This article summarizes results from the project LEGECARTAS (Lectures géoarchaéologiques des cartes anciennes) of the CNRS (2017-2019), the subproject The Thracian harbour city Ainos of the DFG SPP 1630 Harbours – from the Roman Period to the Middle Ages (2011-2017), as well as the annual archaeological excavations supported by the Turkish Ministry of Culture, the Edirne administration and Museum, and the Istanbul University. We acknowledge the support by Thomas Schmidts, Mainz, and Martin Seeliger, Frankfurt, PI and staff member, respectively, in the SPP 1630 Harbours.

1. An Old Insular Setting between the Hebros and the Aegean

3The land resources, together with the abundant seafood, have attracted people since the Neolithic. The oldest site is Hoca Çeşme, 2.5 km E of Enez on a 35 m high and 150 m large plateau, whose name recalls its freshwater springs. This fortified settlement, which dates to the 7^th-5^th millennia BC, was unearthed by Sait Başaran and Mehmet Özdoğan from 1988 until 1993 (Fig. 6; Özdoğan 1996: 336-337; Başgelen, Özdoğan 1999: 217-220; Özdoğan 2000). During the first phase of habitation, the ramparts were built of stones of various sizes; they are still preserved on a length of 55 m. They protected round huts with rock foundations and walls of adobe and woven branches. The earliest Neolithic ceramics indicate connections with Anatolia, while later layers contained materials related with the Balkan Sésklo and Karanovo cultures. We may suppose that this sedentary community took advantage of the nearby sea and rivers, but we are still ignorant about the reason why they did not occupy the Enez promontory itself. The earliest sherds discovered by Sait Başaran in Enez, under the Byzantine/Genoese/Ottoman castle (“Acropolis”), date back to Chalcolithic times only (4^th millennium BC), like on several other neighboring sites. It is, however, impossible to say if there was continuity or, on the opposite, discontinuity of indigenous occupation until the arrival of the Greeks. In fact, so-called “Thracian” gray ceramics were found in the lowest layers of the “Acropolis”; but as yet, no structure has been identified as belonging to a Thracian settlement preceding the Aeolian city and no precise chronology could be established for these phases of habitation (according to A. Erzen in Naumann et al. 1983: 241, and in Naumann et al. 1984: 212-213; cf. Baralis 2016: 32-33). This apparent chronological gap raises questions about the value of the Greek historical traditions referring to one or two Thracian cities that would have preceded the Aeolian foundation.

4The hypothesis that we are currently testing by geophysical surveys and geoarchaeological corings in order to explain the weak indigenous presence on the peninsula is that during the postglacial rapid marine transgression, Ainos had turned into an “estuarine island”, which only later on became landlocked by a tombolo (isthmus). The isthmus allowed an easy connection with the hinterland, offering the Greeks a settlement site which corresponded to their maritime needs – as it has been understood since the 19^th century (e.g. Slade 1833: II.383-384). Yet, because of its flat and low topography, the isthmus remained swampy, as shown by the current name of its NE shore, “water terrace” (Su Terazisi), and by the numerous ancient and modern drainage canals on both sides, toward the brackish Taşaltı lagoon and Lake Stentoris (Schwardt et al. forthcoming, Fig. 7-8). In fact, this insular nature has proven to be most favorable for Ainos: in the 1950s, the isthmus was artificially cut in order to deviate part of the Hebros waters into the Taşaltı Gölü and the Dalyan Gölü. This stopped inundations and refreshed the water of the lagoons.

5According to the preliminary results of the palynologic studies of Lyudmila Shumilovskikh, when the sea surrounded Ainos, the impact of the Neolithic and Chalcolithic communities on their environment remained limited. Coring AIN 50 (40°43’12.72” N, 26°05’19.48” E, 0.17 m a.s.l., Fig. 9) on the presently amphibious shore of the Taşaltı lagoon indicates the presence of a climax vegetation with open deciduous oak woods, pine and hazel in the 7^th-6^th millennia BC. These natural open oak woods are different from the pine forests of the last 600 years. This vegetation change is probably due to the intensive anthropogenic impact of wood clearing for agriculture and pasture. The resulting strong soil erosion caused rapid delta progradation, siltation of the lagoons and harbors, and finally, the cessation of river navigation (still possible with maritime ships until the second half of the 19^th c.: Admiralty 1917: 21; Slade: II.378; Hasluck 1908-1909: 249-250; cf. De Boer 2010).

