type, in contrast to those found among the material remains of the free Carps, which are of the Abarbarian@ style. The same may be said about a considerable part of their pottery. A Sarmatian influence, characteristic of the remains of the Carps in Moldavia, was not ascertained at Soporu de Câmpie. In 5 out of 189 tombs, i.e., in 2.6% of all tombs, coins were found and it is assumed that these were used according to the Greco-Roman custom of putting an Aobulus of Charon@ in the mouth of the deceased.

On the other hand, there is ample evidence to indicate that the people at Soporu de Câmpie preserved their old traditions, notwithstanding the fact that they were living in a Roman province. The total number of tombs in this cemetery is 193. Out of these, 189 date from the 2nd and the 3rd centuries and are considered to belong to ADaco-Romans@. (Four tombs from the 5th century are not considered Daco-Roman.) A total of 167 tombs contain earthenware and in 62 of these, this is of the Dacian style. The differences as compared to the earthenware of the free Dacians are not essential:

Protase concludes that the people who left these material remains Awere able to organically assimilate only minor elements of the rich repertoire of the Roman provincial tradition.@

This is true of entire Dacia Traiana: THERE ARE NO INDICATIONS OF A PROCESS OF GRADUAL ADOPTION BY AN INDIGENOUS, NON-ROMAN POPULATION OF ELEMENTS OF ROMAN CULTURE AND CIVILISATION beyond the use of objects and assimilation of Roman style, essentially in the same way as was the case in almost all of contemporary Europe.

Those ten objects presented in 1966 by Protase as indicating the presence of Christians in Transylvania in the 4th century were described above (p. 173B175). Since then, a number of similar discoveries were made, for example crosses and casting moulds for crosses, objects with different Christian symbols, Christian monograms on pottery pieces, etc.

If the ancestors of the Rumanians had adopted Christianity in Transylvania in the 4th century, then they must have moved to Bulgaria in the 10th century. It is namely scarcely possible that they could have been organized in the Bulgarian Orthodox Church, taking over Slavic as the language of the religious services, (and borrowing more than 70 religious terms from Bulgarian), during that century the Bulgarian state ruled over Muntenia and southern Transylvania after 830 AD (cf. above, p.22). B The Rumanian Orthodox Church belonged, until the early 18th century, to the ecclesiastical jurisdiction of the archbishopric of Ochrida.

North of the lower Danube, the low number of coins from the period between 271 and 305 AD may be explained by the bad economic situation of the Roman Empire, and the decrease and gradual disappearance of the coins in the 5th cent- ury was certainly caused by the severe crisis and final collapse of the Empire. These factors are recognized also by Protase and Preda, who argued that the coins indicated a Daco-Roman population (cf. above, p. 176B178). However, their reasoning does not take into account other important factors. Roman coins were not only found near former Roman towns or camps and they are by no means restricted to the territory of the former province. Those of a lower value (the bronze coins) were used as money by many European populations in the period in question. Roman coins were discovered as far as in Ukraine (in the territory of the Cerneachov culture) and on the Crimean peninsula. In the most eastern areas, they are rare, but their number increases considerably west of the Dniester. In the territory of present day Rumania, the free Dacians, who migrated to the former province after 275 AD along the MureÕ and the superior valley of the Olt, left Roman coins in their settlements and in the Banat, the Sarmatian Iazyges Aused extensively, for a long time, Roman coins@.

In the case of hoards of coins, drawing conclusions is made difficult also by the uncertainty about basic circumstances connected with the finds. Many of them were found a long time ago and lack proper descriptions of essential data such as their integrity and unity. A number of them might have been imported much later from the territory of the former Roman Empire (cf. the reservations made by Protase, above, p. 177). According to Protase, some of the hoards accumulated entirely in the 4th century may belong to the Aautochthonous population@. This is not impossible, but this population turns out to have been the free Dacians, who were not Latinized.

The two inscriptions (Qvartine vivas and the ex voto with the inscription: Ego Zenovivs votvm posvi, cf. above, pp. 178B180) do not tell us anything about the ethnic character or the language of the population of Transylvania in the 4th century. Contrary to what Condurachi & Daicoviciu (cf. above, p. 179B180) assert, the ex voto was not produced in Dacia. Both inscriptions were imported from the Empire, but it is not even certain if this occurred in the 4th century, later imports cannot be excluded. The finds were isolated, without any distinct archaeological site. The ex voto was probably produced in northern Italy; similar objects were found in Aquilea, Bonyhád (Hungary), Ljubljana (Slovenia), etc. Most probably, these pieces reached Transylvania through trade or by plund-ering soldiers (cf. Latin inscriptions found in other parts of Europe, above, p. 222).

A serious objection against the concept on which Rumanian archaeologists base their research is that they restrict investigation to the territory of present day Rumania. In view of the fact that it is generally admitted, at least by linguists, that the Rumanian language and people also developed south of the lower Danube (cf. above, p. 140B142), a similar picture of archaeological finds, identical or at least very similar cultures, should exist on both sides of the lower Danube or even at a considerable distance from it towards the south (that is to say, if the material excavated in Rumania really showed the ancestors of the Rumanians.)

