Actual State of Research of the Csörsz- or Devil’s Dyke. In: Identităţi culturale locale şi regionale în context european. Studii de arheologie şi antropologie istorică. Eds.

Eszter Istvánovitz, Valéria Kulcsár

Actual State of Research of the Csörsz- or Devil’s Dyke

ESZTER ISTVÁNOVITSJósa András Múzeum Nyiregyháza, Ungaria [email protected]

VALÉRIA KULCSÁRUniversity of Szeged, Szeged, [email protected]

Interpretation of the Csörsz Ditch (or Devil’s Dyke) belongs to the neuralgic points of the archaeological research of Roman Age Barbaricum in Hungary.

Th e complex research of the ditch started in the 1960s had a number of premises (works of Flóris Rómer, Róbert Frölich or Ferenc Pulszky).1 Importance of the Csörsz Ditch in the Roman Age is well shown by a number of studies on the question. In 2003 the summarizing volume published originally in 1983 was re-edited on the occasion of the XIXth Congress of Roman Frontier Studies organized in Pécs.2

Scholar research of the Csörsz Ditch goes back to the 19th century. Th e fi rst to collect writt en sources was Flóris Rómer.3 Th e earliest data appears in the diploma of Hungarian king Béla IV, from the year 1267 that affi rmed the deed of foundation of the Százd abbey from 1067. Th e “great ditch” mentioned here was identifi ed by the scholars with the Csörsz. Under the name Csersz it was mentioned in István Székely’s chronicle from 1558, at the events of year 718. Th e information we get from here is the following. King Csersz made this ditch to protect the country before the time of Att ila. Th e ditch was deep, and on the top of the vallum, trees were grown in order to preserve the soil of the earthwork. Th ere were gates with guards. Th e dyke protected people from the enemy. Th ere were several dykes between the Danube and Tisza, but later the ditches fi lled up.4

Th e Csörsz Ditch is an earthwork consisting of 3–4 vallum and fossa. Th e territory encircled by it is about 60, 000 sq.km. From the west and south the Danube borders this territory. Th e ditch starts above Budapest at Dunakeszi and runs in eastern direction towards Ároktő. East of the Tisza River, before reaching Újfehértó the dyke turns to the south and continues its way almost in N-S direction down to the Danube, towards Viminacium (Kostolac, Serbia). Th e length of its track is about 520–550 km. At several places several lines (two, three and even four) can be observed. During the complex research started in the 1960s, a total of 1260 km long earthwork was revealed. Th e dyke protected the population of the Great Hungarian Plain from the north and east. Ditches situated north or east of the vallum were 2–4 m deep and 5–9 m wide. Th e foundation of the vallum could be 7–10 m; its height was about 2.5–3.5 m. According to some calculations around 10–15 million cubic meters of soil were necessary for the erection of this earthwork. Th is could take one year for 83, 300 people (taking into consideration a fi ve-day week).5

1 Rómer 1876. 39–77; Frölich 1887; Pulszky 1891.2 Garam–Patay–Soproni 1983; Garam–Patay–Soproni 2003.3 Rómer 1876, 39–77, map between pages 42 and 43; Balás 1961, 331.4 Székely 1854, VII. 6.5 Patay 1969, 111; Garam–Patay–Soproni 1983, 13–17.

ROMANI ȘI BARBARI LA GRANIȚA IMPERIULUI

312 | ESZTER ISTVÁNOVITZ, VALÉRIA KULCSÁR

Beside the so-called longitudinal dykes there are also so-called interior dykes.6 Th eir system is less investigated and it is very questionable whether there was any historical connection between the Csörsz and the so-called interior or Roman dykes. A part of the interior earthworks appears on the military map compiled on the order of Austrian Emperor Joseph II. Here it is called “minor Roman dyke”, while there is another marked “major Roman dyke”.7

Th e whole length of the so-called “minor Roman dyke” can be found in Bačka, Serbia.8 It starts at Apatin, runs to E-SE along the line Doroslovo (Doroszló) – Srpski Miletić (Militics) – Ratkovo (Parabuty/Paripás) – Depotovo (Deszpot Szent Iván/Úrszentiván) – Kulpin (Kölpény) – Alpár – Bački Jarak ( Járek/Tiszaistvánfalva) until Gospođinci (Goszpodince/Boldogasszonyfalva). Here it turns to the north and parallel with the Tisza runs as far as Čurug (Csurog). From Gospođinci parallel and west of it, starts another line running north along Bačko Gradište (Tisza-Fölvár/Bácsföldvár) – Bečej (Óbecse) – Bačko Petrovo Selo (Bácspetrovoszelló/Péterréve) where it reaches Tisza. At the end of the 19th century its height reached 3–10 m, it was 16–20 m wide at the bott om and 4–8 at the top. Ditches running together with the vallum could be well seen. At the Apatin section the ditch ran from the northern side and starting from Doroslovo there were ditches at the both sides of the vallum observed, while in the vicinity of Kisač (Kiszács) the ditch got to the southern side of the vallum. Together with the ditch the bott om of the earthwork reached 20–25 m. Its fi rst section was marked on the maps 65 km long, but originally – if we take into consideration also the missing, probably destroyed sections – it could reach 100 km. Th e northern, independent section of the earthwork is of similar structure. It is 29 km long and the ditch is situated on the eastern side. At certain points there are missing parts. Th e line of this section is not straight, but “winding”.9 According to Kálmán Gubitza’s records dyke sections of diff erent length could be observed also in other parts of Bačka (such a short section is situated at Ravno Selo (Ósóvé), another one from Apatin to Sonta (Szond), and a third one from Svetozar Miletić (Nemesmilitics) to Čonoplja (Csonpolya) and through the territory of Kljajićevo (Kerény) towards Crvenka (Cservenka). Th ese sections were similar in their structure to the “minor Roman dyke”.10