6Two important environmental changes occurred during Antiquity: between ca. 5000 and 2800 BC (i.e., in Chalcolithic times), mixed deciduous oak woods with elm, lime tree, hornbeam, and beech became dominant, either as an effect of climate change or early anthropogenic clearings: the last remains of these wild oak and elm forests were mentioned by modern travelers in the Hebros valley up to the 19^th c. (e.g. Keppel 1831: I.252-253). However, this already altered ecological environment was strongly modified by the settlers of the Greco-Roman times. From the 7^th c. BC to the 3^rd c. AD, intensive deforestation, iron-plough agriculture, goat herding and charcoal production contributed to the degradation of the vegetation to macchia and even phrygana. This points to the presence of larger communities, who used the land and water resources not only for local needs but also for large-scale trade, including the use of timber for ships and sea salt for fish conservation. The dramatic ecological change is a good illustration of the major environmental impact of the Greeks and later the Romans, who exploited the best natural sites in their Mediterranean networks.

2. Sea, River, and Land Connectivity: Streets, Necropoleis, and Harbors

7Ainos was the city of Hermes Perpheraios (Call. Iamb 7, cf. Bousquet 1948). The famous inscription attesting the existence of naukleroi (Miller 1873; Dumont 1892: 437 nr. 103), the coins (Tekin 2007, Baker 2013) and the various ceramics (Lätzer-Lasar 2016) discovered in Ainos recall the city’s status as a hub between the Aegean and the North, through the Hebros to the hinterland, as drawn on the Tabula Peutingeriana (Fig. 10). The archetype of this compilation of Roman itineraria picta probably goes back to the mid-4^th c. AD, but the road network shown in Thrace was constructed on previous paths by the Romans between the 1^st c. BC and 2^nd c. AD. In the system developed around the Via Egnatia, Ainos appears as a nodal point of three roads, in the middle of the stations of Dymis/Feres (?) to the N (across the Hebros), Plotinopolis/Didymoteicho and Hadrianopolis/Edirne to the NNE (up the Hebros), and Zorlanis/Keşan to the NE (to Byzantium/Istanbul). Sait Başaran discovered several segments of ancient paved roads outside the city. The first is at the E extremity of the isthmus connecting the peninsula to the mainland, on the N shore of the Taşaltı lagoon. The second and third segments are on the SE shore of Lake Stentoris, at Gala Gölü Milli Parkı, and near Çeflik Köyü. One of these two points could correspond to Pliny the Elder’s Stentoris harbor, a station from which one could cross the river to Pheres, Doriskos, or Traianopolis (Başaran 1999; cf. Külzer 2008: 192-204; Külzer 2011). Further geophysical prospections and geomorphological corings are, however, necessary in order to reconstruct the puzzle of the S branch(es) of the Via Egnatia, determine the former natural environment, and establish the chronology of the road building.

8To date, the city’s large and wealthy necropoleis are the best indicator for the trajectory of the exterior roads and, more generally, for Ainos’ vast economic networks. Excavated by Sait Başaran since 1982, three main funerary zones of Greek and Roman times extend from the S and E shores of the promontory along the isthmus and the modern way to Keşan. The earliest Greek materials, unearthed from the Su Terazisi necropolis (on the isthmus, Fig. 7-8), are W Anatolian (Orientalising), Corinthian, and Attic vases, accompanied by local imitations. They show that by the end of the 7^th c. BC and during the 6^th c. BC, Ainos’ trade networks with the E, S, and N Aegean were already well established. A change occurs by the beginning of the 5^th c. BC, when the N Ionian and Aeolian importations are replaced by the Athenian black- and red-figure cups, amphoras, hydrias, and lekythoi. Yet the Micrasiatic connection persists until Roman times, as attested by marble reliefs illustrating the funerary banquet of the deceased with his familia (in the Edirne Museum).

9The Ainians not only bought but also adapted the imports in their own way, sometimes even without parallels in the rest of Thrace. A good example is the so-called Clazomenian sarcophagi of the 6^th c. BC, imported throughout the N Aegean up to Chalkidiki and Abdera. Terracotta sarcophagi remain one of the possible burial choices in Thrace up to the Black Sea, throughout Classical and Hellenistic times. However, in Ainos, the quantity and quality of their polychromic painted decoration are unique. Also, while bronze hydrias were often used as incineration urns in the 5^th c. and especially in the 4^th c. BC in central Greece and throughout Thrace up to the Black Sea, in Ainos they are buried in small stone sarcophagi, similar to those in which they would have been casted (Fig. 11).

