DRAWING THE LINE: AN ARCHAEOLOGICAL METHODOLOGY FOR DETECTING ROMAN PROVINCIAL BORDERS
K. da Costa
Mommsen published Die Provinzen von Caesar bis Diocletian in and although our understanding of the limits of official authority (provin- cia)isnowmorenuanced,1 itdoesnotappearthathistorianshavemade very much progress in defining the spatial limits of territorial provinces. Indeed, the standard geographical reference states clearly, at least of the eastern provinces, that “Provincial boundaries are approximate and in many cases, very uncertain”.2 Provinces were, nonetheless, the essence of the empire: next to the standing army, the administration of the provinces was what kept the empire together for over years. While it is more or less clear how provinces were accumulated—from inheritance through conquest to acquisition—it is much less clear what they were for. Not one single ancient source describes the rationale behind the definition of territorial provinces, nor the reasons behind the transfer of territory from one to the other. Our only reference is the unreliable testimony of Lactantius that Diocletian chopped up the Empire to give more jobs to his cronies.3 In terms of the functions of provinces there has been relatively little scholarship using documentary evidence to assist in clarifying the prob- lem of territorial assignment, although prosopography gives invaluable information on the roles of officials in provinces,4 and there remains
1 Inter alia F. Millar, The Emperor in the Roman World (London , nd ed.); E. Meyer-Zwiffelhoffer, Politikôs archein. Zum Regierungsstil der senatorischen Statthalter in den kaiserzeitlichen griechischen Provinzen (Stuttgart ). 2 T. Elliott—R. Barckhaus, ‘Egypt after the Pharaohs: Late Roman Egypt (map)’ at www.unc.edu/awmc/awmcemap.html (, last accessed ). 3 Lactantius, de mortibus persecutorum .; C. Roueché, ‘Provinces’, in P. Brown— G. Bowersock—O. Grabar (eds.), Late Antiquity: A Guide to the Post Classical World (Cambridge ), f. 4 Prosopographia Imperii Romani, –; PIR2 http://www.bbaw.de/forschung/pir; Prosopography of the Later Roman Empire, Cambridge: vol. , A.H.M. Jones, J.R. Mar- tindale, J. Morris (eds.) ; vol. , J.R. Martindale (ed.) ; vol. , J.R. Martindale (ed.) .
K. Da Costa - 9789004215030 Downloaded from Brill.com09/28/2021 09:09:39AM via free access k. da costa much to be mined from other documentary sources, as recently demon- strated by Sipilä.5 It is time to use other data, and a potentially fruitful starting point is to better improve our knowledge of provincial borders, so that, by observing the changes over time, we might be able to calculate the reasons for those changes. It is still a commonplace to ascribe to Dio- cletion the major provincial reorganizations in Late Antiquity, and yet, as the history of change in Judaea/Palaestina and Arabia shows, small to large sections were reassigned both before and after his reign.6 Provincial territory was defined from the inside out, as it were, with cities and their associated lands being assigned to a province, and the outer extent of all those territories forming the border. Some of this terri- torial knowledge is preserved in a variety of ways, somewhat better in the west than the east.7 The province in which particular cities were placed is usually known from general historical documents, or more specific texts such as itineraries, geographical works or even church council atten- dance lists,8 and these often also indicate dependent settlements. Inscrip- tions are invaluable sources. City (or other) territories can be defined by cadestrations or boundary markers.9 Some provincial boundaries can be calculated where they cross major roads, based on milestones, or where there were settlements at the border.10 Land deeds show clearly