Th e “major dyke” starts at Novi Sad (Újvidék), runs towards N-NE, fi rst in the square between Temerin – Gospođinci – Bački Jarak – Žabalj (Zsablya), then north of Gospođinci it crosses the “minor dyke” and reaches the Tisza between Čurug and Bačko Gradište. On its whole length, a ditch follows the vallum from the western side. Róbert Frölich measured it at two points and found that its eastern slope was 5.3–9.5 m, while the western one was 9–13 m. Th e ditch was 8.6–10.3 m, the top of the vallum: 1.3–5 m.11

North of the Bačka earthworks, in Hungary the territory between the Danube and Tisza (historic name of the region: Cumania Minor) is crossed by another earthwork, SW from the mouth of Körös. On 19th century maps it starts from Baja, runs towards Csépa, then, east of the Tisza, on the northern bank of the Körös River reaches Transylvania. At the same time, on Flóris Rómer’s map this dyke started from Bátmonostor to N-NE. It could be seen in the vicinity of Kunszentmárton, then turning to the east at Szelevény and running parallel with the Körös River up to Szeghalom. According to a datum recorded at Bodoglár, a ditch followed the vallum from S-SE.12

Work-group organized by Pál Patay for the research of the Csörsz, studied the cross-dykes of Cumania Minor and also the one running along Körös Valley. In the course of this work they mapped the remains of the dyke running from Baja and determined that it goes in the direction of Szeged oriented almost W-E, so it has no connection with the earthwork along the Körös. Its structure – the fact that there are ditches at both of its sides – reminds the “minor Roman dyke” in Bačka. Despite the cross-section cuts made at diff erent parts of the earthwork they did not succeed in determining its age. In one case they detected a pit with shards from Sarmatian or Árpádian Age. Th e pit was suggested to be younger than the ditch.13 Despite of the uncertainty of the dating, the earthwork was determined as a road of Imperial Age.14

6 Cf. map of Bács-Bodrog County made by Pál Gönczy (htt p://szilagyinfo.googlepages.com/Bacs-Bodrog_varmegye_terkepe.jpg/Bacs-Bodrog_varmegye_terkepe-full.jpg); Rómer 1876, map between pages 42 and 43; Frölich 1887, 21; Gubitza w.y. 11–13.

7 Considering size and structure data see: Frölich 1887, 22–24, 28–29, 210, 307. 8 At the start of the research, this territory was part of Hungary. Th at is why the names of places appear here in several

variants, according to the fi rst records. We also refer to the modern name of the sett lements in question.9 For the size and structure data see Frölich 1887, 22–24, 28–29, 210, 307.10 Gubitza w.y. 11–13 – mainly based on Gyula Cziráky’s research.11 On the size and structure data see: Frölich 1887, 133, 138.12 Gallina 1999; Patay 2005; Patay 2006 – with earlier references.13 Patay 2005, 399.14 Patay 2005.

Actual State of Research of the Csörsz- or Devil’s Dyke | 313

Th e dyke section observed SW from the Körös mouth, in Cumania Minor is SW directed. On the basis of aerial photos, maps and fi eld surveys a line between Szank – Kiskunhalas – Tompa – Csikéria – Bajmok can be reconstructed. In all probability it ran into the line recorded at the track Svetozar Miletić – Sombor (Zombor). Th e SW oriented earthwork in Hungary could be hardly traced at the end of the 1970s; it was preserved in a much worse state than the ones in Bačka or the Csörsz Ditch.15

So, the SW oriented earthwork in Cumania Minor, comes to an end long before it would reach the Tisza, 30–35 km east of it. Th e vallum-fossa running at the northern side of the Körös follows its line, so it can be hardly doubted that these two sections are related. Th e latt er section joins the Csörsz Ditch.16

In the Carpathian Basin we know several other earthworks. Among them on the fi rst place we should mention the work of recent years done by Alexandru Matei and Róbert Gindele, who made a series of cross-section cuts at vallum and fossa running some 50 km from the Porolissum part of the Dacian limes, at Supuru de Sus. According to A. Matei’s opinion, ditches crossing the valleys of rivers Crasna and Zalău obviously refer to Roman building activities. Th e defense system badly damaged by agricultural activity is 2–2.5 m deep and 3–4.5 m wide.17

Th e “minor Roman dyke” described above was suggested to continue in Transdanubia (west of the Danube). We know further vallum sections from Transdanubia, e.g. from counties Tolna, Vas or Somogy.18

We can also mention such fortifi cations from South Slovakia and Transylvania.19 But there are also earthworks – among them the one called Devil’s Dyke (so-called Brazda lui Novac20) – at the Lower Danube (about 300 km long), and at the territory between Dniester and Prut, and more to the south the so-called Erkesija vallum between the Black Sea and Marica (the earthwork in Bessarabia and the Eskerija make almost 140 km).21

According to our present knowledge, we cannot prove either that the earthworks described above – among them the so-called longitudinal ditches – were erected according to one concept, or that they are at least more or less of the same age. Several scholars underlined their common feature: they border plain territories. At the same time, in each case we have to examine one by one when they were made and what was their function. Th ere are three possible explanations to the latt er question:

1) Dams against fl oods.2) Roads.22

3) Marking and/or defending a border.If we deal with chronology, we should take into consideration the excavations related to the dykes,

the system of their structure, the territory crossed or bordered by the dykes and historical data of the actual period. From our point of view, the question is whether there are structures erected in the Imperial Age among the dykes. Roman Age dating is generally spread; this is an old conventional opinion. If this dating is acceptable, then we have to answer the question: was it constructed and planned by Romans or Barbarians? Already at the beginning, we have to emphasize an important counter-argument against the role of Roman engineers, pointed out already by Róbert Frölich. He underlined that along a Roman limes there should have been traces of castra, castella or burgi, not like in the case of the dykes in question.23 Along the dykes of the Barbaricum no such edifi ces were found.24

15 Rómer 1876, map between pages 42 and 43; Soproni 2003, 59, footnote 119. Recently on the Cumanian dyke, with earlier references: Gallina 1999; Patay 2006.