Protase presented 22 settlements as indicating the continued existence of ADaco-Romans@ in post-Roman Dacia Traiana. Out of these, six show the material remains of FREE DACIANS OR CARPS who migrated to the former province towards the end of the 3rd century: Bezid, Cip|u, Comol|u, MediaÕ, Reci (in Transylvania) and Chilia in the region of ArgeÕ. At Sic, a settlement of Dacians during the Roman domination continued possibly until the end of the 3rd century. Seven of these settlements were FOUNDED AFTER 275 AD (the abandonment of the province by the Romans): Archiud, Bratei, Cluj-M|n|Õtur, Iernut, Mugeni, NoÕlac, and Soporu de Câmpie. Two (Cioroiul Nou and VerbiÛa) are situated in Oltenia south of the Furrow of Novac, in an area dominated for some decades during the 4th century by the Roman Empire (cf. above, p. 224).

Thus, six settlements remain north of the Furrow of Novac which were a) inhabited during the Roman domination and with more or less probability also after its end (in several cases only until the end of the 3rd century) and b) in which the material remains do not clearly indicate free Dacians or Carps. This is a very low figure, and even of these, two can scarcely be taken into consideration. At Porumbenii Mici, the existence of the settlement after 271 AD is questionable; at most, it may have existed a few decades at the end of the 3rd century. In VeÛel, the only sign assumed to indicate life in the 4th century is a silver fibula.

A survey of the data presented by Protase PCD 1966 and Horedt 1973 shows a similar discontinuity of the villae rusticae (rural farms) from the time of the province. Only one of these is assumed to have been inhabited after 275 AD (Iernut) and in another (Rah|u) is this possible.

Out of those 13 cemeteries or single tombs presented by Protase (cf. above, p. 164) as showing the material remains of Daco-Romans, two (Chilia and St|neÕti) show the material remains of free Dacians who migrated to the area after the Romans had left. The cemetery at Bratei dated to the 4th century was discussed above (pp. 165B169); remains suggesting the Sântana de MureÕ culture as well as of free Dacians predominate there. In MediaÕ, an urn of cremation from the end of the 3rd century and assumed to belong to the Carpic population was discovered. Of the remaining nine sites, 6 are single tombs and date from the end of the 3rd century or (in the case of Cluj) the mid-fifth. Only one cemetery, that found at Iernut, was in continued use after 275 AD; from the mid-third century to about 400 AD. At Moigrad, a cemetery was most probably used during the Roman rule, and its use after 275 is questionable.

Fig. 2. The time span during which the SETTLEMENTS attributed to a ADaco-Roman@ population in post-Roman Dacia Traiana may have been inhabited. Maximal limits of time, earliest start and latest abandonment, are given. (After D. Protase, Problema continuit|Ûii în Dacia în lumina arheologiei Õi numismaticii, 1966.)

Fig. 3. The time span during which the CEMETERIES attributed to a ADaco-Roman@ population may have been in use. (On the basis of D. Protase, Problema continuit|Ûii în Dacia în lumina arheologiei Õi numismaticii, 1966.)

III. THE 5thB7th CENTURIES

Those settlements in which Daco-Romans were assumed (cf. above, pp. 159B164), were abandoned partly already in the first quarter of the 4th century and latest towards its end, obviously attacked by the Huns. Thus, the population of the former province was, less than a century after the abandonment by the Roman Empire, once again exposed to radical changes in its dwelling places.

Nevertheless, many different explanations have been put forward in the course of time. They often contradict each other. Even the period in which this flight is supposed to have taken place is given vaguely and differently by different authors. Condurachi & Daicoviciu (cf. above, p. 234) state that the attacks of the Huns and of the Avars caused the Asteady disappearance of the Daco-Roman farming settlements@, which implies that the process started in the second half of the 4th century. According to IR 1960 (p. 627), the Daco-Romans left their old places during the 5th and the 6th centuries. The explanation put forward in IR Compendiu 1969, p. 100, asserts that the Daco-Romans fled to Adistant valleys, among hills and mountains, in one word in places better protected from the ceaseless predatory invasions of the migrating troops.@

As shown above, IR Compendiu 1969 defended the theory of the flight to the mountains. More recently, archaeologists in Rumania deny it. In the 1974 edition of IR Compendiu, the passage about this is omitted, although it is not yet denied: writing about the Gepidae living together with ADaco-Romans@, it is asserted that Abeing different populations, basically inimical, they had few points of contact, which explains the absence of Old Germanic words in the Rumanian language@. Eugenia Zaharia disposed of the theory of the flight into the mountains entirely, stating that it belongs to the 19th century, and that the archaeological finds show

This is not a quotation, in fact, no scholar of the Rumanian language would even dream of asserting this. The Arumanians living in northern Greece cannot have been living there since their Romanization, but must have migrated there later, since that territory is situated south of the Jire…ek-line, the frontier between Latin and Greek in the Balkan peninsula. The question put by K. Jaberg is to be answered as follows: The Arumanians migrated, after the period of Common Rumanian, to the region northwest of Trikkala. As a Romance language, their language contained many Latin words, among others, also those mentioned above. The difference as compared to the rest of Rumanian today has nothing to do with the Arumanians. Its explanation is to be sought in the other dialects. Northern Rumanian, for example, has lost cîÕári and uses only strung|, it has replaced Latin faba with other words, etc. The explanation of the preserved Latin words in the region of the MunÛii Apuseni, in CriÕana, MaramureÕ, and the Banat, is similar. In fact, the impossibility of the argument regarding the ancient Latin core areas appears already in the above quotation from Giurescu & Giurescu. It contains the information that those Latin words appear not only in the region of the MunÛii Apuseni, Awhere the Roman settlements were most dense@ (PuÕcariu), but also in CriÕana and MaramureÕ, where such settlements were entirely absent, because those areas never belonged to the Roman Empire. Rosetti has pointed out the fact that other Latin words were forgotten or replaced by foreign loan- words in Transylvania but preserved in other areas of Rumania, along the lower Danube, in Moldavia or in Dobrogea:

To give an example of how these dialectal differences are to be interpreted, we consider the dialectal distribution of a number of words. The word for ´snow´ is nea (< Latin nivem) in western Transylvania, CriÕana, and the Banat; z|pad|, of Slavic origin, in Oltenia, Muntenia, southern Transylvania and Dobrogea; and om|t, (cf. Russian omät, Ukrainean om�t) in Moldavia and north-eastern Transylvania. For ´abdomen´, there is in the northern part of the country, pântece (pâncete) (< Latin pantex, -ticem), in the south-west, foale (< Latin follis, -em), and in the south, burt|. The word of Latin origin, p|curar (< Latin pecorarius) is used by speakers of Rumanian living in the intra-Carpathian territory, with the exception of southern Transylvania, where cioban (of Turkish origin but transferred to Rumanian via Bulgarian) is used. It obviously extends from Muntenia and Oltenia, where a homonymic clash occurred between p|curar ´shepherd´ and p|curar ´merchant of crude oil´ B in those territories, oil was exploited. In the case of the word for ´woman´, it is also possible to follow the changes. In this case, the original Latin word was preserved in Rumanian spoken today in the Banat, CriÕana, and the centre of Transylvania: muiere (< Latin mulier, -erem). In Muntenia, Moldavia, southern and northern Transylvania, it was replaced, starting with the 17th century, by femeie, a word also of Latin origin, but with the sense of ´family´ whose sense changed in the same century to that of ´woman´. Lastly, in F|g|raÕ, where many Rumanian peasants were ennobled by the Hungarian kings, boreas|, from boier ´nobleman´, with the suffix -eas|, replaced the word of Latin origin.

Nea exists also in Arumanian, Meglenorumanian, and Istrorumanian: neao, n„u|, and n„, n„vu, respectively. The inherited Latin word familia exists also in the south: Arumanian fumeal´e , Meglenitic f|mel„l´|, fum„l´|, and there with the original sense of ´family´. Also Latin mulier is found in the south, and was earlier general in the Rumanian language: Arum. and Meglenitic mul´ari, Istrorumanian mul´„re.

In principle, the explanation of other phenomena put forward as arguments for a Acore region@ in north-western Transylvania is similar to that of the words. Thus, the habit of putting a k between s and l B sclab instead of slab, etc., as well as the form renunculus (besides reniculus) existed in certain areas of Late Latin; also in part of the Balkan peninsula. Given the different forms in the Latin of the Balkan peninsula, it is only natural that they are found also today in the language of the Rumanians B as it is natural that the Rumanian grammar is predominantly Latin, or that there are Latin words in Rumanian.

The phenomena put forward in favour of the Kerngebiet theory are chosen arbitrarily from a vast number of facts of language, such as differences in pronunciation, the preservation or replacement of Latin lexical elements by loanwords in the course of time in the different areas, etc. Some examples of these factors were given above. Although this theory was initially put forward by linguists, present day Rumanian linguists do not accept it. As shown above, Rosetti refuted it definitely and Matilda Caragiu MarioÛeanu, in her monograph about the Rumanian dialects, does not even mention it. The same is the case with other publications on Rumanian dialectology as for example the studies by B. Cazacu, collected in one volume (1966). If it has been discussed here in some detail, this is because the theory of core areas seems to be one of the main arguments in favour of the theory of continuity accepted also by some western scholars (cf. above, pp. 195B198). E. Illyés, in Ethnic Continuity in the Carpatho-Danubian Area, 1992, pp. 272B290, after having given a more detailed review of this theory, also refuted it.

2. OTHER LINGUISTIC ARGUMENTS

The arguments mentioned in IR Compendiu 1969 (cf. above, p. 182), concerning the Christian terms of Latin origin B namely that in Dacia Romana, Christianity was the bearer of the Latin language B are not acceptable. These terms were inherited by Rumanian in the Balkan peninsula (Albanian concordances, etc, cf. above, pp. 111-115). The theory that the word biseric| proves that the ancestors of the Rumanians did not live in the Balkans in the 4th century was not mentioned in the 1974 edition of this treatise. As shown above, basilica has also been borrowed by Albanian.

The theory based on the difference between the treatment of Common Slavic consonant groups *tj, *dj , and a number of Common Slavic vowels (cf. above, pp. 182-183) cannot be accepted. The Bulgarian pronunciation reaches far to the southwest to Macedonia. Besides, to refer to the Aoldest and most numerous@ Slavic elements in Rumanian, as if the oldest elements would also be the most numerous, does not correspond to reality. (The oldest Slavic influence includes about 70 lexical elements, which existed in Common Rumanian. Most numerous Slavic loanwords in Northern Rumanian are those borrowed later, in the 11thB 12th centuries (cf. above, p.101).