5 J. Sipilä, The Reorganisation of Provincial Territories in Light of the Imperial Decision- Making Process: Later Roman Arabia and Tres Palaestinae as Case Studies (Helskinki ). 6 E. Kettenhofen, ‘Zur Nordgrenze der provincia Arabiae im . Jahrhundert n. Chr.’, Zeitschrift des Deutschen Paläestina-Vereins () ff.; T.D. Barnes, The New Empire of Diocletian and Constantine (Cambridge ), ; G.W. Bowersock, Roman Arabia (Cambridge ), ; ; P. Mayerson ‘Justinian’s novel and the reorganization of Palestine’, Bulletin of the American Schools of Oriental Research (), –; Y.Tsafrir—L. Di Segni—J. Green, Tabula imperii Romani: Iudaea—Palaestina: Eretz Israel in the Hellenistic, Roman and Byzantine periods: Maps and Gazetteer (Jerusalem ), . 7 e.g. J.-C. Beal ‘Territories des cités antiques: notes de géographie historique en region lyonnaise’, Revue des Etudes anciennes / (), ff. 8 e.g. Ammianus Marcellinus, Roman History –; the Bordeaux Pilgrim—P. Gey- er—O. Cuntz ‘Itinerarium Burdigalense’, in P. Geyer—O. Cuntz (eds.), Itineraria et Alia Geographica (Turnholt ), ff.; Eusebius, Onomasticon. 9 S.L. Dyson ‘Settlement patterns in the Ager Cosanus: the Wesleyan University sur- vey, –’, Journal of Field Archaeology (), ff.; J.-F. Breton, ‘Les inscrip- tions forestières d’Hadrien dan le mont Liban’, IGLSyr ., (Paris ), nos. –; M. Sartre, ‘Appendice: Bornes du territoire ou marques de propriété?’, Syria (), f.; J. Seigne, ‘Les limites orientale et meridionale du territoire de Gerasa’, Syria (), ff.; F. Millar, The Roman Near East, bc–ad (Cambridge ), ff. 10 R. Laurence, ‘Milestones, communications, and political stability’, in L. Ellis—
K. Da Costa - 9789004215030 Downloaded from Brill.com09/28/2021 09:09:39AM via free access a methodology for detecting roman provincial borders that boundaries of individual properties were well known,11 as we should imagine, given that property was taxable. In theory then, the limit of ter- ritory dependent on each city was known, based on tax records of land holdings, and thus, the provincial borders were known fairly exactly, even if there was no physical marker. But relatively few land deeds survive, and most of the rest of this type of written information fades away by the th century. That the lines of provincial borders were known in antiquity, even if this knowledge has not survived for us, conforms with our understand- ing of the interest of Romans in boundaries of many types. We know that Roman law was sophisticated enough to distinguish conceptually between the finis (limit) and limes (boundary) of land, and between land delimited by a natural feature and land measured out.12 Markers of land depended on whether the property was ager arcifinius or ager limita- tus: the former delimited by natural features, such as mountain ridges or rivers, the later by termini of stone or wood.13 Rivers had particular connotations,14 but in the more arid east, Kennedy has reminded us that watersheds might be as important.15 Ruling on territorial disputes or con- ducting audits of provincial territory was a common duty of provincial officials, implying both a record of land holdings, and the use of survey- ors to determine claims.16
F. Kidner (eds.), Travel, Communication and Geography in Late Antiquity: Sacred and Profane (Aldershot ), ff.; Toponyms: Ad Fines in Dalmatia: R.J.A. Talbert (ed.), Barrington Atlas of Greek and Roman World (Princeton ), Map E. 11 From Nessana: PColt (Nov. , ad; C.J. Kraemer, Excavations at Nessana III: The non-literary papyri (Princeton )) a notice of the transfer of land from the brothers Abraham and Abu-Zunayn, sons of Sa"ad Allah, grandsons of Valens, to a fellow soldier Thomas, son of #Awidh, grandson of Ammonius. The boundaries of the land were: E: property of Abla, son of Darib; N: the same; W: property of Zeno, son of Firsan; S: the desert (i.e. not fixed). 12 O.A.W. Dilke, The Roman Land Surveyors: an introduction to the agrimensores (Newton Abbott ). 13 E. Hermon, ‘Le concept d’ager publicus et l’équivalence ageroccupatorius/ ager arcifi- nius chez les Gromatici’,in D. Conso—A. Gonzalès—J.-Y. Guillaumin (eds.), Les vocabu- laires techniques des arpenteurs romains (Franche-Comté ), ff. 14 D. Braund, ‘River frontiers in the environmental psychology of the Roman world’, in D. Kennedy, The Roman Army in the East (Portsmouth RI ), ff. 15 D. Kennedy, ‘The identity of Roman Gerasa: an archaeological approach’, inG. Clarke (ed.), Identities in the Eastern Mediterranean in Antiquity (Sydney /), ff.; . 16 e.g. Pliny the Younger, Trajanic Letters ; G.P. Burton, ‘The resolution of territorial disputes in the provinces of the Roman Empire’, Chiron (), ff.