16 Gallina 1999, 74; Patay 2006, fi g. 1.17 Matei–Gindele 2004. According to the authors’ opinion, this section of the limes can be connected with the Körös

Valley dyke. Opposite of Zsolt Gallina, they situate the ditch north of the dyke, in which case the dyke protected the territory from the north. According to their hypothesis, this could be the border of Decebal’s land, and that was the territory where Trajan formed the earliest limes later surrendered by the Romans. We have to notice that in the course of the research between 1977–79 it was possible to reconstruct a section between Szank–Kiskunhalas–Tompa–Csikéria–Bajmok. Sándor Soproni noted that its earlier dating to 358 could not be proved. An important argument for the dating is that on the basis of a cross-section cut made at the crossing point of the Csörsz Ditch and the Körös Valley dyke, the latt er is younger (Soproni 1978. 118).

18 Rómer 1876, map between pages 42 and 43, Frölich 1887, 22; Gallina 1999, 69.19 See the series of articles by István Dénes (Székelyföldi törésvonalak) in Acta Siculica 2001, 2004, 2006 and 2007.20 Th is one starts from Ploieşti and runs as far as river Ialomiţa, from here it can be traced in the direction of Siret.21 Recent summary of these dykes: Fiedler 1986 – with earlier references.22 Recently Mihály Nagy again suggested that in the Hungarian Plain missing W-E oriented routes, moves in N–S direction

were possible on the dykes of the Csörsz both for Romans and Barbarians (Nagy 1997, 156.).23 Frölich 1887, 211.24 In the case of the Bačka earthworks, descriptions mention that the dykes could have connected hillforts. However, no

serious research has been conducted on the question since Gyula Cziráky who, based on his excavation results, dated these hillforts to the Late Migration Period. (Gubitza w.y. 12–13)


Let’s now see what can be said on the dykes of the Great Hungarian Plain.

Csörsz- or Devil’s DykeAs we have already pointed out, most of scholars agree that the Csörsz can be reliably dated to the Roman

Age. Th is is the case also with the rest of dykes described above, because researchers thought that out of all peoples of the region, only Romans were capable of organizing works of such scale.25 In this connection we again should refer to the absence of watch-towers and military camps at the dykes, that should make us cautious.

Th e most fundamental and methodological research was conducted on the longitudinal or Devil’s dykes. Based on the topographical and historical sources collected by Vilmos Balás, a work-group organized by Pál Patay in 1962 has conducted a methodological fi eld survey of the dyke and made several cross-section trenches.26 Th e main starting points in dating are always the archaeological excavations. Th e terminus post quem of the existence of the vallum-fossa system are determined by features in the fi lling of which there were shards of sigillata, of painted Roman vessels, fi bulae. Judging from them, building activities must have been started obviously later than the mid–3rd century.27

Th e ante quem is the 11th century: at Oszlár graves dated with the coin of Hungarian king Salamon were dug into the vallum.28 Th e 11th century closing date is supported by a writt en source, the already mentioned deed of foundation of the Százd Abbey mentioning the Csörsz Ditch.29

Two cuts of the Csörsz were made at its Maros/Mureș section in 1966, in the vicinity of Sîmbăteni and Covăsniţ.30 At both places a badly demolished section was examined. In the latt er case, despite of bad preser-vation, remains of 5 ditches could be well traced. Under the widest dyke situated between them (marked D) a continuous layer of shards was found and dated by K. Horedt to the 1st century A.D. Th is feature can be also interpreted as terminus post quem on this section of the Devil’s Dyke.

In connection with archaeological data the following question arises: can we consider the system of dykes a homogeneous work built in a single period? It would be diffi cult to doubt that the vallum-fossa system called Csörsz Ditch – or at least its main line – was constructed on the basis of a general concept. However, it can be strongly discussed what is the chronological relationship between the double, sometimes triple lines of the dyke. It is not clear either, what kind of chronological diff erence is shown by the sections of the dyke crossing each other. It is an additional diffi culty that only certain sections can be measured and mapped in our days. In several places only old topographical data, sometimes aerial photos and excavations provide data for the outlining this or that section of the dyke.31 For example, in the case of the Nyírség part, it was possible to reconstruct an exterior line on the basis of local names and topographic data.32

Back to the chronological data, the main starting point of the dating is a theoretical idea: scholars were looking for a people and/or culture between the 3rd and 11th century, that could be bordered by the Csörsz Ditch. Th ey took into consideration three archaeological cultures of the Hungarian Plain that characterized more or less the territory in question: Sarmatian, Gepidic and Early Avarian. Th e most accepted dating is the one concerning the Roman Age. It was a scholar consensus that before the surrender of province Dacia the construction of the dyke would have been senseless from strategic point of view, either from Roman, or Sarmatian part. Aft er the Gothic invasion of Transylvania it could be in accordance with the interests of both the Empire and the inhabitants of the Great Hungarian Plain.33 Of course, in the case of Roman participation it is also a question on what degree the territory of Barbarians living by the Roman border was sovereign – at least in the eyes of the Romans.34 As to the period following 271, there were several ideas considering this time. Researchers joined mainly two hypothesis:35

25 Fiedler 1986, 458 – with further reference.26 Balás 1961; Balás 1963; Garam–Patay–Soproni 1983; Garam–Patay–Soproni 2003.27 Garam–Patay–Soproni 2003, 49–56. 28 Soproni 1978, 113, footnote 9; Garam–Patay–Soproni 1983, 52; Garam–Patay–Soproni 2003, 53.29 Balás 1961, 331. Th e diploma comes from the time of king Béla IV, but the original deed of foundation dates to 1067,

so this is the year we should consider as the fi rst writt en datum of the Csörsz Ditch. 30 Detailed report and interpretation: Horedt 1968.31 Such is the section of the Csörsz situated at Dunakeszi – today not visible – that was traced during preventive excavations

of Road 2/a: Istvánovits–Kulcsár 2002.32 Istvánovits–Kulcsár 2000, 248.33 Soproni 1978, 115.34 Vö. Mráv 1999, 92–93; Kiss 2008, 6–7.35 Mócsy–Fitz 1990, 47; Soproni 1978, 113–137; Garam–Patay–Soproni 2003, 57–70; Mócsy 1975, 100–101. Summary

of chronological arguments, with further references: Tóth 1983, 68–69.