There is significant difference between Rumanian archaeologists and historians on the one side and linguists, on the other, regarding this question. The affirmation in IR 1960 and of Giurescu & Giurescu, that the Slavs living in the territory of present day Rumania were assimilated to the Rumanians in a relatively short time after the 10th century (cf. above, p. 183) cannot be correct. The same applies to the result which Eugenia Zaharia reached on the basis of archaeological finds B the assimilation in the 8th century. This is clearly shown by the analysis of the sound pattern of the South Slavic elements in Rumanian, presented in detail also by Rosetti (ILR 1986, cf. above, pp. 98B101). If IR 1960 would be right, only about 70 South Slavic words were found in the Rumanian language, essentially the same in all dialects. The result reached by Eugenia Zaharia is an example of the danger of drawing conclusions regarding language from material remains. The disappearance of the Slavic contact already in the 8th century would not even permit this relatively weak Slavic impact. As shown above, p. 101, the South Slavic influence on Northern Rumanian increased after the Common Rumanian period, reaching its peak in the 11th and the 12th centuries.

Rumanian historians assert that the Rumanian language was exposed to Slavic (entirely or at least predominantly) north of the Danube. Rosetti presents all the facts but does not draw the logical conclusions. The passage dealing with this in Rosetti ILR 1986 (cf. above, p. 184) may be summarized as follows (with the present author´s comments in brackets): Friedwagner and others, who deny the possiblity of influence in the north, Amake a mere assumption because we have no direct information on the connections between the Romanized population and [the migratory peoples].@ (The problem is that we have no information about a Romanized population in the period in question north of the Danube.) Rosetti continues the argument by saying that for those who admit that the Rumanian language developed on a large territory of Roman colonization, the existence of contact between such a population and the Slavs also north of the Danube appears to be beyond doubt. (This is not, of course, a linguistic argument and no valid argument at all, because the assumption of a large, both south- and north-Danubian territory in which the Rumanian language developed is not proved. On the contrary, it is denied exactly by the Slavic influence upon Rumanian, which excludes the territory of present day Rumania, cf. below.) The next argument presented is the assumption that Athe Romanized population was preserved in Dacia after the Roman domination ceased.@ (This is not impossible, but it has little relevance for the question discussed; cf. above, p. 209: Latin-speaking populations remained after the Roman retreat in Pannonia, Noricum, Britannia, etc., but they did not survive for many centuries. As regards the statement of Petrovici, it should be made clear that the Roman domination lasted 600 years SOUTH of the lower Danube. Several circumstances indicate that the original areas of the Vlachs were situated much more to the south).

Thus, Rosetti showed in this passage, although indirectly, that there is no LINGUISTIC argument for the assumption of a South Slavic influence upon early Rumanian north of the Danube.

There is, in fact, specific evidence showing a South Slavic influence on Rumanian exerted in the Balkan peninsula. There are words of Latin origin borrowed by South Slavic and then from this language by the Rumanians. Rumanian sigur, matur, and oÛet derive from South Slavic, as shown by the accent on the first syllable in sígur and mátur, (as opposed to Latin secúrus and matúrus) and o instead of a and ts instead of c(e) in oÛet (cf. Latin acetum).

Thus, two main conditions were established: (1) The Rumanian language contains South Slavic elements with a sound pattern from the 7thB8th centuries and the most numerous Slavic elements in Northern Rumanian show the sound pattern of Middle Bulgarian (11thB12th centuries). (2) Before the 12thB13th centuries, only SOUTH Slavic elements were transferred to Rumanian.

Considering the historical situation, these facts indicate that Rumanians cannot have been living before that period (a) in central and northwestern Transylvania. Archaeological remains excavated there show namely western and eastern Slavic cultures and the placenames of Slavic origin in that area show a western Slavic pattern. (b)The Rumanian language also lacks Ukrainian elements of a pattern shown before the 12thB13th centuries, which excludes also (at least northern) Moldavia. (c) But not even the Bulgarian influence upon Rumanian could have been exerted north of the lower Danube, because the Bulgarian domination was too short there. The Bulgarian state was founded in 679 south of the lower Danube and expanded in a southward and southwestward direction. In the 9th century, the Bulgarians occupied present day Muntenia and southern Transylvania. However, latest at the beginning of the 10th century, Bulgaria lost its power north of the Danube. The Bulgarian domination lasted at most a century in Muntenia and southern Transylvania, obviously too short for the impact of South Slavic on Northern Rumanian. After the Bulgarians, Petchenegs occupied Moldavia and the Valachian plain. The Hlincea I culture in southern Moldavia showed signs of ravages around 930 AD and disappeared some decades later. This, as well as the end of the Dridu culture on the Valachian plain not long after 1000 AD is ascribed to Petcheneg attacks (cf. below, pp. 278B279). The Transylvanian basin and most of the valleys were, beginning with the 10th century, inhabited by Hungarians. From the end of the 11th century, the Cumans dominated the extra-Carpathian territories (cf. below, p. 280). Thus, in the period of the most intense Bulgarian influence on Northern Rumanian, the Bulgarians no longer ruled north of the lower Danube, Slavic archaeological cultures disappeared there, Turk peoples were living in the extra-Carpathian territories and Transylvania was organized into the Hungarian kingdom and was increasingly populated by Hungarians. Neither the absolute amount of the South Slavic influence (cf. above, pp. 102B108) nor its increase upon Northern Rumanian after the 10th century can be explained under these circumstances.

