AGNIESZKA TOMAS, LIFE ON THE FRONTIER: ROMAN MILITARY FAMILIES IN LOWER MOESIA

The epigraphic evidence from four Lower Moesian legionary bases provides information about military-civilian society living along the Lower Danubian frontier zone. The inscriptions on stone originating from these four sites informs not only about the names, age or status of the people living in the border zone, but may be helpful in recreating the emergence of Roman provincial society in the military garrisons. The epigraphic evidence collected from Oescus, Novae, Durostorum, and Troesmis includes only the inscriptions related to the people who were not active soldiers and officers serving in the legions. The aim of the paper is to find out whether the epigraphic evidence may bring any additional information concerning the civilian and military families, ethnic and social composition of the inhabitants living near legionary bases, conditions of life, and the nature of the settlement.

Military Museum in Bucarest 3 . Since 1960 regular archaeological surveys have been carried out on the site, but the architectural remains were excavated mainly within the fortress. The only remains unearthed extra muros castrorum is a large residential building of an official character and late antique basilica with a burial ground, in the western part of the canabae 4 . Salvage excavations were made in the mithraeum 5 , in the very poorly published shrine of Liber Pater 6 , as well as in three cemeteries where only a few burials were unearthed 7 .
The camp in Novae was laid out on a high Danubian escarpment, on three terraces sloping downward toward the Danube. The legionary fortress covers 17.99 ha, but the spread of finds around it are traceable on another 80 ha. The area adjacent from the west is a flat and easy accessible, while the eastern side of the fortress is surrounded by the irregular terrain with flat land and a deep gully in the north-east and elevated terrain on the south-eastern side. Toward the end of the 3 rd century, a new circuit of defensive walls was added on the eastern side, enlarging the fortress by an additional 8 ha. The layout of the fortress had changed then and the legionary camp and attached civil settlements developed into the Late Roman town.
Exploring and reconstruction of civil settlement around Novae is feasible only with the use of multidisciplinary approach and requires integrated methods of landscape archaeology supplemented by archive and museum studies. A three-year project of non-intrusive surveys in the extramural area carried out in three spring seasons from 2012 to 2014 8 was supplemented by museum and archive investigations, which included inscribed stones. The epigraphic monuments, in their major part published, were mapped in the GIS database together with some other non-inscribed finds. The texts which refer to the extramural settlement or the civil inhabitants were catalogued, analysed, and compared to the evidence from other Lower Danubian legionary bases.
Recent discoveries in Novae shed light on supply of provisions for the army in the 1 st century AD. Large pits explored in the central part of the camp contained, among others, large quantities of amphorae sherds which allowed to reconstruct entire forms, some of them unique 9 . Long distance wine trade in this early period is also attested by a funerary monument of a negotiator vinarius and his wife 10 .
However, archaeology and epigraphy seem to prove that wine was not exclusively imported product from afar. It seems that the surroundings of Novae is an area where the tradition of viticulture goes way back to the Roman, or perhaps, even pre-Roman times. As it has been shown years ago by the late Jerzy Kolendo, such a hint may be a relief on one of the tombstones of a certain family of Atronii, possibly freedmen, which shows a falx vinitoria -a tool used in collecting grapes (a billhook) 11 . A popular subject in funerary arts could be just regarded as a decorative motif, but some other hints indicate that it was not a matter of coincidence. Members of the Atronii family are listed on a famous album of the Bacchii vernaculia monument of the members of Bacchic congregation, discovered in the rural hinterland of Novae 12 . 8 On multidisciplinary approach and methods used in Novae, see A. Tomas 61, 2010 (2014), pp. 79-83. 10 IGLNov 100. Wine was imported also in barrels; see J. Kolendo, Symboles des fonctions militaires et des métiers sur les monuments funéraires de Novae, camp de la legio I Italica (Moesia inferior), Novensia 22 (2011), pp. 28-31. 11 Ibidem, CIL III 6150 = 7437 = 12346 = ILBulg 438 = AÉ 2004, +1197.
Recent discoveries add to this small, but significant list some more intriguing evidence. An inscription on a statue base set up by two civilians living in the canabae, a woman and a man, described Jupiter as fructuorum conlator 13 . The crops mentioned in the text should refer to this deity and indeed, one of the traditions known from ancient sources informs about the first fruit of a harvest which the Latins offered particularly to Jupiter. According to this tradition, the offerings consisted of wine or grapes and the act of consecration -in the form of a libation to Jupiter -took place on 23 rd April as vinalia priora or urbana, and on 19 th August as vinalia rustica 14 .
Therefore, it is possible that the statue and the base was dedicated during the celebration of the vinalia.
Last but not least, all these hints are supported by the results of laboratory analyses made recently. Vine seeds were identified in soil samples dated to the pre-Roman period, as well as in the contents of a local hand-made pot discovered in the earliest layers of the camp 15 . All the subsequent layers of the legionary fortress contained vine seeds, too. It is noteworthy that the natural conditions in the area are very good for the cultivation of grapes and wine production, as it is continued nowadays by Svištov Winery.