K. Da Costa - 9789004215030 Downloaded from Brill.com09/28/2021 09:09:39AM via free access k. da costa
In contrast to the written sources, archaeological data, particularly ce- ramics, are abundant in every province. Distribution patterns of locally produced ceramics have long been recognized as significant indicators of local economic activity,17 and the opportunity exists to use this material to address the problem. The use of ceramic patterning to examine the extent, or nature, of imperial influence in provinces, has been used in pre-Roman Levantine studies, and in Meso-America. The persistence of cultural regions in the southern Levant since the Neolithic has now been documented.18 Allow- ing for the effects of various pre-Roman imperial authorities, ranging from Egyptians to Assyrians, these cultural regions can be taken as the natural trading zones of the southern Levant, against which the results of the project presented here can be evaluated. Work on the distribution of distinctive pottery in pre- and Imperial Aztec polities is particularly relevant from a methodological perspec- tive.19 There are many striking parallels between the anthropology and archaeology of Meso-America and the Roman Empire: in both cases written documentation is heavily biased towards elite classes, and cities rather than rural areas. Traditional studies have followed the written material, and only in recent decades has the larger, undocumented world of the uninfluential population been examined. In this respect, the vastly greater documentary evidence from the Roman empire, which includes, for instance, personal letters, epigraphy and sermons, has led to extensive examination of the non-elite, well before such issues have been raised in America. On the other hand, possibly because scholars of the Aztec empire have been employed in anthropology rather than Classics depart- ments, and because the written sources are so limited in central America,
17 K. da Costa ‘Byzantine and early Islamic lamps: typology and distribution’, in P. Watson—E. Villeneuve (eds.), La céramique byzantine et proto-islamique en Syrie- Jordanie (IVe–VIIIe siècles apr. J.-C.) (Beirut ), ff.; B.D. Shaw, Rulers, Nomads, and Christians in Roman North Africa (Aldershot ); L. de Ligt, Fairs and Markets in the Roman Empire (Amsterdam ); D.P.S. Peacock, Pottery in the Roman World: an Ethnoarchaeological Approach (London ); H. Howard—E.L. Morris (eds.), Produc- tion and Distribution: a Ceramic Viewpoint (Oxford ). 18 S.K. Kozlowski—O. Aurenche, Territories, Boundaries and Cultures in the Neolithic Near East (Oxford-Lyon ); P. Bienkowski ‘The north-south divide in ancient Jordan: ceramics, regionalism and routes’,in T. Potts—M. Roaf—D. Stern (eds.), Culture through Objects: Ancient Near Eastern Studies in Honour of P.R.S. Moorey (Oxford ), ff. 19 M. Hodge—L. Minc, ‘The spatial patterning of Aztec ceramics: implications for PreHispanic exchange systems in the Valley of Mexico’, Journal of Field Archaeology (), ff.
K. Da Costa - 9789004215030 Downloaded from Brill.com09/28/2021 09:09:39AM via free access a methodology for detecting roman provincial borders there is a stronger emphasis on, and use of, archaeological materials and methodology in that field of scholarship.20 Hodge and Minc studied the distribution patterns of selected high- quality ceramics made in limited centres and marketed widely but never in vast quantities. They used the collected material of earlier surveys, and collated data from sites. Their definitions of polities depended on written sources although they stressed that “in the future, archae- ological data gathered for the purpose of detecting city-state political boundaries could be used to evaluate the ethnohistorical accounts of the extent of political territories”.21 Their results revealed separate distri- bution patterns of distinctive ceramics in the Early Aztec period, when autonomous city-states belonged to two main confederacies, with very limited exchange between the larger units. In the Later Aztec Imperial period, the patterning changed to a much more homogenous market- ing system, although some regional differences remained. Although the results are different from the archaeological data described below, they do indicate that material culture patterns can be related to political, as well as social structures, depending on the artifact class and level of quantifi- cation studied. In the southern Levant, patterns of