Consequently, THE SOUTH SLAVIC INFLUENCE ON RUMANIAN BEFORE THE 12thB13th CENTURIES WAS EXERTED SOUTH OF THE LOWER DANUBE.

Historical records and abundant archaeological finds testify to the presence of Goths and later of Gepidae in most of present day Rumania from the end of the 3rd to the 7thB8th centuries. Some historians have affirmed that these populations did not have close contacts with the ADaco-Romans@, others asserted a symbiosis between the two populations (cf. above, pp. 190). Rosetti stated that there are no Old Germanic elements in the Rumanian language, and that, based upon historical records and archaeological finds, Athe permanent habitation of the populations of Germanic language north of the Danube and their symbiosis with the local Romanized populations are well-proved facts.@ This section in Rosetti´s ILR 1986 does not contain any answer to the question: how can these two facts be reconciled with the theory of continuity? (The affirmation that Old Germanic elements could have been borrowed by the Rumanians also south of the Danube is no sufficient explanation, nor is it valid.) Also ILR 1969 evades to answer this question. Iordan & Manoliu, although criticising the argument presented more than a century ago by Rösler, that the absence of Old Germanic words in Rumanian is not compatible with the theory of Daco-Roman continuity, do not give any explanation of the fact itself.

The Burgundians dominated the eastern part of southern France for about a century until they were defeated by the Franks (532B534 AD).

The Old Germanic influence on the Romance languages was considerable, with hundreds of lexical elements. The largest number (more than 500), of these were left by the Franks, followed by the Longobards, with about 300. Out of words from the language of the Visigoths, about 130, from that of the Ostrogoths, 70, and from Burgundian, about 50 words are still extant in the Romance languages. (These figures are taken from Întroducere în lingvistica romanic|, by I. Iordan & Maria Manoliu, 1965, p. 276, who quote Gamillscheg, Romania Germanica, 1934B1936. According to Iordan & Manoliu, the figures are somewhat exaggerated.)

From the viewpoint of the modern Romance languages, the Gallo-Romance idioms contain most words of Old Germanic origin, followed by the Italian dialects and, lastly, the Ibero-Romance languages. However, the number of personal names of Old Germanic origin in Spanish and in Portuguese is considerable (cf., for instance, Tagliavini Orig Lingu Neolat 1969, p. 305). In Rumanian, as shown above, there are no Old Germanic elements (not even personal names).

In the Balkan peninsula, the role of the Old Germanic peoples was of a very limited importance (cf. above, p. 16). In contrast, large areas north of the lower Danube were during the 4th to the 7th centuries inhabited and dominated by Old Germanic peoples (Goths and Gepidae). Assuming with Rumanian archaeologists that they were living together, in the same settlements, with Daco-Romans, i.e., the ancestors of the Rumanians (Mugeni, LechinÛa de MureÕ, etc.), one would expect to find Old Germanic elements in the Rumanian language. The number of these would not necessarily be as high as it is in the western Romance languages, but the total absence of this influence is very difficult to reconcile with the idea of a symbiosis lasting several centuries. One may assume, as was usually done earlier, that the Daco-Romans fled to the mountains and did not have contacts with the Germanic peoples, who lived in villages in the valleys. But a shepherd population is not entirely independent from peasant groups; and the Vlach shepherds proved, both by their Romanization as well as by their borrowing of a tremendous amount of Slavic elements, contact with surrounding populations. However, there are the Basques living in the Pyrenean mountains whose ancestors were once subdued by Old Germanic peoples and their language lacks, in spite of this, any Old Germanic influence. The absence of Old Germanic elements in Rumanian may thus be considered an argument ex silentio, and as such, not decisive in itself, although by no means without significance.

In chapter III, p. 191B193, a number of hypotheses were presented concerning the placenames and names of rivers in present day Rumania.

Poghirc assumed about five placenames recorded in ancient texts that they may exist even today, because their names are similar and the geographical situation of these settlements is the same as assumed for the settlements mentioned in ancient texts. These are Drencova, Hîrsova, Mehadia, Oltina, and B|roi.

Drencova, in the Banat, is believed to continue Dric(c)a, mentioned by Iordanes. It may be a non-identified tributary of the Danube. Another river, a tributary of the Tisza, is Dregkwn or Drhkwn. There are several places with this and similar names in Rumania. Iordan mentions the following: Dranov, Dranovatul, Drânceni, Drâncova or Drencova (in the Banat), Drencea (in the region of IaÕi), Drincea, and Drinova. It is of course not probable that all these names would derive from ancient Dric(c)a or Dregkwn. Iordan derives these names from Slavic dren ´cornel tree (Cornus mascula)´ (Bulgarian dr�n ´id.´ and dr�nov ´of cornel tree´). Many Slovakian, Serbian, and Croatian placenames contain this word.

HîrÕova is assumed to continue ancient Carsium in Dobrogea. Also in this case, there are several villages with similar names: besides that in the region of ConstanÛa, there is a village with the same name in Vaslui, and in the region of Craiova, there is CruÕovul. This name exists also in Arumanian: CruÕuva. Before the 15th century, HîrÕova was written in Slavic documents Hrušova or Hruševa. Assuming that HîrÕova continues ancient Carsium, the c > h change must be explained. Poghirc proposes that an Aautochthonous@ h was in Latin written c, because h did not exist in Latin in that period. However, all places with this name cannot continue ancient Carsium but are evidently of Slavic origin, from Slavic hruša, kruš(/)ka ´pear tree´. Many placenames all over the territories populated by Slavs derive from this word. Iordan´s conclusion is that HîrÕova in the region of ConstanÛa may continue, Aindirectly and at least formally@ ancient Carsium, but it was modified by Slavic hruš(k)a.