Mapping epigraphic finds can be sometimes the same fruitful, although not easy. This concerns, for example, the evidence of Bacchic cult at Novae. A number of finds is relatively large. Due to the discovery of the inscribed base we know that Liber Pater or Bacchus was venerated in the central part of the fortress and the map of all finds seem to confirm this fact (Fig. 3). Another concentration of finds shows that he was venerated also in the eastern part of the canabae. Moreover, his shrine which was excavated in the past, existed 2 km to the south-west of Novae. Bacchic cult was particularly popular and strong in Novae and this popularity is well seen both in small finds and in inscribed monuments.
Mapping finds related to the solar cults seem to reflect the topography and confirm that such method may bring valuable results. As the map shows, Mithraic finds were discovered, i.a., reused in the western defensive walls, very possibly moved from the mithraeum which was placed around 300 m to the south-west from the fortress (Fig. 4) architectural remains and the texts of three altars discovered there, allowed to date the shrine to the end of the 2 nd and the beginning of the 3 rd centuries 16 .
Quite mysterious cult of Deus Aeternus, a deity worshipped mainly in Dacia, is also well attested at Novae. Deus Aeternus very possibly joined variety of deities believed to be powerful and eternal 17 . Two out of the four inscribed monuments dedicated to this deity were found in the Episcopal basilica and two others were discovered in another place -to the south-east from the legionary camp (Fig. 5). It is possible that the finds from the basilica originate from the western part of the canabae. Carrying stone from the western part of the canabae to the place on the south-eastern side of the fortress would be very difficult due to the shape of the terrain and inconvenient transportation. Thus, we can assume that another place of cult may have existed somewhere to the east of the camp.
The cult of Isis and Sarapis is attested at Novae by sculpted and inscribed monuments. It is possible that the place of worship dedicated to Egyptian deities existed in the western part of the settlement. Although the finds related to his cult, i.e. an altar or a base dedicated to Sarapis, an architectural element bearing dedication to Isis and a fragment of a relief with his representation 18 were found reused in three distant places, we can consider some additional hints. Firstly, the architectural element which must have belonged to some shrine was found found in the ruins of the Episcopal basilica, which is closer to the western side of the fortress. Secondly, the mentioned altar or a base is a heavy stone difficult to transport, hence we may assume that it was found not far from the place where it was originally set up.
The evidence of extramural life is not always so helpful, but it may provide some information which allows to imagine the variety of religious, social, and cultural life, as well as its quality or limitations. Sometimes mapping a certain type of finds may be an aimless procedure. Such an example seem to be finds related to the cult of Hecate called also Dea Placida. So far, we know about six finds related to this goddess 1) an altar dedicated to Dea Sancta Placida by augustalis municipii, discovered to the east of the East Gate of the legionary camp 19 , 2) an altar dedicated to Placida by praefectus legionis and his wife, discovered in the extramural residence in the western canabae 20 , 3) an altar dedicated to Numen Placidae, a stray find 21 , 4) a votive tablet bought by the Historical Museum in Svištov in 1905 22 , 5) a votive tablet bought by the National Museum in Sofia, according to the seller, found in the nearby village of Vardim 23 , 6) a fragment of a votive tablet kept in the Historical Museum in Svištov 24 . Although only two of these finds have an established archaeological context, it is possible that their potential spread would be random. Hecate, as a goddess of fate and witchcraft, underworld and ghosts, had shrines in households and on crossroads. She was venerated in liminal places and in liminal moments of life, like childbirth and death. Such a broad realm could be the reason for the evidence of her cult to be found in many places within the site. What is interesting, is that the cult of Hecate is well attested in Dacia inferior, not in Moesia 25 , hence Novae is an exception within the entire province.
Apart from the evidence of cult practice, we can trace some features of the civil settlement. One of the inscriptions from Novae possibly mentions the renovation of a water source (fons) by two veterans, certainly inhabitants of the canabae 26 . Two altars dedicated to Quadriviae -protective deities of crossroads discovered near the West Gate 27 . Both altars were set up by civilians, hence they may have stood outside the camp, most probably at the crossroads to the west of the West Gate. All these hints, combined with the spread of the archaeological finds visible on the surface, lead to the conclusion that the main part of the canabae was situated to the west of the fortress.
Novae provided a quite significant amount of epigraphic finds, thus also people's names. Although some inscriptions are only partly preserved and some of them do not inform whether they give a name of an active soldier or a civilian, we can assume that at least 148 persons we know by name might have lived extra muros castrorum. Among them at least 22 veterans, 25 members of military families, 79 civilian inhabitants, and 22 persons whose military or civilian status remains unknown 28 . The tombstones dated to the 1 st or the beginning of the 2 nd century mention 46 people living most probably in the canabae. The 2 nd -century inscriptions name 76 people, 69 of them civilians. The inhabitants of the canabae left far more inscribed monuments than the military and the majority of them is dated to the 2 nd century. This period should be treated as a moment of the greatest development of the settlement.