differing classes of ceramics seem to be showing an equally uneven distribution. The substantial production of cooking wares from Kefar Hananya is not known outside Judaea/Pal- aestina.22 Distribution patterns of Golan ceramics drop significantly on the eastern edge of the region, although there is no major topographic barrier.23 Ceramic lamps produced during the Byzantine period, from the rd to the early th centuries, seem to be restricted to either Palaestina or Arabia.24 Watson has shown that bulk importation of ceramics into Pella, in Palaestina, from the important production centre of Jarash, in Arabia, did not occur between the rd and th centuries, although by the early th century, most of Pella’s ceramic supply was from Jarash.25
20 These admittedly sweeping assertions may not be completely true of work on the Roman empire in the western parts of Europe, especially Britain, but it does seem that study of the eastern Roman empire is still in thrall to written history. 21 Hodge—Minc , op. cit. (n. ), , note . 22 D. Adan-Bayewitz, Common Pottery in Roman Galilee: A Study of Local Trade (Tel Aviv ). 23 M. Hartal, The Material Culture of the Northern Golan in the Hellenistic, Roman and Byzantine Periods (Jerusalem; unpublished PhD ). 24 K. da Costa, ‘Economic cycles in the Byzantine Levant: the evidence from lamps at Pella in Jordan’, Levant / (), ff. 25 P. Watson, ‘Change in foreign and regional economic links with Pella in the seventh
K. Da Costa - 9789004215030 Downloaded from Brill.com09/28/2021 09:09:39AM via free access k. da costa
The cities were relatively close and linked by a major Roman road. Some other factor, before the late th century, acted as a barrier to local trade. The patterns are best explained not by topographic features nor by simple distance from production centres. They seem bounded by the approximate line of provincial borders, in the few places where these can be reasonably reconstructed. The most likely explanation is the imposition on major provincial borders of a customs duty.26 Our knowledge of the collection of indirect taxes is rather patchy.27 Of these, customs duty, portorium,waslevied— at . or —on the Imperial frontiers, and also within the Empire; the rate is not certain, but probably .– .28 Our information, while heavily biased towards Egypt and the early Empire, shows that taxes, tolls and levies had a conspicuous effect on small-scale economics and local trade. It seems clear that the customs duty on major borders, such as Arabia/Palaestina, but not internal borders (between the three Palaestinas), remained in place until the late th century. By making it uneconomical to import local ceramics from neighbouring provinces, the duty distorted trade patterns. This distortion can be harnessed to map the location of the unknown sections of the provincial borders. The Borders of Arabia and Palaestina (BAP) project, a case-study in an area overlapping part of the border between Palaestina Secunda and Ara- bia, is developing an archaeological methodology to allow a more pre- cise definition of provincial territory based on this distortion to ceramic trade.29 Provincial borders in this part of the Empire are relatively well known, although the entire south-east corner of Palaestina Secunda’s century ad: the ceramic evidence’, in P. Canivet—J.-P. Rey-Coquais (eds.), La Syrie de Byzance a l’Islam VIIe–VIIIe siecles (Damascus ), ff. 26 M. Cottier et al. (eds.), The Customs Law of Asia (Oxford ). 27 R. Delmaire, Largesses sacrées et res privata: L’aerarium impérial et son administra- tion du IVe au VIe siècle (Rome ); I.W.J. Hopkins, ‘The city region in Roman Pales- tine’, Palestine Exploration Quarterly (), ff.; W. Goffart, Caput and Colonate: Towards a History of Late Roman Taxation (Toronto ); A.H.M. Jones—P.A. Brunt (eds.), The Roman Economy. Studies in Ancient Economic and Administrative History (Oxford ). 28 Probably ., according to A.H.M. Jones, The Later Roman Empire –: a Social, Economic and Administrative survey (Oxford ), ; ; P.J. Sijpestejin, Customs Duties in Greco-Roman Egypt (Zutphen ); De Laet, in the major study of portorium, was unable to comment on customs duty after Diocletian, due to a lack of written evidence: S.J. de Laet, Portorium: étude sur l’organisation douanière chez les Romains, surtout à l’époque du Haut-Empire (New York ). 29 K. da Costa, ‘Roman provincial borders across Jordan’, Studies in the History and Archaeology of Jordan (), ff.