1) András Mócsy: the dyke was constructed around 294, in the course of the fortifi cation works by Galerius and Diocletian.36 Th e Achilles heel of this theory is that the defense dyke could not been constructed without the knowledge and agreement of Sarmatians, more than that, they had to take an active part in this work. However, it was Diocletian’s time during which hostile Sarmatians att acks followed each other.

2) Sándor Soproni: the dyke was constructed in 322 in connection with Roman military campaign in the Hungarian Plain. Following him, several other scholars thought that the period between 324 and 332 was suitable for the construction of the Csörsz Ditch.37 Aft er 332 this was hardly possible, because then the long lasting Sarmatian civil broke out. In the political situation formed at the second half of the 4th century the dyke would have run at another track. By that time its route would have lose its sense.

Both András Mócsy and Sándor Soproni tried to connect the large scale Roman construction works on the Barbarian part of the Danube Valley with the erection of the Csörsz Ditch. Th at is to say, they inter-preted the Roman edifi ces on the left bank of the Danube and the Csörsz as parts of the same defensive system concept.38 Further research – on the fi rst hand the one by Barnabás Lőrincz – has shown that counter-fortifi cations/bridgeheads of controversial function, situated on the left bank39 can be well dated on the basis of brick-stamps. Judging from them, the earliest edifi ces can be dated aft er the mid-350s (Constantius II). Th e majority of the bricks were dated to the 370s, that is to say, to the time of Valentinian’s building projects.40 So, the erection of the counter-fortifi cations cannot serve to support theories of Mócsy or Soproni.

Apart from the date of erection, most of the scholars agreed that the Csörsz Ditch was the advanced defensive system of the Empire formed as a protection against the Germanic danger. It would have been initiated and directed by Romans on Sarmatian land. András Mócsy was correct to point out that the dyke should have been in use for a long time, because its line had been changed for several occasions.41 However, as the time of surrender scholars suggest Valentinian’s reign42 or the defeat at Adrianople in 378.43 Th is makes not more than 2–2.5 generations’ time even in the case of the earliest dating.

It is a characteristic feature of the dating problem that Jenő Fitz in connection with Soproni’s hypothesis concluded that his dating was too early, and suggested the reign of Constantius II and Valentinian I.

In the course of the large scale preventive excavations of the recent decades the track of the Csörsz Ditch was cut at several points. Most of them are still unpublished. Research was rather uneven at this problem, because several archaeologists did not clean the ditch for diff erent reasons. Among them the site of Mezőszemere–Kismari-fenék can be mentioned, where a section of the Csörsz was traced but not cleaned “for fi nancial reasons”.44 Th is is quite regrett able, because in the case of the neighboring cemetery part, the publi-shers of the site formed a determined opinion according to which this burial ground was used by Sarmatians who, in Roman alliance, were guarding the Csörsz.45 Th is hypothesis was based on the numerous Roman fi nds (among them weapons) of the cemetery and on the closeness of the Csörsz Ditch. If we agree with this idea, than it would contradict the datings accepted before, the ones by András Mócsy and Sándor Soproni. Th e reason for this is that the cemetery cannot be dated earlier than the second half of the 4th century. Andrea Vaday solved this controversy by the theory according to which the dyke was erected aft er 358,46 thus joining the dating by Fitz mentioned above. Taking into consideration that we have no information on any serious fi ghts between the Barbarians of the Hungarian Plain and of the Romans in the period in question, this suggestion can be seriously considered. It is also supported by the fact that during the reign of Valentinian large fortifi cation works went on in the Danube region.47 Aft er the events around the death of Valentinian – a

36 Mócsy 1972. 95–96. In that time two camps were built at the territory of the Barbaricum opposite of Bononia and Aquincum, that was connected with the erection of Csörsz Ditch by A. Mócsy. Th e summary of references: Tóth 1983, 69, footnotes 5–6 and 11–12.

37 Soproni 1978, 113–137. Summary of literature: Tóth 1983, 69, footnotes 7–10. 38 Soproni 1978, 117.39 Th ey were mostly considered to be fortifi cations built for keeping the Sarmatians off the border and for demonstrating

power (Bertók 1997), but recently M. Nagy suggested that they functioned as storage rooms (Nagy 1999, 121–122). 40 Lőrincz 1999. Among them the recently excavated fort at Dunakeszi (Mráv 2002).41 Mócsy 1975, 101.42 Mesterházy 1990, 62–63; Mesterházy 1994, 283–284.43 Visy 1989, 23.44 Domboróczki 1997, 41.45 Vaday–Domboróczki 2001, especially: 111, 116–122, 204–206, see also: Vaday 2002; Vaday 2003. Cf. Prohászka

2003. 84. 46 Vaday 2003, 212.47 Among them, beside the above mentioned counter-fortifi cations, the edifi ce at Hatvan-Gombospuszta (Mráv 2003a).


new Quadian-Sarmatian att ack against Pannonia – Sarmatians lost their signifi cance. Soon aft er the Great Hungarian Plain was invaded by the Huns.

Among the few excavated sites situated at the Csörsz Ditch we can mention the vicinities of Csincse. In 1994 Klára Fischl cut the dyke with several trenches. Her description is not explicit. No fi nds supporting the Roman Age dating came to light. At the section where the cuts were made, smaller ditches followed the vallum at a 200 m long track. From the publication we get controversial information, whether the earthwork and the small ditches were synchronous or not. Fischl suggested that in the Avarian Age the earlier dyke was strengthened and aft er that the vallum got to the opposite side of the ditch. Th at means that in the 9th century on the track between Csincse and Gelej the vallum was situated at the northern side and the system of ditches supplemented by smaller ditches ran on the southern side. Th at is to say, the dyke was used by people living north of it to defend themselves from an enemy arriving from the south; no traces of the Roman Age vallum remained by this time.48 In this case the problem of dating has not been resolved and no new information on the Roman Age was gathered.