It is expectable that the surroundings of the legionary fortress should be inhabited by the civilians and the military families. Surprisingly, the names of the relatives of the soldiers and officers occur not as often as of the civilians. This fact may be partly explained by the so-called "marriage ban" on active soldiers 29 , However, from the reign of Claudius soldiers could have established heirs in their wills, including children born in illegitimate relationships 30 , hence their relatives should appear on tombstones. The children of soldiers and of veterans raised their own families and were commemorated by their own families, who were not necessary military. Perhaps some women who were in relationships with soldiers and veterans moved to the countryside or did not leave any trace of their "military marriage", if they remarried 31 . 28 A. Tomas, Living with the Army I. Civil Settlements near Roman Legionary Fortresses in Lower Moesia, Warszawa 2017, p. 127. 29 In fact one cannot speak about a ban. The soldiers (milites) did not possess the legal ability to enter a full Roman marriage (iustum matrimonium). They had such status probably until AD 197 and -in consequence -their informal families had no claims to their estate unless they were named heirs in the wills. The officers and perhaps centurions, were possibly allowed to enter into such marriages also before this date. Such privilage is based on the evidence given by Herodian . Some scholars consider the information given by Herodian as unreliable, however; see A. Tomas Only 10 out of the 42 women known from Novae were certainly related to the military 32 . There are two cases of presumably active soldiers who commemorated their wives. The first involves a freedwoman commemorated by a trumpeter (cornicen), most probably of the First Italic Legion. The cornicen could be a veteran, in fact, but this presumption should be treated as the same possible as the fact that he was an active soldier who was married and the monument should be dated to the 3 rd century. The second case -of a centurio who commemorated his one-year-old son, does not necessarily have to be dated to the 3 rd century, as it was in the primary publication, as centurions with families are attested before the end of the 2 nd century 33 .
The inhabitants of the canabae arrived to Novae from various parts of the Empire. One of the earliest monuments is a tombstone of an Etruscan family from Crustumiae Arniae 34 . Etruscan origin has also been attested for a producer of clay pipes through stamps bearing the name Arrius and dated to the Flavian period. The Arrii family has been attested in the rural hinterland of Novae, namely at Butovo, which was a big production centre for pottery and building materials, hence it has been suggested that the family was involved in clay manufacturing there 35 . It is possible that Arrius started his business in the canabae and his descendants developed it in Butovo, which had strong economic ties with the legionary fortress 36 . Apart from people from Italy, Spanish Asturae also lived at Novae, serving in Moesia in the 1 st century AD 37 . Among veterans were former soldiers who did not 32 Ibidem, This monument was dated in IGLNov by J. Kolendo to AD 201-250, but Fr. Feraudi-Gruénais in the Epigraphik Datenbank Heidelberg (henceforth: EDH) follows the dating to AD 151-200 as proposed by S. Conrad (Die Grabstelen aus Moesia inferior. Untersuchungen zu Chronologie, Typologie und Ikonografie, Leipzig 2004, p. 232, No. 393;Taf. 102, 4 = EDH, No. HD42193). 34 The monument was found together with another gravestone of a cornicen leg. VIII Aug. (IGLNov 81). Such a find context became the basis for an argument presented by G. Forni that the commemorated family not only may have lived in the canabae of the Eighth Augustan Legion, but even might have been related to an individual from the military of that legion, but J. Kolendo Studies, 7, Bucharest 2010, pp. 167-169. serve in the First Italic Legion, but decided to settle down at Novae, like a certain Thracian named Tarsa Bassus, a former miles cohortis I Tyriorum sagittariorum, who lived in the canabae just after Trajan's Dacian Wars 38 . A former centurio, who had served in the Eleventh Claudian Legion based at Durostorum decided to settle down in Novae, as well 39 . In the first half of the 2 nd century, a soldier previously enrolled in Claudian municipium Virunum in Noricum (now Zollfeld in Austria) retired and died at Novae, where he was buried by his friend 40 . Another veteran born in Ancyra lived at Novae until he was a very old man 41 . In the 2 nd century AD, a certain Aurelius Victor arrived in Novae from distant Piroburidava and raised a family with Aelia Dionysia.