Mehadia is assumed to derive from Latin Ad Mediam, although not from Latin medius but Aprobably from an autochthonous *Mehedia, Mehadia, with intervocalic h not recorded by the Romans@. This hypothesis cannot be accepted, because, as indicated by the first mentioning in a document (1323: Myhald, Mihald, etc.), this name derives from the Hungarian personal name Mihály, with the suffix -d.

B|roi, also in Dobrogea, derives probably from ancient Bireo, Beroe.

Thus, out of four placenames and a river name for which Aa perfect geographical correspondence@ is assumed, two are of Slavic and one is of Hungarian origin. The rest, two placenames, most probably continue the ancient name, but the settlements in question (as also HîrÕova) are situated in Dobrogea, south of the Danube. They belong to the large number of Roman settlements along the southern shore of the Danube whose names were preserved B in a Slavic sound pattern. These Roman towns, although exposed to the ravages of the barbarians, existed until the end of the 6th century and a Latin-speaking population lived there in the period of Slavic colonization. This explains the fact that these, together with a large number of Latin placenames in the Balkan peninsula, were not forgotten. Many Latin names disappeared also in the Balkans, in those cases in which the settlements were abandoned, and occupied by the Slavs after some time had elapsed. To such settlements, the Slavs gave their own names, for example Brani…evo ´the tower of the defender´ for Viminacium (cf. above, p.20). Of course, one should also take into account that changes may have occurred by chance, reducing the number of preserved names in certain areas. However, north of the lower Danube we find not only a reduced number of inherited placenames but a total absence B not a single name of a Roman town or any other kind of settlement was preserved. The most obvious explanation of this is that the Slavs did not find Latin-speaking inhabitants when they migrated to these territories in the 6thB7th centuries.

It is generally considered that the ancient names of the great rivers were handed down to Rumanian by the Slavs. In the course of time, however, hypo- theses were put forward assuming that these names were inherited directly. Poghirc has tried to show that Slavic or Hungarian were not necessary to explain the Rumanian names MureÕ, Olt, SomeÕ, TimiÕ, and Tisa. The argument is based on assumed sound changes which would have occurred in Alate Daco-Moesian@ between the 3rd and the 6th centuries: -si- > š and a > o (u). Ancient Marisia and Timisia changed to MureÕ and TimiÕ, respectively, Samus to SomeÕ and Alutus to Olt. A difficulty is Tisia, which, according to this hypothesis would have changed to *TiÕ. This may be explained, argues Poghirc, by a different pronunciation by the local population of this name, not corresponding to what was recorded.

The basis of this reasoning is very weak and uncertain, which is, in fact, also pointed out by Poghirc: AWe emphasize the hypothetical nature of these phonetical transformations, based upon a scanty and uncertain material.@ One may add that if Dacia Traiana had had a Latin-speaking population in those times, all these names would have been used by this population IN A LATIN FORM and a sound change in ADaco-Moesian@ between the 3rd and the 6th centuries would not have affected them.

Another hypothesis is that old Rumanian names were TRANSLATED by the Slavs, and in Transylvania by the Hungarians (cf. above, p. 192). There is, however, no proof that this mechanism was instrumental. It is not known whether the name of a tributary of the BistriÛa, Repede, existed before the Slavic population of the area. It might as well have been given by Vlachs who migrated there much later. (´Fast´ is a usual name given to mountain rives and brooks.) It was originally Dr|ganu who assumed that from Latin Alutus, Rumanian Alt resulted and this was the name of the river until the Slavs and the Hungarians came to the region and changed it to Olt. The Rumanians then borrowed the Slavic or the Hungarian form: Olt, while the Germans, when they in the 13th century settled in Transylvania, borrowed the old Rumanian form Alt.

This is at most a possibility, although very unlikely. It would be justified to formulate such hypotheses in a territory in which most of the geographical names are of Latin or of Rumanian origin. This is, however, not the case here: NOT A SINGLE PLAECNAME OR GEOGRAPHICAL NAME OF PROVED LATIN ORIGIN AND INHERITED DIRECTLY BY RUMANIAN IS FOUND NORTH OF THE LOWER DANUBE. Assumptions that merely permit the possibility of such origin are in these circumstances meaningless. For the most probable explanation of the river names cf. below, p. 272B 273. The example presented by PuÕcariu (cf. above, p. 192) of Petra translated by the Slavs into Kamena has no relevance for the territory of former Dacia Traiana. This settlement in Dobrogea belongs, together with B|roi and Oltina, to the Roman settlements along the southern shore of the lower Danube, whose names were preserved (in a Slavic sound pattern) obviously because the Slavs found there a Latin-speaking population.

One has tried to explain the absence of inherited Latin names north of the lower Danube by assuming that the ADaco-Romans@ adopted the names given by the Slavs (cf. above, p. 192). These explanations cannot be accepted.