Relatively recently Zsolt Mráv, interpreting the history of the unfi nished Roman military camp at Göd concluded from archaeological data (topography and chronology of Quadian sett lements on the territory between the dykes of the Csörsz) that the inner line of the earthwork system was erected under Constantine the Great as earliest, and under Constantius II as latest, somewhere at the Quadian-Sarmatian border.49 In this case the exterior, further two lines of the Csörsz would have been constructed on Quadian territory a period later, as Mráv noted, that is to say in the time of Valentinian. According to him, the Romans advanced the line of the Csörsz in order to mark the Sarmatian border. We can hardly regard this idea. It would be problematic to mark the border between Sarmatians and Quadians, taking into consideration that for centuries they had joined ventures. Writt en sources emphasize their alliance, brotherhood in arms and that they had mutual understanding because of being neighbors, their similar customs and weaponry.50 We can also add the infor-mation deriving from a peace treaty made during Constantius II’s reign: the leading role of the Quadians (Araharius) in this region is clear on the disadvantage of Sarmatians (Usafer).51 Considering this, it is hard to believe that Quadians would peacefully tolerate the advance of the dyke on their territory, while a pure plan of deploying a camp led to a military confl ict. Th e next question arising in this case, what was the sense of constructing such a vallum-fossa system hardly supported by any of the Barbarians. Th is is so, even if we know from the Notitia Dignitatum that in 365 as the latest there was a Praefectus legionis, Transiacinco (occ. XXXIII.65.) and judging from this somewhere opposite of Aquincum a territory under Roman military administration existed.52

In connection with Zsolt Mráv’s hypothesis the following problem is to be considered: exterior dykes (at some points two and even three ones) at further parts of the Csörsz would then also come from Valentinian’s time. In this case Romans tried to advance the complete border of the Sarmatian land. Keeping in mind the strengthening Germanic pressure, it would be hard to imagine this.

Now, if we examine what kind of serious and relevant arguments were published by scholars supporting the Roman Age dating of the Csörsz Ditch, we’ll fi nd the following:

Th e basis of the Roman Age dating is the idea according to which such a large scale project needed a well organized state. Th e only one that could be thought of between the 3rd and 11th century, is the Roman Empire.53 We should note that the territory populated by Sarmatians spread out of the line of the Csörsz, e.g. in the valley of river Ier (Ér) it reached Carei, while the territory occupied by the population of the Early Avarian sites almost exactly suit the line of the Csörsz Ditch.54

Aft er all, it is not surprising that serious doubts have arisen among West European scholars consi-dering the accuracy of the dating, so they emphasize this dating to be hypothetical.55

Future research should take into consideration also folklore data on the Csörsz Ditch, not forgett ing the tale mentioned in the already mentioned 16th century chronicle by István Székely. István Borzsák has

48 Fischl 1995, 34, 36.49 Mráv 1999, 80–101; Mráv 2003; Mráv 2005.50 Tacit. Ann. XII.30, Dio LXVII.5, Zos. I.48, Ammian. XVII.12.1, XVII.12.12. Th e data have been collected by Zsolt Mráv

who also mentioned that archaeological fi nds are so mixed on this territory, that the border between the two peoples can be hardly determined (Mráv 1999. 95–96).

51 Ammian. XVII.12.52 Mráv 1999, 99–100.53 On the diffi culties of the dating and uncertainties see: Fiedler 1986, 459–460.54 ADAM Karte 3.55 Ditt rich 1987, 25–27.


drawn att ention to the similarity of the tales cycles with the legends connected with the monarchs of the Ancient East famous for their channel constructing activity as e.g. Semiramis.56 In this connection it is to be examined whether the fi gure of Csörsz could be connected to some historic personality related to the image of the “grand builder”. It would be interesting to make a deeper study of the tale version according to which Langobardian king Rád invited Avarian prince Csörsz to be an ally, who, instead of war booty asked for the beautiful Délibáb (Mirage), the daughter of Rád. However, the shrewd Rád would give his daughter to the prince only if he took home his bride at a waterway. Csörsz commanded his people to dig a ditch, but one night he was killed by a thunder, so the bride, being a mirage, fi lled the dry ditch only with the shadow of the water. We could hardly mention any other Hungarian tale that is about a Langobardian and Avarian king.

In our opinion the question about the time of construction of the Csörsz Ditch and its function can be solved only aft er the evaluation of the recent excavations and new fi eld research.

The so-called „Roman dykes” in Bačka

Dating of the dykes of Bačka to the Roman Age originates from the description and map by Ferdinand Marsigli. He marked castellum in four cases at the eastern side of the dyke. However, Róbert Frölich deter-mined that these phenomena did not belong to edifi ces of castellum type. Th ey are open from the side of the dyke; the ditch was traced on their interior side. Th eir long shape (parallel with the dyke the earthwork was 240 m long, the width was under 50 m) these features had some other function.57

Th ough Frölich did not consider Marsigli’s dating to be relevant, he accepted the Roman Age dating. He examined the function of the dyke from this aspect. He excluded the possibility of its being a dam, because it was not “nivellated”. Because of the winding character of the earthwork Frölich did not believe that it could serve as a road, especially because it did not cross the deep, watery parts, and the width of its top was not even. Finally, from topographic point of view, its line was not reasonable.58

We think that Frölich’s arguments were not utt erly convincing. Th e fact of two ditches on both sides of the vallum supports mainly the idea of the road function. Th ere could have been bridges at the points where the dyke broke at deeper places. From topographic point of view we cannot forget in the case of the „minor dyke” that Apatin, which is one of the fi nal points of the earthwork, has a special strategic importance at the Danube: it is situated above a marshy area opposite of the Dráva mouth. Th is line of the dyke runs up to Čurug that, judging from its topographic situation also seems to be a „good choice”: marshy washland at the point where Tisza gets narrow and could provide a suitable crossing to the other side of the Tisza. Th e N-S line of the „minor dyke” in the north ends at a place of similar topography at Petrovo Selo. Th e „major dyke” runs along the line „cutt ing” the marshy mouth region of the Tisza, connecting the two rivers. As we could see, at Čurug it ends at a very suitable part. As to Novi Sad, its key importance is well known.