The majority of the civilian inhabitants of Novae bore the names of Roman citizens, while the smallest group is formed by people bearing peregrine names (6%). Among them Greek, Thracian, Celtic, Etruscan, and possibly Illyrian 42 . Thracian names in the canabae are attested only in four cases and all of them belong to men who were related to the army. Thracian and Dacian female names are absent among the civilians living near the base. This, however, does not mean that the local population was a small minority absent in the canabae. It is quite obvious that the epigraphic habit is a phenomenon of Roman and Greek culture and was practiced by the Romans and those who were Romanized at least superficially. Those who were Romanized include people who changed their native names into Roman ones. Such a case is that of famous Aurelius Statianus, a former soldier in a fleet and a producer of building materials, known from two inscriptions discovered at Novae, whose origin is recognized thanks to a military diploma in which he is named Apta 43 . In some cases, the names of children and other members of the family provide a hint suggesting the origins of a certain person. Such is the example of a boy from Novae whose name, Mampsalachanus, may be of African or Eastern origin, while his parents bear common Roman names. Another case refers to a veteran of the First Italic Legion who settled in the countryside. name, his daughter's and brother's names are Thracian 44 . Similarly misleading is the case of a certain Antonia Optata from Novae, whose customary name, Maura, may indicate either a dark complexion or African roots, but this is not indicated in her "official name" 45 . The majority of the families living in the canabae must have been ethnically mixed. Hence, a change of a name seems to have been unexceptional 46 . We can imagine that in the civilian-military environment of the canabae, Roman homines novi could and wanted to possess Roman-sounding names as an expression of their social promotion.
Greek names which prevail among the civilians does not mean that the civilian inhabitants of the canabae were people of Greek or eastern origin.
Slaves sold at the market at Ephesos, Delos or Delphi were very often renamed, and their new names could be Roman or Greek, but only Greek names can be recognizable as those given to the slaves 47 . Among the Greek servile names recorded in Novae, we find Glaphyrus, Elpidephorus, Chresimus, Timocrates, Abascantus, Onesiphorus, Eutyches and Therapo. Sometimes only the names of the freedmen' family members, are helpful in determining the possible low social status or slave origins of people. Freedmen who bear Roman names are recognizable in cases when they belonged to rich Italic families, i.e. Caeserni, Metelli, or Paccii, but also those who are probably related to imperial slaves. Caeserni originated from Aquileia, but in the 1 st century they settled down in Emona, and later in other  33 (1982) 1985. The name Optatus appears in the Western provinces, mainly in Hispania, southern Gallia and Italia (Nomenclator,p. 209). The assumptions concerning the presence of the Mauri at Novae were based on the fact that the First Italic Legion was supposed to participate in their revolt during the reign of Antoninus Pius. However, such theory is very doubtful, since it is based on two finds of stamped building materials: CIL VIII 10474, 13 et p. 911, Numidia, Constantine / Cirta, now lost, and the other, of doubtful provenance, kept in the Museum in St. Germain. On the stamps and their interpretation, see E. Ritterling, s.v. Legio, RE 12, 2, 1925, col. 1407-1417 As was the case of a sailor serving in Egypt, see P. Weiβ, Zu Vicusangaben und qui-et Namen auf Flottendiplomen des 3. Jhs, ZPE 130, 2000, pp. 279-285. 47 Some Greek names were typical for freedmen although they were not popular among the Greeks. Moreover, six out of ten slaves from Rome bore Roman names which made them undistinguishable among freeborn people; H. Solin Pannonian cities, including the military sites of Brigetio and Aquincum 48 . In Novae, a couple bearing such a name is attested in the 2 nd century. As it was stated above wine trade is very well attested at Novae. Among other occupations, we find a civilian medic and local wine producers. The descendants of rich freedmen played a very important role in the local society as shown by the example of the Oppii. The altar discovered at Nicopolis ad Istrum set up by two members of this family -one of them a priest of Mater Deum (i.e. Cybele) and the second of Liber Pater -is very important evidence of the social and religious life of the canabae. The text of the inscription provides information that the two priests set up an altar under the standards of Novaensium Oppianorum 49 . It has been suggested that the Oppii could have been firemen as one of them was a priest of Cybele, in which dendrophori played an important role, and dendrophori were once considered to be involved in fire-fighting, the same as fabri and centonarii 50 .
However, the theory about the connection between fire-fighting duties and the fabri (not dendrophori) based on Plinian testimony has been rejected widely by scholars 51 . We can assume that the mentioned vexillum was in all likelihood the standard of a sacral collegium 52 , not that of firemen or any other paramilitary association 53 . However, their religious positions may have been linked with some other activities in the canabae, e.g. social or economic. It has been suggested that the centonarii could have been connected with trade and the provision of supplies to frontier towns 54 . Perhaps in frontier towns the members of these collegia were active both in religious life and in business of various kinds.