People who migrate to new territories usually borrow the geographical names and the placenames they find there. The numerical circumstances and even the cultural level of the populations in question have a subordinated role in this process. The decisive circumstance is the existence of a name (of a town, of a river, etc.). Thus, for example, the English borrowed very many names of rivers and streams in America, Africa, and elsewhere. This group of nouns is the most easily transferred element of language. In American English, Athe numerous names of Indian origin are almost the only loanwords from that source.@

In a series of former Roman provinces, the Latin-speaking population was in the course of time replaced by speakers of other languages. In many of these provinces, placenames of Latin origin are still in use. In England, names ending in Bchester, Bcaster (Manchaster, Lancaster, etc.) continue Latin Bcastra, the ending Bcoln (Lincoln, Old English Lindcylne) derives from Latin colonia. In Noricum and in Raetia, the Slavs and the Germans borrowed many Latin placenames and names of rivers and streams from the Latin-speaking population they found there in the 6thB7th centuries. About 200 of these are still in use in the territory of the former Roman provinces: Latin Laureacum > German Lorch, Lentia > Linz, Bataua (Castra) > Passau, Licus > Lech, (Ad) Pontes > Pfunzen, Celeusus > Kelsbach, etc. In Pannonia, the situation is varying in the different areas, depending on historical circumstances. In the northeast, the changes of population occurred more abruptly and more often, and, consequently, there are no placenames of Latin origin in that part. In the west, however, there are: Latin Vindobona > German Wien, Poetovio > Ptuj, Siscia > Sisak. The Hungarians found a number of such names when they, at the end of the 9th century, took possession of Pannonia. They borrowed several river names: Latin Arrabo > Hungarian Rába, Mursella > Marcal, Sala > Zala, etc., and Savaria was in use (along Hungarian Szombathely) until the 19th century.

In the Balkans, the Rumanians have preserved a number of ancient place-names in a sound pattern which shows direct inheritance: S|run|, L|sun, Fl|rina, etc. This indicates that they have lived in (or were in contact with) those areas continually since the Roman times, which implies Roman-Rumanian continuity. North of the lower Danube, several toponyms of Scythian, Sarmatian, Petcheneg, and other origin were transferred to Rumanian B and in Transylvania, to Hungarian B via Slavic. The earliest toponyms borrowed directly by the Rumanians from people living north of the Danube were (1) from the Cumans, in the 12thB13th centuries (cf. below, pp. 280B281), and (2) from Slavic, showing the Bulgarian sound pattern in approximately the same period (cf. below, p. 248).

The explanations given (cf. above, pp. 191B192), in themselves more or less valid, can only explain why not ALL Latin toponyms were preserved in the former Roman provinces. They are not sufficient to account for the total absence of inherited Latin placenames and geographical names in Rumanian spoken north of the Danube. The absence of such names is not compatible with the assumption that Rumanian is the continuation of Latin spoken in Dacia Traiana.

This argument is erroneous. There are several Rumanian words of South Slavic origin with an older sound pattern: m|gur|, Õchiau, scump, munc|, dalt|, etc. (cf. above, p. 98B99). These words were borrowed early, probably before the 9thB10th centuries, and were not Aadapted to the actual pronunciation@ (in Slavic). Words borrowed from a foreign language usually develop according to the rules of the borrowing language, not those of the original one. This is true, of course, also regarding placenames and geographical names, which, moreover, are among the most resistant elements of any language. This fact gives them their great significance in historical research.

All the placenames and geographical names north of the Danube borrowed by Rumanian from Slavic are of a much later sound pattern (of the 11thB12th centuries and later) than the above-mentioned words (from the period before the 9thB10th centuries). Since it is known that Slavs were living in the territory also earlier (in the 7thB10th centuries), this circumstance confirms the conclusion above, that Rumanians did not live north of the Danube in those centuries.

The above conclusion applies to all areas north of the lower Danube. However, with regard to the hypothesis that the ADaco-Romans@ were, during a certain period, living in far away valleys and high mountains, the situation in these areas should be analysed separately.

If the ancestors of the Rumanians had been living, in the period of Slavic colonization, in the region of the high mountains, we may assume that they could possibly have forgotten the names of the towns. They would have preserved, however, at least some names of villages and names of rivers and streams in the territory in which they were living (allegedly the Transylvanian Alps [MunÛii Apuseni] and the southern Carpathians). The rivers ArieÕ, Ampoi, SomeÕ as well as Cerna, SebeÕ, DâmboviÛa, flow in this region. The upper course of the SebeÕ, a tributary to the MureÕ, situated in the area of mountains about 2000 m high, is in the heart of this region; it is called Bistra, a name borrowed from Slavic. Also Târnava (Mare and Mic|), MureÕ, and Olt should be known by a population living in those mountains. However, not a single of these names was by Rumanian inherited from Latin. As shown below, pp. 271B273, they are either ancient names, or names of Slavic or of Hungarian origin, and ALL WERE BORROWED BY THE RUMANIANS FROM ONE OF THESE TWO LANGUAGES.

The situation in the mountainous region south of Or|Õtie (Hung. Szászváros) has been investigated by Mircea Homorodean. This author tried to find ancient Dacian names in the region around Sarmizegetusa, once the centre of the Dacian state. The result was negative: not a single geographical name or placename of Dacian or of Latin origin was found. (It may be added that, along with Rumanian names, a considerable amount of Hungarian and also German [Transylvanian Saxon] names were found in the area.)