K. Da Costa - 9789004215030 Downloaded from Brill.com09/28/2021 09:09:39AM via free access a methodology for detecting roman provincial borders
Figure border is undocumented. The aim of the project was to collect ceramics of the rd to th centuries from sites in the area of the supposed border route (figure ). The overall corpus from each site will be categorized by reference to the known corpora from Pella (Palestinian)andJarash(Ara- bian). The border must lie between the Palestinian and Arabian sites. Sites selected for sampling had previously been identified in earlier surveys of north-west Jordan, although in most cases little or no pot- tery had been published. From the nearly sites of the Roman to early Islamic period documented in the region, settlements which appeared to be small towns were the priority, as these would be expected to con- tain the largest range of ceramics. Two field seasons, and , have been undertaken, with the aim of collecting at least , sherds of the Byzantine (rd – early th centuries) from each site. However, of the twenty sites sampled, only around ten will produce reliable statistics because of collection difficulties. There were fewer sites in the eastern half of the sampling area, and many of these had significant modern or mediaeval occupation over the entire area. There were also fewer Byzan- tine and Roman sites in the eastern part of the study area than the west, although more mediaeval material—a settlement pattern change which is
K. Da Costa - 9789004215030 Downloaded from Brill.com09/28/2021 09:09:39AM via free access k. da costa intriguing in itself. The focus of processing to date has been on the coarse ware body sherds—exactly those ceramics normally unsampled or discarded in conventional survey. It is this which distinguishes the project from the methodology used, for instance, by Hodge and Minc to discuss market types and integration in the Aztec Empire. The . tax on provincial borders in the Roman Empire seems to have had little or no inhibiting effect on luxury or expensive goods, including ceramic fine wares such as African, Cypriot or Phocaean Red Slip products. It is only at the level of bulk trade in low-profit common wares that the distorting effect seems to appear. Processing of the material still continues, and only preliminary results are presented. One reason coarse (or common wares) are not prioritized in conven- tional survey or sampling is the tremendous difficulty of close dating, particularly when corrective data from excavations is unavailable. The BAP project therefore has been using very broad date ranges, and there are clearly potential problems for interpretation, given the known his- tory of border changes at more frequent intervals than we may be able to detect ceramically. Leakage of ceramics across the border has also been anticipated, particularly since we have sampled sites quite close to the hypothetical border line, and it is quite feasible that small quantities of material crossed over. However, the identification of a corpus as Palaes- tinian or Arabian will depend on general ratios of wares across the entire sample, rather than the presence of a few distinctive pieces. As so few sites in the case study area have been excavated or published, our treat- ment of the coarse wares must remain very general. However, it appears from the initial results that the methodology is able to indicate differ- ences in corpora, and these correspond to the presumed provincial allo- cations of each site. Given that the same circumstances exist across the Empire—abundant coarse ware ceramics and a customs duty on major borders—this methodology should be applicable elsewhere in order to more precisely define the line of a provincial boundary. Only those sites where over a thousand sherds of the Byzantine period have been catalogued have been included in these preliminary results. Coarse ware sherds have been divided into handmade and wheelmade categories, into ribbed and unribbed if wheelmade, and in each case, by thickness, more or less than mm. With the same processing protocol for each site, and with the largest possible quantities collected in the time available, we believe that minor fluctuations in cataloguing will be evened out. Bodysherds in each category have been counted and weighed. Results so far have been collated at a very general level of ware
K. Da Costa - 9789004215030 Downloaded from Brill.com09/28/2021 09:09:39AM via free access a methodology for detecting roman provincial borders
Figure definition. The final processing of diagnostic sherds will help in some way to gauge the bodysherds, although it is not possible in many cases to tell if rim sherds come from ribbed or unribbed (or ribbed and unribbed) vessels. Figuredemonstratesthatcountsandweightsofsherdsdonotnec- essarily provide the same results—Kh. Sittat had large numbers of small sherds, whereas Maqati" and Ba"un had large numbers of large sherds. These results will help characterize each site, and are used to normalize results. Plots of part of the ceramic corpus at each site, showing the main groups of coarse ware sherds from storage and simple table vessels (al- though no plates and few cups have so far been recovered) and the dis- tinctive Late Roman (LR) Palestinian Bag Shaped Amphora (Brown Slipped, White Painted BSWP), are graphed in figure . Sites have been grouped based on supposed provincial affiliation: Nasar to Dohaleh are thought to be in Palaestina;Ba"un to Tor Hanna should be in Arabia. Of the eleven sites plotted, seven have similar profiles—the most common wares are the orange terracottas, followed by brown, pale and grey. Nasar has extraordinary quantities of the LR amphora in BSWP ware and it may very well be that some of the grey wares catalogued at that site are also BSWP sherds. Nasar is one of the closest settlements on the main road to the border crossing from Palaestina to Arabia. All the supposed Arabian sites have very low