Some of the inhabitants of Novae must have enjoyed wealth. One of the tombs, which was built for a veteran and his wife, was located on a plot 48 A. Móczy, Pannonia and Upper Moesia, London-Boston 1974 (2014) In epigraphic evidence, women are usually recorded directly following the men. However, the women sometimes came first if they were the founders of the tombstone or the deceased to whom it was dedicated. In other cases, the reason behind such a placement might have been the higher social position of the women 57 . Although in this case the monument was erected by their heirs, who carried her gentilicium (Aurelii), the fact that she was mentioned first before her husband, may have had also other reasons, like the influential position of the family. Looking at the shape of the tombstones and their decoration, we can observe common motifs, but also some peculiarities which perhaps were specific for Novae. The gravestones from Novae commemorate mostly men, very rarely children, and usually they were born in the military families. Tombstones depicting the deceased were infrequent and it is worth noting that we do not know of any portrait of a child from Novae, while in Dacia funerary reliefs from various military bases show military families with children 58 . The quality of the stone should not be considered as the reason behind such differences. Funerary monuments from Novae were usually carved using relatively high-quality limestone, while those from Dacia were made not only with limestone, but also sandstone and andesite of an inconvenient porphyritic texture. We can assume that the funerary habit either was not focused on portraying the deceased or the local workshops were not specialized in such reliefs 59 . Moreover, among the few portraits discovered at Novae the majority had been intentionally damaged at some point in the past. 55 Veterans were a relatively wealthy social group; see K. Królczyk

OESCUS
Oescus was the legionary base of the legio V Macedonica, probably from the reign of August or rather Tiberius 60 . In the times of Nero, the Legion was to take part in the Armenian expedition together with the newly created First Italic Legion, and was absent in the military base between 62-71 61 . At that time, auxiliary detachments may have stationed there 62 . The military base existed up to the end of the second Dacian Wars of Trajan, when the Legion was transferred to Troesmis, i.e., between 102 and 106 63 . The foundation of the colony is dated between 106 and 112 64 .
The archaeological remains of the legionary fortress are almost certainly placed below the buildings of the later colonia 65 . The majority of researchers accepted the theory that the principia were situated at the place where later was the forum 66 , and -as a camp built in the Augustan-Tiberian period -may have had an irregular pentagonal plan. Still, its exact area and plan remains unknown and the line of defensive walls are reconstructed on the very weak premises 67 . The exact location of the canabae is a debated subject.
The scholars rather exclude the northern side of the fortress, as this area was swampy in the past, although some finds from the Late Bronze Age and Iron be placed on the territory of the later annex (Oescus II), but this presumption is based only on the fact that some residential buildings dated to the 3 rd century were discovered there 69 . For obvious reasons the epigraphic and archaeological evidence of civil settlement is far less than that at Novae. The epigraphic evidence dated to the 1 st century AD provides information about 21 persons: 9 veterans and members of their families, 9 civilians, and 3 persons of undetermined military or civilian status 70 . Among seven veterans, five certainly served in the Fifth Macedonian Legion, one in the Fourth Gallic Cohort and one in undefined unit.
One of the monuments was found embedded in the church of the presentday village Bajkal 71 . It has been suggested that the commemorated veteran could be buried in his property outside Oescus 72 and this assumption cannot be excluded.
At least five freedmen attested at Oescus form the biggest group of civilians known from the 1 st century. The earliest known monument commemorates a freedmen of a centurio leg V Mac. Other freedmen are Hermes, a former slave of a veteran G. Iulius Longinus, Eutyches, and a freedmen of the Moesian legate from the times of Domitian. A partly preserved name of [---]aculia -perhaps -l(iberta) is the only female example of a former slave. One monument commemorates a veteran of an unknown name, but gives the names of certain Braetii, the family originating from northern Italy (Venetia et Histria) 73 , hence possibly merchantmen or freedmen. Nothing can be said about the heres of the duplicarius alae Pansianae 74 , apart from the fact that his testator originated from Treveri.
Only by chance we receive information concerning life in the canabae of Oescus. Unnamed heres who commemorated a signifier of the Legion, 69  Q. Philippicus from Edessa 75 set up the monument according to the decision (arbitratu) of Antonius, who was an architect (architectus) and Titus, a leather worker or a tanner (coriarius) 76 . Antonius and Titus were arbiters in inheritance obliged to control the heres in execution of the Philippicus' will. It is noteworthy that both arbiters were identified by their gentilicia and occupations instead cognomina or origin. This small detail shows their important role in the services provided to the army.
One of the most interesting inscriptions is a tombstone of L. Freius l. lib.
Faustus, who was described as lixa of the Fifth Legion. A term lixa is not fully recognized, but certainly refers to an army sutler 77 . No civilian women are attested on the stone monuments from the 1 st -century Oescus.

DUROSTORUM
Prior the encampment of the legion, an unknown auxiliary unit was deployed at Durostorum, but probably not earlier than in the Flavian period 78 . In AD 79 legio III Gallica marched along the Danube, and perhaps it encamped at Durostorum 79 . It has been suggested that C. Iulius Apo[---], veteran, who commemorated himself and his wife, most probably at the turn of the 1 st and 2 nd century AD was discharged from one of the units based in this early period 80 . However, excavations have not proved the existence of an auxiliary fort.
The Eleventh Claudian Legion was detached to Durostorum between or just after Dacian Wars of Trajan, but the exact moment of its arrival is a subject of a debate 81 . The castra probably cover an area of about 22  which would be 4 ha more than the Trajanic camp at Novae. The ancient layers of Durostorum are covered by a modern city of Silistra, which is an obstacle for any regular excavations. Only salvage investigations are possible and the results of them were published in a few articles and book chapters.