10The Su Terazisi necropolis has been continuously used from the first generations of ἄποικοι up to late Roman times, for both inhumation and cremation, in all kinds of graves – from the simple deposition of the body into the ground or into a stone frame, to reinhumation or incineration and deposition of the remains in amphoras, hydrias, pelykai, pithoi, terracotta and stone sarcophagi, without any inventory or with smaller vases, jewelry, and coins. Goods of different origins are usually combined in the same grave. This mixture of funerary rituals and inventories is also attested in another necropolis, at the NW end of the isthmus on the slopes of the promontory dominating the Taşaltı lagoon. On several terraces, graves of different forms have been carved from the late Archaic-Classical to the Byzantine times (Fig. 12). We did not find the ancient fortification closing the urban space in this zone. Therefore, we do not know the precise location of the city gate for travelers coming by land from the isthmus. However, the lines of tombs and the presence of an exedra suggest the existence of a path along the edge of the city through the necropoleis.

11Ainos’ necropoleis extend from the SE slopes of the promontory on the isthmus, along the road to the E, at least for 2 km until Çakıllık, where Sait Başaran found inhumation and incineration graves from the 5^th-4^th c. BC. Ainos was by then Athens’ ally in the Delian League and later in the Second Attic League. Two discoveries point to the richness of some Ainians at that time: the quantity and quality of bronze hydrias used as incineration urns, and the discovery of a marble funerary lion (Fig. 13).

12Further archaeological and geoarchaeological prospections are needed in order to map all the funerary sites around Ainos and evaluate the demographic and economic fluctuations in time. The only conclusion that we can as yet draw is that the necropoleis confirm the prosperity of the ancient and Byzantine city as well as the importance of the isthmus relating it to the mainland. The presence of several tumuli illustrate the symbolic value of the isthmus, as probably the only terrestrial connection of Ainos with the mainland. The most important is the mysterious tumulus assigned since the 1^st c. AD to the legendary Trojan prince Polydoros (Plin. NH 4.43) and observed by the modern travelers since the 15^th c. (Bertrandon de la Broquière, in Schefer 1892: 173-174; Cyriacus of Ancona, in Bodnar 2003: Diary II, 104-106; see also Fig. 2). Nonetheless, until now, no funerary chamber could be found by excavations, non-invasive geophysical methods, or coring attempts.

13Today, there is a second road on the sand bar between the Taşaltı Gölü and the Dalyan Gölü. As yet, we do not know if a bridge already existed in ancient and Byzantine times or if this NS passage at the connecting shores of the two lagoons was only possible by (flat) boats. The oldest walls that we could observe under the 20^th c. road, and the wood that we could recover from this structure and date by ¹⁴C are from the Ottoman times (16^th-17^th c.). This matches the Byzantine archaeological evidence for the opening of the Taşaltı lagoon to flat-bottomed boats, by which construction materials were brought for the so-called Kral Kızı basilica. This also corresponds to the modern maps, showing the opening of the lagoons until the 19^th c. and the existence of a Byzantine and Ottoman road following the seacoast, passing the Gümrük Kervanseray (18^th c.) toward Gallipoli (Fig. 2).

14The geomorphological corings revealed a marine / lagoonal influence in the Taşaltı lagoon during the 3^rd c. BC and beyond. Considering the age / depth model and the sea level evolution, it can be assumed that at coring site AIN 5 (= AIN 50, 40°43’12.72” N, 26°05’19.48” E, 0.17 m a.s.l.), the water depth was at least ca. 1.5 m before the turn of the eras, and ca. 1-0.8 m during medieval times (Fig. 8, 14). Thus, throughout Antiquity and medieval times, it would have been possible to cross this lagoon with relatively flat boats. Roman constructions and tombs have been observed on the S shore of the Dalyan Gölü, up to the natural salines used until the 20^th c. But they do not have the richness of the contemporary graves discovered on the slopes of the city’s promontory, and could be assigned to the population whose existence was directly related to the exploitation of the lagoons for fish and salt.