Th e dating is problematic. Zsolt Gallina pointed out that if the dyke did continued in Transdanubia, then the Roman Age dating cannot be supported.59 Th is is true only in the case, if we consider the dyke to be a defensive fortifi cation, a frontier line. But if it served as a road, it could easily continue on the right bank of the Danube.

If these dykes used to be roads and were really constructed in the Roman Age, then they could provide outstandingly good means of marching through the Barbaricum for the Roman troops that crossed Sarmatian territory on several occasions. However, all these are vague hypotheses, like Róbert Frölich’s witt y idea according to which the dyke was constructed by the Romans interfering into the fi ght between the Arcaragantes and Limigantes. In this case Ammianus would have referred to the „major Roman dyke” when saying that Constantius „...pitched his camp near Acimincum, where a loft y mound was raised to serve for a tribune; and some boats, loaded with soldiers of the legions, without their baggage, under command of Innocentius, an engineer who had suggested the measure, were sent to watch the channel of the river, keeping close under the bank; so that, if they perceived the barbarians in disorder, they might come upon them and surprise their rear, while their att ention was directed elsewhere.”60 In this case the interpretation of the source speaks against the argumen-tation of the hypothesis, namely the phrase according to which Barbarians asked the emperor „to allow them to cross the river and come to him”. If, aft er crossing the river they met the emperor where the mound was

56 Borzsák 1996, 286–303.57 Frölich 1887, 208–209.58 Frölich 1887, 306–307.59 Gallina 1999, 70.60 Ammian. XIX.11, Frölich 1887, 309–310.


raised, it would mean that Ammianus located their place of habitation to the right bank of the Danube, which is hardly probable.

We should mention one further hypothesis. Uwe Fiedler suggested that the „minor Roman dyke” can be connected with the earthworks of the Lower Danube, and in this case these dykes could have been erected during the wars fought by Bulgarians against Avarians.61

We can assume that similarly to the problem of the Csörsz Ditch, these questions can be solved by further thorough fi eld works and excavations.

Dykes in Cumania Minor, between the Danube and Tisza and in the Körös Valley

Th e dating of the Körös Valley ditch is even more uncertain than the case of the Csörsz. First, Sándor Soproni connected it with Gepidians and, based on historical aspects, dated it around 358.62 Later he modifi ed this opinion and suggested a dating later than the 4th century.63

Zsolt Gallina, who recently summarized the research of the Cumanian ditch, accepted the dating to Roman Age judging from the fact that its line connects to the track of the Csörsz Ditch embracing the Hungarian Plain. In this case it could be erected between 271 (surrender of Dacia) and 378 (batt le of Adrianople). At the same time he associated the erection of the dyke with the defense of the trade route between Lugio and Csongrád (this is one of the suggested routes functioning between the 2nd and 4th cc.).64 However, it is diffi cult to interpret this idea from historical point of view. It is not clear who was the enemy against which the northern part of the Hungarian Plain must have been protected; who could att ack from the south. We should not forget that on the territory between the dyke and the Roman province we fi nd Sarmatian material, the same we have north of the dyke. As an evidence to show that even at the end of the 4th century Sarmatians populated this region, it is enough to refer to Ammianus Marcellinus who wrote in connection with the events of 374/375 that „Th eodosius the younger, Duke of Moesia, ... in many encounters defeated and vanquished the Free Sarmatians (so called to distinguish them fr om their rebellious slaves), who had invaded our fr ontier on the other side...”65 Th is information att ests to the conclusion made on the basis of archaeological fi nds: the region neighboring Moesia was in the hands of Sarmatians (more concretely: the Arcaragantes). Diff erent groups of Sarmatians – even if know that in this period there was a civil war on their territory – hardly could erect the ditch. And even if they did, it could not be the part of the same system with the Csörsz Ditch, as it was suggested by Gallina.

So, we can assume that despite of its several hundred years old history, the research of the dykes of the Great Hungarian Plain is at its very beginnings. It is not surprising that Alexandru Matei turned to this topic with a great interest. In connection with the excavations in Supuru de Sus he tried to gather as much infor-mation as possible on the Csörsz Ditch (especially on its W–E section) and on the Körös Valley dykes. His early death is a great loss also for the international research of the dykes...

References

ADAM Archäologische Denkmäler der Awarenzeit in Miteleuropa. Hrsg. József Szentpéteri. Varia Archaeologica Hungarica XIII: 1–2. 2002.

Balás 1961 Balás, Vilmos: Az alföldi hosszanti sáncok. [Die Längswälle der Tiefebene.] Régészati Füzetek II. 9, Budapest 1961.

Balás 1963 Balás, Vilmos: Die Erdwälle der Ungarischen Tiefebene. Acta Archaeologica Academiae Scientiarum Hungaricae, Budapest 15. 1963. 309–336.

Bertók 1997 Bertók, Gábor: Ripa Sarmatica: Late Roman counterfortifi cations on the left bank of the Danube. In: Roman Frontier Studies 1995. Proceedings of the XVIth International Congress of Roman Frontier Studies. Eds. W. Groenman-van Waateringe – B. L. van Beek – W. J. H. Willems – S. L. Wynia. Oxbow Monograph 91. Exeter 1997. 167–166.

Borzsák 1996 Borzsák, István: Dragma 2. Válogatott tanulmányok. Telosz Kiadó, Budapest 1996.Ditt rich 1987 Ditt rich, Ursula-Barbara: Die Wirtschaft sstruktur der Quaden, Markomannnen und Sarmaten

im mitt lere Donauraum und ihre Handelsbeziehungen mit Rom. Münstersche Beiträge z. antiken Handelsgeschichte VI: 1. 1987. 9–30.

61 Fiedler 1986, 461–463.62 Soproni 1978, 118.63 Garam–Patay–Soproni 2003, 61.64 Gallina 1999, 74, 69, 78–80.65 Ammian. XXIX.6.


Domboróczki 1997 Domboróczki, László: Mezőszemere–Kismari-fenék. (Heves megye). Régészeti Füzetek I: 51. 1998. Archaeological Reports 1997. Magyar Nemzeti Múzeum, Budapest 2001, Nr. 33, 41–42.