The major part of the canabae is situated to the north from the military base and its estimated area is approximately 60 ha 82 . Five buildings with baths have been unearthed so far within the area of the canabae 83 . The buildings discovered in the canabae seem to have been solid, many of them had heating system, water supply and sewers 84 . The settlement was provided with paved streets with canals. The castra and the canabae were destroyed in AD 170, i.e., during the Costoboci raids 85 . The settlement existed throughout Antiquity, with the most flourishing residences in the 4 th century AD. During the 6 th century AD the buildings in the extramural area changed their function into manufacturing sites 86 . The same at Novae, at a distance of 2-3 km from the canabae, another settlement (vicus) existed. The second settlement which possibly developed from the local site may have had a name of Gavidina, known from an inscription 87 .
The epigraphic evidence provides quite significant information concerning the status of the civil settlement and its importance. Among them are texts which mention the members of the magistrate. It is a debatable subject which settlement (canabae or vicus) was granted municipal status 88  Herculanus, a veteran, bears a dedication made to Mercury for the veterans consistentibus huius loci (AÉ 1974, 571 Eight freedmen or people of servile origin are attested at Durostorum, among them two people of the Oppii family, a father and a son, who dedicated a temple, a statue and an altar dedicated to I.O.M. for the health of the emperor A. Pius and caesar L. Verus 96 . Although they did not state their positions or any relations to business, it is obvious that they must have been wealthy men and that they were of servile origin.
One inscription from Durostorum should draw attention again. Its text published in 1887 by G. Tocilescu was reconstructed according to the testimony of an inhabitant of Silistra 97

TROESMIS
Troesmis was a base of the Fifth Macedonian Legion which was dispatched there from Oescus after the end of the Dacian Wars and remained there up to AD 162 when it was sent to the Parthian campaign, and after AD 166 moved to Potaissa. It is possible that prior the arrival of the Fifth Macedonian Legion ala I Pannoniorum was garrisoned here 101 , but the earliest inscriptions come from the times of Hadrian 102 . The detachments of the legio I Italica were based at Troesmis after 167 103 .
Thanks to the recent non-intrusive surveys it has been established that the legionary fortress was situated on a plateau adjacent to the high riverbank sloping down to the waters of the Danube. Two longer sides of the camp were identified, as well as the shorter side in the north-east 104 , but the total length of the fortress remains unknown due to the erosion of the riverbank.
The civil settlement was stretched out to the north-east of the fortress. Its topography is recognized only to a certain degree as the basis for the plan is the image obtained from the geomagnetic surveys. Several streets and rather densely packed quarters of buildings are visible there. Very modest archaeological evidence remain in contrast with the rich epigraphic evidence. However, the the municipal status granted between 177 and 180 must certainly refer to the period when the legion was dispatched to Dacia.
Nevertheless, the magistrate of the canabae is considerably well attested there.
According to the inscriptions, more or less 65 people may have lived in the canabae legionis. The vast majority of them (50) were people related to the army -among them veterans and their families. A considerable group of the soldiers and veterans who served in the Fifth Macedonian Legion originated from the East (Asia, Bithynia, Syria, and possibly from Iudaea 105 ). Others were born in Italy, Pannonia and Lower Moesia (Oescus). Such a composition reflects local enrolment made in the 60's of the 1 st century AD when the Legion was stationed in Bithynia, and later was sent to supress the Judean revolt. The latest recruitment before arrival to Troesmis was made in Oescus, most probably in the canabae 106 .
Predominance of the Bithynians in Troesmis fits well into the ethnic pattern of Dobruja where the significant part of population originated from Asia Minor, particularly from Bithynia and Pontus 107 . Economic connections may have been based, among others, by the veterans from Troesmis 108 .
The earliest attested members of the magistrate of the canabae most probably originated from Italy, Pannonia, and the East 109 , hence represents well the ethnic composition of the settlement. The evidence mentioning local magistrate seem to indicate that the veterans played significant role in selfgovernment and constituted the local elite. Contrary to other sites, freedmen are less attested than the veterans. It is worth to note, that the stone used for setting up the monuments was imported from distant areas of Dobruja, which means that it was more expensive and not available to a common inhabitant of Troesmis. Only wealthy group of inhabitants could afford it, and among them were veterans who were members of the magistrate.
Perhaps, the composition of local elites (including tradesman, augustales etc.) was more diverse, but stone monuments were left by this specific group.

CONCLUSIONS
The canabae of Oescus were the first civil-military settlement in Moesia.