15The abundance and variety of finds in these necropoleis raise the question as to the major harbors through which important quantities of foreign products arrived in Ainos (cf. Schmidts, Vučetić 2015: Brückner et al. 2015; Rabbel et al. 2015; Brückner, Schmidts 2019). At least one harbor has been known from Antiquity to Ottoman times (Fig. 3-4). Its natural bay, at the W extremity of the promontory, presented several advantages: as Ainos’ shores were mostly marine in Antiquity, the position of this harbor was precisely at the crossing of the Aegean routes to the SW (Samothrace), NW (the Samothracian peraia, Thasos), SE (Chersonese), and NE (toward the Hebros). The maritime gate of the medieval fortification of the “Acropolis” is opened to the NW of this protected harbor, or in other words, toward the cape between the open sea and lake Stentoris. The site is close to the “Acropolis” and the littoral plain where an agora could have been installed (W of the so-called “Pan Cave”). The “Acropolis” mount offered the best protection against the strong N-NE winds. This potential harbor area is still easily identifiable today, at the foot of the N gate of the castle, between the two series of towers that were joined by fortification walls in medieval times, but seem to have been erected, at least in part, on the ancient (Hellenistic?) foundations. The medieval configuration of the wall, however, fits the literary texts that indicate the vulnerability of the city toward the sea and the importance of the walls down the “Acropolis” (Procop., Aed. 4.11; Agath. 5.22 p. 192 Keydell). Further geoarchaeological prospections and excavations are necessary in order to determine the shape and chronology of the harbor walls and check their connection with a mole, in case a mole existed in ancient or medieval times.

16The geoarchaeological research of the last few years has taught us that the inner area between the two walls was already silted up during Classical Antiquity. Marine strata have been discovered at 4.85-2.75 m below the present sea level (b.s.l.), but they date back to the 3^rd millennium BC; therefore, the area of the coring site was already silted up when the Greek settlers arrived. The underwater topography of the presumed harbor basin progressively sloped down W from the shoreline marked by the most NW tower (“A” for Hasluck 1908-1909). Near the tower, in the corings AIN 115 and 131 (40°43’24.68133” N, 26°04’41.73392” E, 0.586 m a.s.l.), marine and lagoonal strata at ca. 5-2.5 m b.s.l. date from the 5^th millennium BC. At the arrival of the Greeks, the water level, determined after our preliminary estimation of the former sea level, was about -1 to -1.5 m. Therefore, as much as we can say before carrying out further research, the ancient and medieval foundation of this tower may have been nearshore. The harbor walls, at least in their medieval form, mainly protected the landing site, accessible by flat-bottomed boats or by foot. In September 2019, the core AIN 146 and two geoelectric profiles allowed us to locate the ancient and medieval harbor basin in front of this landing space. However, the magnetic and seismic prospections further to the W, in the lagoon, have not revealed any underwater structure. If a mole ever existed, several explanations are possible: the current techniques cannot reveal structures due to the methane gas formation in today’s marshy environment; a wooden mole has since rotten away; or the mole was situated at another location.

17During the last millennia, eroded sediments from the cliffs of Cape Sarpedon have been transported N by the coastal currents and re-deposited first as sand spits and later as sandbars, which eventually closed the ancient sea gulfs, thus forming Dalyan Gölü, and caused the abandonment of the main harbor in the 18^th c. As its modern name indicates, the Dalyan Gölü became a fishery. At the same time, the progradation of the Hebros delta required the installation of a river harbor to the N of the Dalyan sandbar at the place still known as İskele. The seagoing vessels had to anchor in the sea, while flat-bottomed riverboats could sail through Lake Stentoris up the Hebros until the 20^th c., when alluviation and intensified irrigation lowered the water level and ultimately blocked navigation on the S mouth of the river (Dumont 1892: 204-205; Hasluck 1908-1909: 249-250).