K. Da Costa - 9789004215030 Downloaded from Brill.com09/28/2021 09:09:39AM via free access k. da costa
Figure quantities of this ware, exactly as we would expect given that its distri- bution should fall off over distance, but might be expected to cross the border in small numbers, given that it was the contents of the vessel that were traded, rather than the amphorae themselves. This would explain the quantities at Ba"un, the first major settlement in Arabia across the Palaestinian border along the major trade road to Jarash. Nonetheless, the fact that LR is a Palaestinian amphora is emphasized by the quanti- ties at Fara and Dohaleh, both as far from Pella as Sittat and Maqati". Apart from Nasar’s corpus, all the other Palaestinian sites have quite similar profiles, although the overall numbers are much lower at Mah- rama, a site with significant Mamluke upper levels. In contrast, there does not seem to be any consistency amongst the Arabian sites—Ba"un has no pale or grey pottery to speak of, Maqati" has large amounts of pale terracotta wares, Sittat is dominated by brown wares, and Abde and Tor Hanna have profiles similar to Palaestinian sites, with the notable exception of low quantities of BSWP LR amphorae. However, if the proportions of the two main ware groups, orange and brown terracottas, is compared, the pattern is clear. Figure shows that Palaestinian sites have an average ratio of orange to brown wares of ., and the lowest ratio is . at Dohaleh, another site with large amounts of Islamic material. In contrast, Arabian sites have an average ratio of ., and two sites have a ratio of less than . The position of Dohaleh is equivocal. The site is one of the more easterly of the case study area, in a region where the line of the border
K. Da Costa - 9789004215030 Downloaded from Brill.com09/28/2021 09:09:39AM via free access a methodology for detecting roman provincial borders
Figure is least well known. The site has been excavated by Saleh Sari, and some of the pottery published.30 There are reasonable quantities of LR BSWP amphorae sherds at the site, the range of lamps from Dohaleh is similar to that at Pella,31 and according to Avi-Yonah, the border lay to its east.32 Certainly, the milestones of the road from Jarash to Dera"a, which lay entirely in Arabia, were east of Dohaleh, and if the border ran along river beds, there are several wadis to the south and east of Dohaleh and Ya"amun which would serve as boundaries, as indicated in figure . On the other hand, churches at Yasileh and el-Husn, to the north of Dohaleh, excavated in the s, have inscriptions said to be dated by the Arabian era. Husn was a dependent village of Irbid (ancient Arbela), and so it seems likely that the border was much further west than Avi-Yonah suggested. It may be that the major north/south watershed of northern Jordan was the border, and that Dohaleh lay in Arabia. Recalculating the averages of the ratios of orange to brown wares with
30 S. Sari, ‘Preliminary report on the results of the excavations at Kh.Dohaleh-al- Nu"aymeh, st season, Summer ’, Annual of the Department of Antiquities of Jordan (), ff. (Arabic); S. Sari, ‘Dohaleh, a new site in northern Jordan. First season of Excavations, ’, Liber Annuus (), ff. Prof. Sari kindly gave permission for the BAP team to sample Dohaleh—sampling squares were placed away from the Yarmouk University trenches. 31 da Costa , op. cit. (n. ), . 32 M. Avi-Yonah, The Holy Land from the Persian to the Arab Conquest (bc–ad) AHistoricalGeography(Michigan , rev. ed.), , maps & : the list of sites in the text does not match either of the maps, both of which are substantially different from each other—Avi-Yonah never addressed this discrepancy.
K. Da Costa - 9789004215030 Downloaded from Brill.com09/28/2021 09:09:39AM via free access k. da costa
Figure this configuration, as shown in figure , increases the average orange: brown ware ratio of Palaestinian sites to ., decreases that of Arabian to ., and shows Dohaleh fitting neatly into an Arabian pattern. As Dohaleh is the only site sampled by the BAP project in this conjectural zone with usable quantities of Byzantine ceramics, the question of the border line must rest until full processing of the site is complete. However, even these preliminary and incomplete results indicate the value in the time-consuming process of intensive cataloguing of body sherds, and the potential data which can be obtained from them.
Durham, May
K. Da Costa - 9789004215030 Downloaded from Brill.com09/28/2021 09:09:39AM via free access