Fiedler 1986 Fiedler, Uwe: Zur Datierung der Langwälle an der mitt leren und unteren Donau. Archäologisches Korrespondenzblatt , Mainz 16, 1986, 457–465.

Fischl 1995 Fischl, Klára: Előzetes jelentés a Csörsz-árok kutatásáról Csincsén. [Vorbericht über die Forschung des Csörsz Grabens bei Csincse. Ďđĺäâŕđčňĺëüíűé đĺôĺđŕň îá čńńëĺäîâŕíč˙ő Ęŕíŕâű-×¸đńŕ â ×čí÷ĺ.] Somogyi Múzeumok Közleményei, Kaposvár XI, 1995, 33–46.

Frölich 1887 Frölich, Róbert: A bácskai u. n. római sánczok. Archaeologiai Értesítő, Budapest VII, 1887. 19–30, 132–138, 207–23, 304–310.

Gallina 1999 Gallina, Zsolt: A kiskunsági Ördög-árok. In: Halasi Múzeum. Emlékkönyv a Th orma János Múzeum 125. évfordulójára. Szerk. Szakál Aurél. Th orma János Múzeum – Halasi Múzeum Alapítvány, Kiskunhalas 1999, 67–82.

Garam–Patay–Soproni 1983 Éva Garam – Pál Patay – Sándor Soproni: Sarmatisches Wallsystem im Karpatenbecken. RF II. 23, 1983.

Garam–Patay–Soproni 2003 Éva Garam – Pál Patay – Sándor Soproni: Sarmatisches Wallsystem im Karpatenbecken. RF II. 232, 2003.

Gubitza w.y. Gubitza Kálmán: Bács-Bodrog vármegye őskora. In: Bács-Bodrog vármegye II. Magyarország Vármegyéi és Városai. Szerk. Borovszky Samu. Budapest without year, 1–27.

Horedt 1968 Horedt, Kurt: Zur Frage der Datierung der grossen Erdwälle im Karpatenbecken. Kwartalnik Historii Kultury Materialnej XVI: 1. Warszawa 1968, 39–54.

Istvánovits–Kulcsár 2000 Iranian-Germanic contacts in the Sarmatian Barbaricum of the Carpathian Basin. In: Die spätrömische Kaiserzeit und die frühe Völkerwanderungszeit in Mitt el- und Osteuropa. Hrsg. Magdalena Mączyńska – Tadeusz Grabarczyk. Łódź 2000, 237–260.

Istvánovits–Kulcsár 2002 Istvánovits, Eszter – Kulcsár, Valéria: Th e history and perpectives of the research of the Csörsz Ditch (‘Limes Sarmatiae’). BAR S 1084, 2002. Limes XVIII – Proceedings of the XVIIIth International Congress of Roman Frontier Studies held in Amman, Jordan (September 2000). A conference held under the auspices of the Department of Antiquites of the Hashemite Kingdom of Jordan, Th e Council for British Research in the Levant and the Department of Archaeology at the University of Liverpool. Edited by Philip Freeman, Julian Bennett , Zbigniew T. Fiema and Birgitt a Hoff man, volume II, 625–628.

Kiss 2008 Kiss Magdolna: Gót vezéregyéniségek a késő Római Birodalomban. Getica-kutatások. Vivarivm fontivm 2. GeniaNet Könyvkiadó, Pécs 2008.

Lőrincz 1999 Lőrincz Barnabás: A későrómai hídfőállások bélyeges téglái Valeriában. / Die Ziegelstempel der spätrömischen Brückenkopff estungen in der Provinz Valeria. In: Pannoniai kutatások. A Soproni Sándor emlékkonferencia előadásai (Bölcske, 1998. október 7.) Szerk. Gaál Att ila. Wosinsky Mór Múzeum, Szekszárd 1999, 53–68.

Matei–Gindele 2004 Fortifi catia romana de pamant de tip burgus, atasata valului şi şantului roman descoperit la supurul de Sus, Jud. Satu mare = Th e Earth burgus-type fortress, att ached to the Roman ditch and vallum, descovered at supuru de sus “Dealul soarecelui/mouse hill” (Satu Mare County). Acta Musei Porolissensis, Zalau 26. 2004, 283–307.

Mesterházy 1990 Mesterházy Károly: Münzdatierter spätkaiserzeitlicher Gerätfund aus Hajdúnánás-Tedej. Alba Regia. Az István Király Múzeum Évkönyve 24. 1986–1988. (1990) 53–66.

Mesterházy 1994 Mesterházy Károly: Pénzekkel keltezett későcsászárkori eszközlelet Hajdúnánás–Tedejről. [Ein mit Münzen datierter spätkaiserzeitlicher Werkzeugsfund aus Hajdúnánás–Tedej.] Néprajzi tanulmányok Ikvai Nándor emlékére I. Studia Comitatensia, Szentendre 23. 1994, 273–292.

Mócsy 1972 Mócsy, András: Das letzte Jahrhundert der römisch-barbarischen Nachbarschaft im Gebiete des heutigen Ungarn. A római-barbár szomszédság utolsó évszázada hazánk területén. Cumania I. 1972, 83–102.

Mócsy 1975 Mócsy, András: Pannonia a késői császárkorban. Apollo Könyvtár 4. Akadémiai Kiadó, Budapest 1975.

Mócsy–Fitz 1990 Mócsy, András – Fitz, Jenő: Eseménytörténet. In: Pannonia régészeti kézikönyve. Szerk. Mócsy András – Fitz Jenő. Akadémiai Kiadó, Budapest 1990, 31–51.

Mráv 1999 Mráv Zsolt: „Valentinianus ... in ipsis quadorum terris quasi romano iuri iam vindicatis aedifi cari praesidaria castra mandavit.” (Amm.Marc., XXIX 6,2) I. Valentinianus kvád külpolitikája egy vitatott Ammianus Marcellinus szöveghely tükrében. / „Valentinianus ... in ipsis quadorum terris quasi romano iuri iam vindicatis aedifi cari praesidaria castra mandavit.” (Amm.Marc., XXIX 6,2) Die quadische Aussenpolitik Valentinians I. im Spiegel einer stritt igen Textstelle bei Ammianus Marcellinus. Szerk.Gaál Att ila. In: Pannoniai kutatások. A Soproni Sándor emlékkonferencia előadásai. Bölcske, 1998. Wosinszky Mór Múzeum, Szekszárd 1999, 77–111.