The idea of the development of Rumanian north of the Danube is not compatible with the Rumanian dialects. If the ancestors of the Rumanians lived, during the 4thB6th centuries both south and north of the lower Danube, there would be at least some dialectal differentiation in their language between the territories south and north of the river, as a consequence of the Roman limes; even if one would assume more significant contact across this frontier. Such differences in Latin have been caused by the division of the Roman Empire into two halves at the end of the 4th century, and the frontier between the two halves was much less of an obstacle of contact between peoples than was the Danubian limes. The dialectal division of Rumanian (presented above, pp. 119B123) shows, however, no traces of a division caused by the Roman limes. It goes, instead, between Arumanian with Meglenitic in the southern parts of the Balkans on the one hand, and Northern Rumanian with Istro-Rumanian on the other. These two dialects are spoken in Istria, in the Timok valley, and were spoken during the Middle Ages in a large area between Skopje and Prizren up to Niš and Sofia, and also to the west of this territory (cf. above, pp. 26B33 and map No 3).

Northern Rumanian is now spoken mostly north of the lower Danube, where it is the mother tongue of about 20 million people. However, in spite of this, and in spite of the large territory in which its speakers are living, this idiom shows a remarkably weak dialactal differentiation (cf. above, pp. 124B131). Most relevant in this context would be the situation in ancient periods. Unfortunately, data are extant only from the 15th century on. The opinions concerning the dialectal differentiation of the texts written in the 16th century differ. Some authors assume more phonetic differences as compared to the present situation, others, considering mainly the lexical elements, deny the existence of sub-dialects in that period:

Similar connections must have existed, for example, also between the early Albanians B a typically shepherd population B and in spite of this, their language developed, on a much smaller territory than present day Rumania, two distinct dialects (Tosc and Gheg). The Rhetoromance language has a number of dialects. Bulgarian developed marked dialectal differences since its speakers settled, in the 7th century, in the northeast of the Balkans. Intensive contacts are therefore scarcely sufficient to explain the low dialectal differentiation of Northern Rumanian.

These observations may be applied to Northern Rumanian, independently corroborating the conclusion of emigration to the territories north of the lower Danube by groups of Vlachs relatively recently (i.e., starting in the 11th century.) It is hardly possible that a population living in such a large territory as Rumania from the 3rd century AD to the 15th (and the 20th) century i.e., for 1200B1700 years as assumed by the theory of continuity, should not have developed any significant dialectal differences in their language.

PuÕcariu tried to explain this mainly by the following circumstances: (1) The Vlach shepherds followed the ways of transhumance (les voies de transhumance) from Transylvania across the Carpathain mountains to adjacent plains: the Banat, the Valachain plain, and Moldavia. (2) In Transylvania proper, there were no Rumanian political, cultural, or religious centres around which the Rumanian inhabitants of this province could have aggreagated. The Hungarian and the Saxon towns were for them only market places, administrative and jurisdictional centres. PuÕariu affirmed also that ALes cartes de l´Atlas linguistiques nous permettront d´étudier l´expansion des Roumains de Transylvanie dans toutes les directions.@

According to this, the Moldavian and the Muntenian sub-dialects would be the continuation of the Transylvanian sub-dialects. This, however, is an interpretation of the information provided by the maps. The maps show only that there is no Transylvanian sub-dialect. Instead, Rumanian spoken in the south of Transylvania is essentially the idiom spoken in adjacent Muntenia and Oltenia, and in the northeast, that of Moldavia. This suggests an opposite direction as compared to that suggested by PuÕcariu. Speakers of Rumanian living in Moldavia migrated in the first place to northeastern Transylvania, and those from Muntenia, to southern Transylvania. The main criteria of a Transylvanian dialect transferred to Moldavia and Muntenia by emigrants would be (1) a number of specific traits only found in Transylvania (because even assuming large-scale migrations, it is unlikely that ALL features of a dialect specific to a certain territory would be transferred to other areas) and (2) the presence of at least a number of specific Transylvanian features in both Moldavia and Muntenia.

There are also specific lexical elements whose distribution in Rumania indicates transfer from the extra-Carpathian territories to Transylvania. Thus, for example, the speakers of Rumanian in the Banat, most of the MunÛii Apuseni, CriÕana, and the region of Satu Mare use the inherited Latin word nea (Latin nevis) ´snow´. In the northeast of Transylvania, om|t is used, as also in Moldavia, a loanword from Ukrainian. In Muntenia and Oltenia, z|pad|, (of Slavic origin) is said, and this is the case also in adjacent areas of southern Transylvania. Another word of Turkish origin and transferred to Rumanian spoken in Muntenia and Oltenia via Bulgarian is cioban ´chief shepherd´. In that territory, as well as in Moldavia, crude oil, p|cur|, was extracted and those who sold this product were called p|curar. A homonymic clash occurred between this and p|curar ´chief shepherd´, from Latin pecorarius, which therefore was replaced by cioban. This word penetrated to parts of southern Transylvania (while other areas in that province preserved p|curar). The same map by Cazacu shows that mocan ´shepherd; simple, uncultivated man´ is used in an area in southeastern Moldavia. In roughly the same region, there are many placenames based on ungurean ´people from Hungary´ or ´Rumanian from Transylvania´, which indicates that people from Transylvania did migrate to the territories beyond the Carpathians.

THE ORIGINS OF THE RUMANIANS