Regrettably the state of research and the difficult conditions of archaeological investigations on the site make our knowledge about the early settlement very limited. The canabae at Durostorum must have been a wealthy and prosperous settlement, but it is difficult to excavate its remains. The present city of Silistra hides many secrets, and perhaps some more evidence will come to light in the future. The same mysterious remains the canabae of Troesmis, but in this case we can expect some discoveries, as the site is available for archaeological investigations. Among three settlements presented above, only Novae provides information from the 1 st to the 6 th century and the number of the inscriptions is prevailing. The importance and variety of information achieved from inscriptions makes it possible to reproduce the character of the canabae and its inhabitants.
Comparing the three legionary settlements placed on the Lower Danubian border, we can draw some conclusions, which are not always applicable to all of these sites.
According to their names, women in relationships with legionaries and veterans seem to have been mainly of free status, and this rule applies not only to Novae, but also to other sites in Lower Moesia and in Lower Pannonia 110 . The veterans tended to choose Roman freeborn women with whom they could form a matrimonium iustum 111 . Nevertheless, at Novae we find Valeria Elphis and Iulia, who were the former slaves of their military husbands. A detailed analysis shows that the majority of the freedwomen living at Novae were in relationships with civilians, not soldiers. This phenomenon seems to indicate that the soldiers and veterans made deliberate choices to find citizen wives and did not shun other social groups 112 and former soldiers, who very often had Balkan roots, held on to Roman culture 113 .
The first settlers who left evidence of their presence were foreign traders and their representatives, but also veterans and their families. At the turn of the 2 nd and 3 rd centuries, the people living in the canabae formed a wellintegrated society with a civilian-military elite. This integration is visible not only in acculturation traceable e.g. in archaeological finds but also in epigraphic evidence. Some of the inhabitants of Novae and other neighbouring towns were local military families for generations. This process can be traced as early as from the beginning of the 2 nd century, when in Oescus lived Iulia Tyche, the wife of C. Iulius Celer, veteran of the Fifth Macedonian Legion and the mother of C. Iulius Crescens, centurion of the First Italic Legion. The emergence of local society involved indigenous population, which is traceable in religiosity and evidenced by dedications to deities, such as Apollo Kendrissos, the Thracian Horseman, Sabazius, and Deus Aeternus 114 . The liaisons with the rural hinterland are visible also in people's names, which in some cases sound the same in the canabae and in the rural hinterland. Such an example is Valerius Festus, listed in the Bacchic album from Boutovo, but also appearing in the text of a votive tablet from Paskalevec in the rural hinterland and on a funerary monument from Novae 115 . Perhaps this was one or two or even three different persons, but certainly such repetitive names are an expression of a change in the society which united autochthons and those living in the canabae 116 .
In the 1 st and 2 nd century an opportunity to develop local business was attractive enough to settle down near the fortress. Although the conclusions are very much assumptive, it is possible that the family of Oppii who were very well known traders reached Lower Danubian lands through their freedmen embed in the vicinity of the military camps. Delos was famous for its slave market 117 , but in the east, at least on Delos, the Oppii family was most probably involved in wine trade 118 , and perhaps wine trade was the reason of their activity along the Lower Danube. In the late 2 nd -3 rd c. their descendants might have become influential people in the civilian frontier towns.
The evidence of the civil inhabitants of Troesmis is quite different than those living near the legionary base at Novae and Durostorum. Only eleven freedmen are attested there, two of them former slaves of physicians (medici). The freedmen of the influential families of traders or merchants are not so well attested, and this is quite striking. At the same time, the veterans in Troesmis seemed to play far more important role than in other Lower Danubian sites. This phenomenon should be perhaps analysed against their settlement in the hinterland. Certainly the composition of inhabitants in Novae had influence on the process of municipalisation of the canabae, where veterans' associations seem to have been not as active as in Durostorum and Troesmis 119 .
Inscribed monuments may provide information about the social and ethnic composition of its inhabitants, but sometimes can be helpful in reproducing the landscape around the fortress, the topography of the surroundings, religious life, and the nature of the settlement itself. A statistical approach provides further information concerning the development and prosperity of such a settlement. The prosperity of the legionary settlements reached its peak in the 2 nd century until the Severan period, when it was finally stopped by the barbarian raids. Depopulation and the destruction of the civil settlements combined with the consequences of a series of reforms resulted in the emergence of the Late Roman towns.

Summary
The present paper is a contribution to the civil settlement near four legionary fortresses based on the epigraphic evidence from four Lower Moesian sites: Oescus, Novae, Durostorum and Troesmis. The earliest evidence of extramural settlement in Moesia provides Oescus which was founded at the beginning of the 1 st century AD as a legionary camp and remained the military base until the foundation of the colony at the beginning of the 2 nd century. Novae provides archaeological evidence from the longest period -from the middle of the 1 st century until the beginning of the 7 th century AD. Durostorum became a legionary base in the Flavian period and the same as at Novae, the military unit remained there until the late Antiquity, which was a convenient factor for a more permanent and larger settlement to develop. Both at Novae and at Durostorum existed another settlement in the vicinity of the campthe vicus, located at a certain distance from the fortress. At Troesmis civil settlement may have existed prior to the arrival of the legion and it is possible that the settlement pattern was different there than in other Lower Moesian legionary bases.
