The Carpic, Gothic and Gepid attacks which afflicted the Roman provinces to the north and the south of the Lower Danube are well known from the history of antiquity and had been described in the previous chapter of the history of Roman Dacia. As a result of these attacks, the position of Dacia Superior became critical. The soldiers manning the Roman frontier forts in Transylvania received their pay until the end of the joint reign of Philip I and Philip II; a few garrisons are known to have existed under Decius (249- 251).
The Roman limes in Transylvania was the military installation of a strong, self-confident militant superpower given to taking the offensive. The Impe- rium did not close the passes — with the exception of the Vorostorony Pass
— but controlled them with a chain of advanced turrets, visible from the military forts. The Ldpos Mountains in the north, the Kelemen and GOrgenyi M ountains and the Hargita in the east, the Bereck Alps in the south were a kind of no m an's land — Transylvanian Dacia can be best likened to an enormous classical theatre whose gates and seats Rome ceded to the bar
barian audience, reserving the stage for itself.
The camps, and later forts, of the auxiliary troops were established on the plateaus usually beside or near a major waterway in order to facilitate transport and comm unication, and were located on low hills or terraces which offered an excellent overview of the lowland encircled by the m oun
tains and their passes. In other words, Rome merely kept an eye on the great natural protective ring which — the myth that has persisted into m od
ern historiography notwithstanding — never, for one single moment, really protected Transylvanian Dacia.
This system had collapsed by 160-170, and could no longer be effec
tively held by the Roman Empire under the thrust of the barbarian attacks in the middle of the third century. The successive raids and incursions from 248 onwards ravaged and destroyed the stone-walled forts and towns of Scythia Minor, Moesia and Thracia. The military prowess of the barbarian armies clearly indicates that none of the border forts in Dacia could resist or seriously impede their onslaught. The undefensible eastern Transylvanian limes was relinquished in the 250s without even an attempt at its defence.
About three or four towns and the single legionary castrum of the prov
ince held out for some time in the western half of the province, but only along the main road leading southwards: the garrison of Micia protecting the western exit of the Maros Valley was withdrawn under Valerian (260).
On the evidence of hoarded coins, the last bastion in the heartland of Transyl
vania, the castrum of Apulum, was under attack from 268 onwards, when the m ain road leading through Krasso-Szoreny was also abandoned (the Galacs coin hoard).
C ontem porary literary sources and the archaeological evidence both demonstrate that the Roman army, the soldiers' families and all other civil
ians whose livelihood was strongly linked to the military were evacuated from the province at the beginning of Aurelian's reign (271) at the latest.
Life came to an abrupt end in the forty-eight Roman castella and the settle
ments that depended on and lived off these forts (the so-called vici auxiliari) which formed an arc extending from the Temes region to the Vorostorony Pass. That this was indeed so is borne out by the castella themselves. Those which were not resettled and built in during the Hungarian Middle Ages (e.g. Bereck, Tih6, Kosdly, Varmezd, Magyarigen, and also Pozsesna on the Low er Danube) to this day offer the same desolate spectacle w ith their earthen ramparts and deep ditches as the agri decumates of Baden-W iirttem- berg which were abandoned at roughly the same time, or the border forts of the Antonine Wall in Scotland that had been relinquished somewhat earlier.
They are literally "em pty" both inside and out, for they had long fallen into disuse by the time of the large-scale building activity of the late Roman age.
It is probable in the case of the four municipia that held out the longest — Napoca, Potaissa, Apulum, and Ulpia Traiana — that the small and wretched population groups within or around their walls at first accepted Gothic overlordship. However, the traces of these groups in the archaeological records (a few burials) do not extend beyond the close of the third century A.D. At the same time, the buildings within the stone walls — som e of w hich have survived into the Modern Age — decayed rapidly, and by the turn of the fourth and fifth centuries the principia (military headquarters) in Potaissa was being used as a burial ground by groups of eastern barbarians.
The rural homes, the villae, and the farmsteads of the advocates of roman- ization had perished to such an extent that in the fourth century the Goths had already opened cemeteries on their sites (e.g. Palatka).