Mráv 2002 Mráv Zsolt: Dunakeszi, Duna sor 28. Régészeti kutatások Magyarországon [Archaeological Investigations in Hungary], Budapest 2002, (2004), 205–206.

Mráv 2003 Mráv, Zsolt: Archäologische Forschungen 2000–2001 im Gebiet der spätrömischen Festung von Göd–Bócsaújtelep (Vorbericht). Communicationes Archaeologicae Hungariae, Budapest 2003. 83–114.

Mráv 2003a. Mráv, Zsolt: Hatvan–Gombospuszta Fortlet. In: Th e Roman Army in Pannonia. An Archaeological Guide of the Ripa Pannonica. Ed. Zsolt Visy. Without place of edition, 2003, 207–209.

Mráv 2005 Mráv, Zsolt: Egy erőd, amely sohasem épült fel. Régészeti kutatások Göd–Bócsaújtelepen, a késő római erőd területén 2000–2001 (Előzetes jelentés). [A never-fi nished late Roman fortress. Archaeological excavations in the territory of the fortress at Göd–Bócsaújtelep in 2000–2001 (Preliminary report).] Szántó Kovács Múzeum Évkönyve, Orosháza 7, 2005, 291–332.

Nagy 1997 Nagy, Mihály: Routes in the Carpathian Basin and the Interaction between Romans and Barbarians. [Cesty v Karpatské kotlinĕ a vzájemné vlivy mezi Římany a barbary.] In: Peregrinatio Gothica. Acta Musei Moraviae. Scientiae sociales LXXXII, 1997, 151–162.

Nagy 1999 Nagy Mihály: A pannoniai IV. századi burgus-típusok méretei. / Dimesnsions of 4th century A.D. burgus-types in Pannonia. In: In: Pannoniai kutatások. A Soproni Sándor emlékkonferencia előadásai (Bölcske, 1998, október 7.) Szerk. Gaál Att ila. Wosinsky Mór Múzeum, Szekszárd 1999, 113–140.

Patay 1969 Patay, Pál: Neuere Ergebnisse in der topographischen Untersuchung der Erdwälle in der Tiefebene. Móra Ferenc Múzeum Évkönyve, Szeged 1969: 2, 105–112.

Patay 2005 Patay, Pál: Római út Bácskában? [Römerszrasse in der Batschka?] Communicationes Archaeologicae Hungariae, Budapest 2005, 393–406.

Patay 2006 Patay, Pál: A Bácska–Kiskunság–Körösmenti ördög árka. [Der Ördög árka (Teufelsgraben) in der Batschka, in Kleinkumanien und an der Körös.] Communicationes Archaeologicae Hungariae, Budapest 2006, 107–122.

Prohászka 2003 Prohászka, Péter: Angaben zur Archäologie und Geschichte des Wallsystems in der grossen ungarischen Tiefebene. (Forschungsgeschichter Überblick der Untersuchungen und Ergebnisse aus den vergangenen Zwanzig Jahren.) In: Garam–Patay–Soproni 2003, 79–89.

Pulszky 1891 Pulszky, Ferencz: Régészeti emlékek az Alföldön. In: Az Osztrák-Magyar Monarchia írásban és képben. VII. kötet. Magyarország II. kötete. A Magyar Királyi Államnyomda kiadása, Budapest, 1891, Arcanum DVD

Rómer 1876 Romer, Florian: Résultats généraux du mouvement archéologique en Hongrie. Compte-rendu de la huitème session à Budapest 1876, Musée National Hongrois, Budapest 1878.

Soproni 1978 Sándor Soproni: Der spätrömische Limes zwischen Esztergom und Szenetndre. Akadémiai Kiadó, Budapest 1978.

Székely 1854 Székely István magyar krónikája, 1558. In: Tizenhatodik századbeli magyar történetírók régi kiadások és kéziratok után. Szerk. Toldy Ferenc. Újabb Nemzeti Könyvtár 3. Emich Gusztáv könyvnyomdája, Pest 1854, 2–66.

Tóth 1983 Tóth, Endre: Contra Acinco et Bononia. Arheološki Vestnik. Acta Arcaeologica XXXIII. 1982. Lujbjana 1983, 68–78.

Vaday 2002 Andrea Vaday: Militia inermis, militia armata, Bemerkungen zur Frage des Limes Sarmatiae. In: Zwischen Rom und dem Barbaricum. Festschrift für Titus Kolník zum 70. Geburtstag. Hrsg. Klára Kuzmová – Karol Pieta – Ján Rajtár. Nitra 2002, 201–202.

Vaday 2003 Andrea Vaday: Mezőszemere sett lement. In: Th e Roman Army in Pannonia. An Archaeological Guide of the Ripa Pannonica. Ed. Zsolt Visy. Teleki László Foundation, without place of edition 2003, 209–212.

Vaday–Domboróczki 2001 Vaday, Andrea – Domboróczki, László: Mezőszemere, Kismari-fenék. Spätkaiser-frühvölkerwanderungszeitliches Gräberfeldsdetail. (Mezőszemere, Kismari-fenék. Késő császárkori – kora népvándorláskori temetőrészlet. Mezőszemere, Kismari-fenék. Late Imperial – Early Migration Period Cemetery fragment.) Agria XXXVII, 2001, 5–206.

Visy 1989 Visy Zsolt: A római limes Magyarországon. Corvina, Budapest 1989.


Map of the dykes men oned in the text

1: Csörsz Ditch2: “Cumanian” dyke3: Körös Valley dyke4: “major Roman dyke”5: dyke sec ons reconstructed by Gyula Cziráky6: “minor Roman dyke”7: Transdanubian dykes

Actual State of Research of the Csörsz- or Devil’s Dyke. In: Identităţi culturale locale şi regionale în context european. Studii de arheologie şi antropologie istorică. Eds.

Documents