The state of research and the conditions for archaeological investigations in four sites mentioned above is not the same. The settlement at Durostorum existed throughout Antiquity, Medieval period and survived until nowadays, therefore the ancient layers of the castra and the canabae at Durostorum are covered by a modern city of Silistra, which is an obstacle for any regular excavations. The foundation of a colony at Oescus and the presence of a modern village of Gigen very close to the site makes surveys on the 1 st -century settlement the same difficult. Although the access to the archaeological remains in Troesmis is far better the state of research on the legionary fortress and its extramural settlement is very modest. The accurate position of the castra and the canabae at Troesmis has been established only recently as the result of non-intrusive surveys. The excavations carried out so far, have been focused on the remains of two Late Roman fortresses positioned to the west and east of the 2 nd -century legionary camp. Novae is the only site which has been excavated regularly, but in this case the archaeological excavations were carried out mainly within the legionary fortifications.
The number of inscriptions originating from them varies. This results from their different history (their shorter or longer time of existence) and different state of research. The larger number of inscriptions provides Novae. Based on these texts we know at least the names of 148 persons who might have lived extra muros castrorum (at least 22 veterans, 25 members of military families, 79 civilian inhabitants, and 22 persons whose military or civilian status remains unknown), the majority them are dated to the 2 nd century, when the civil settlements must have been flourishing.
As it is indicated directly in the inscriptions the inhabitants of the canabae arrived to Novae from the various parts of the Empire, i.e. from Italia, Hispania, Noricum, Asia, as well as from the Lower Danubian provinces. In their major part civilian inhabitants of Novae bore Roman sounding names, followed by the Greek names, while the smallest group is formed by the peregrines (6%), but only in four cases they belonged to Thracians, all of them men. The sounding of the names, however, may be misleading, especially in case of those people who were of servile origin. Former slaves and their descendants formed quite considerable group of inhabitants of Novae.
Although the number of inscriptions discovered so far at Durostorum is lower than at Novae, they provide very important information concerning the status of the settlement. We know about 72 people who may have lived in civil settlements near the legionary fortress (25 veterans, 37 civilians and 10 people of unknown status). Almost half of these names are known from three inscriptions dated to the end of the 2 nd and the beginning of the 3 rd century. The family of Oppii draws special attention among freedmen attested both in Novae and Durostorum. At Novae two people bearing such name held positions of priests of Cybele and Liber Pater and at Durostorum the members of the same family were rich enough to build the temple and fund a statue and an altar dedicated to Jupiter. It is possible that the members of this family were involved in trade and provisions to the army.
Although Troesmis was excavated only within these Late Roman sites and only partially, the number of known inscriptions is relatively high. The status of Troesmis is attested by preserved lex municipii. Fifty out of sixty-five people known from the inscriptions were related to the army, with considerable number of veterans, which stands in contrast to the social composition of Novae. The inscriptions from Oescus mention 21 persons, among them 9 veterans and members of their families, 9 civilians, and 3 persons of undetermined military or civilian status. Civilians were mainly freedmen and craftsmen. Interestingly, women are very poorly attested.
Compared with Troesmis and Durostorum, Novae had different social composition. Freedmen and people of servile origin were prevailing and the collegial communities of veterans are so far not attested there, while they were active both at Troesmis and at Durostorum. This phenomenon might have had influence on the process of municipalisation.
Apart from people's names, some texts provide various information concerning everyday life, like trade and agriculture. E.g. local viticulture and import of wine is well attested at Novae. The importance of local wine production is attested by a relatively high number of dedication to Dionysus or Liber Pater discovered within the fortress, but also in the texts discovered in the rural hinterland of Novae. This observation is confirmed by archaeological finds and laboratory analyses of soil samples from the culture layers and from the locally hand-made jar which contained seeds of grapes.
Mapping the finds related to certain deities may allow to recreate places where the monuments were originally set up. Such methodological approach is possible in case of Novae. Bacchic cult is attested inside and outside the fortress, while the finds to another popular cult -of Solar deities were discovered, i.a., reused in the western defensive walls, not far from the mithraeum. This fact seem to prove that the method of mapping in some cases may be a valuable tool. The map of inscribed monuments dedicated to Deus Aeternus -a deity worshipped mainly in Dacia, but also well attested at Novae -shows that the finds may have originally come from the southeastern side of the fortress.
The ethnic composition of the Lower Moesian canabae indicates that the majority of the families living in the canabae must have been ethnically mixed, but the process of cultural homogenisation was gradual. The first camp followers attested in Moesia were those at Oescus, and the epigraphic evidence shows that they were foreign traders and their representatives, veterans and their families. Already at the turn of the 2 nd and 3 rd centuries the inhabitants of the Lower Moesian legionary extramural settlements formed a well-integrated society, with the liaisons with the rural hinterland, as it is attested by the inscriptions from Novae and its hinterland.