There is only one way a "rom anized" population could possibly have survived: if they had resorted to active defence and had withdrawn into hastily erected hilltop forts and fortified sites, as indeed had the provincial population of some areas lying within the empire — the inhabitants of the Balkans and of southwest Pannonia, the population of the Eifel-Hunsriick hill country between the Rhine and the Mosel, and farther westwards, the population of the Ardennes. The terrain of Transylvanian Dacia was cer
tainly extraordinarily well suited for this kind of defence. However, there are no traces of late Roman refugia or fortified places in Transylvania, and the lack of these contradicts the much-quoted "self-defence" theory.
Not one single late Roman or early mediaeval source mentions or knows of a "surviving" Roman population in Transylvanian Dacia: the one-time
Roman towns, settlements and forts perished without exception, not one single language or literary source retained their names into the Middle Ages.
W hat has survived is of linguistically uncertain origin, the names of a few major waterways which the Romans themselves considered a prehistoric heritage: the names of the Temes, the Maros, the KorOs, the Szamos and the Olt. The river names "Am pelus-O m poly" and "Tierna-Cserna" are sim i
larly a legacy of the pre-Roman era. As for the name of the Aranyos River (documented in a charter from 1075: qui dicitur hungaricae aranas, latine autem aureus [which in Hungarian is called 'aranas' but in Latin 'aureus']), which in Hungarian was borrowed from the Iranian, it is "derived" from the Latin aureus only by those who go in for vulgar etymologizing. The name given to this particular river is rather obvious in view of its abundant gold content (Hungarian arany), a fact borne out also by its Slavic name (Zlatna-Zalatna from zlato, "gold ").
The complete disappearance of antique place-names from Dacia proves the complete assimilation of the surviving population fragments, especially if one remembers that, although numerous place-names in Britannia, in the Rhineland and the Upper Danube region of yore are preserved in English, Dutch and German, the Roman population that continued to live there well over two hundred years after the surrender of Dacia was likewise com pletely assimilated. The funerary rite of the Transylvanian "late Rom an"
cem etery (Barathely I) is undoubtedly reminiscent of the so-called scattered crem ation burials form erly practised in Dacia — and extensively docu
mented in Pannonia until Probus (276-282) at the latest — but what must be recalled is that the grave goods (intact vessels, lamps, coins) characterizing genuine Roman burials do not occur in these graves. The Bar&thely graves all contained expressly "barbarian" meat offerings, whilst a few burials yielded fibulae and vessel sherds burnt in accordance with the barbarian rites practiced by the population of the M arosszentanna-Cherniakhov cul
ture. Similar burial rites have been observed in the Carpic-Gothic cem eter
ies of Moldavia (Danceny, Etulija, Baltzata, Hanska-Luterija II and Oselivka).
M arosszentanna-Cherniakhov type finds (vessels, combs, jewellery) have also been found from the settlements belonging to these centuries, and thus the population group using the Barathely burial ground originated from the Barbaricum. It is, moreover, rather improbable that a closed com m u
nity of "surviving Rom ans" could have successfully lain low near one of the most bustling main roads leading through Dacia (between Medgyes and Segesvar).
Relics of the late Roman rites and costumes characteristic of the Tetrarchy and later do not occur on the left bank of the Danube. The so-called cruciform brooches — insignia bestowed by Rome on Roman subjects — only reached the Barbaricum as booty, with sporadic finds of such fibulae am ong the Germanic Quadi, the Sarmatians of the Great Hungarian Plain, the Gepids of the eastern part of the Great Hungarian Plain (Muszka), and the Transyl
vanian Goths (Obrazsa, Lemheny and Vecel). These "crossbow " brooches do not confirm the presence of a Roman population in the Barbaricum, and this is doubly true of specimens that found their way into Transylvanian private collections and museums from Pannonia, Moesia and even Italy in
the course of the past century. Another absurd suggestion — and one that betrays considerable ignorance of human nature — is that the presence of Roman coins is evidence for the presence of "rom anized" Roman persons.
In fact, the Roman coin "circulation" of other regions of the Barbaricum such as the Great Hungarian Plain or the Polish Lowlands shows little, if any, divergence from the circulation of post-271 Roman coins in Transyl
vania.