18During Antiquity and medieval times, however, when the Hebros mouths were still far to the N, somewhere between Ipsala and Gala Gölü, the N shore of the promontory and isthmus formed a very large bay with several indentations, where maritime and riverine ships could be anchored (Fig. 3, 5). We have two types of evidence for assuming that these inlets were possible anchoring points (during seasons when the N winds would allow it): first, several niches sculpted into the rock close to the former N beaches. The votive objects put in these niches could have been related to deities that the seamen of the region took as protectors for their journeys: among them, there was the Thracian hero, represented on several unpublished reliefs now in the Edirne Museum, or the Hebros river-god and the local nymphs on a relief now studied by Dan and Başaran (forthcoming). Second, the coring site AIN 54, N of the “Acropolis” (40°43’35.64208” N, 26°04’34.86147” E, 0.37m a.s.l.), had a water depth of at least 4 m at the beginning of the 1^st millennium BC. From Roman Imperial times until the 17^th / 18^th c., deposits of dark gray sands and black silts, at 3-1.8 m b.s.l., point to a low-energy environment, possibly of a sheltered embayment (harbor?). It is reasonable to assume that this area first served as a landing site for seagoing vessels, and – after the Hebros delta had passed by – as a river harbor. Comparable is the situation at AIN 82 (40°43’42.97” N, 26°04’56.41” E, ca. 0.5 m a.s.l.): strata with marine fauna are attested from ca. 6.8 m until 4 m b.s.l., dating from the early 4^th millennium until the turn of the eras. Lagoonal facies developed in Roman times, showing that the area may first have been a marine, later a fluvial landing site. Further to the E, the geoelectric profiles point to an ancient coastline, now covered by alluvium. In front of it, at AIN 23 (40°43’41.4” N, 26°05’39.6” E, 0.25 m a.s.l.), the age/depth model shows that the water depth was several meters in Greek times. The transition from open marine to lagoonal conditions occurred during Roman Imperial times; during Byzantine times, the lagoon silted up.

19Thus, from Antiquity to medieval times, the Ainos promontory was surrounded by suitable anchoring sites: flat-bottomed boats could anchor in the natural harbor of the Taşaltı lagoon while large seagoing ships anchored on the W and possibly N shores. However, definite harbor moles have not yet been detected by geophysical measurements, corings, or archaeological surveys.

3. A 2600-Year-Old City

20The main result of the geophysical measurements in Ainos is the discovery of a SW portion of the ancient city wall, on the E shore of the Dalyan Gölü (Fig. 15). The “zig-zag” plan of its foundations suggests a Hellenistic date, which still needs to be confirmed by archaeological excavation (Seeliger et al. 2018). The conditions are difficult because of the continuous and intense occupation of the site. Therefore, we know almost nothing about the ancient urban topography outside the “Acropolis”. Rescue excavations in the 1980s revealed a segment of the Roman paved street, covering public water pipes, and a house with frescoes and mosaics, abandoned in the 3^rd c. AD, maybe after an earthquake. The date is confirmed by the ca. 160 silver and bronze coins spread on the floor.

21From Justinian’s time onward, the city walls protected only the “Acropolis”, where the Byzantines, the Genovese family of the Gatellusi, and lastly the Ottomans had the political and religious center of their city (Hasluck 1908-1909; Wright 2014). The best-preserved Greek layer is a large cellar carved into the rock, where Sait Başaran discovered Classical amphorae, drinking vessels, and the terracotta head of a satyr. The Ainians’ passion for wine must have been famous, since in the 3^rd c. BC, Callimachus presented it as the cause of death of his friend Menekrates, compared to a Centaur (Epigram 61).

22Another match between the literary references and the discoveries on the “Acropolis” concerns Apollo: the god is adapted to the commercial profile of Ainos, because he “oversees the village” (he is Apollo “Epikomaios” in Thphr. fr. 97.3 Wimmer ap. Stob., Florilegium 4.2.20) and gives advice to the fishermen to accept the statue and cult of Hermes (Call., Iamb 7). Unpublished fragments of Hellenistic and Roman sculptures of the god suggest the proximity of his temple, as one expects in an Aeolian city. In fact, the Hellenistic and Roman structures under the Fatih mosque – the so-called “Hagia Sophia” church, probably a Saint Constantine, if not a Virgin Mary church (Ousterhout 1985) – could belong to a temple: but nowadays it is impossible to make any speculation about its tutelary divinity. Inscriptions reused in Byzantine walls also indicate the existence of a temple of Zeus as well as of Rome and of the emperor. The publication of the epigraphic corpus by Mustafa Sayar will certainly offer a better picture of Ainos’ political and religious landscape.

23In conclusion, Ainos is a good example for the study of an average Aeolian, Roman, Byzantine, Genovese, and finally Ottoman city, benefiting from its strategic position and exceptional natural resources. Even if the continuity of urban life from Antiquity until the present day destroyed most of the ancient layers, archaeologists are still able to observe the networks of the Greek and Roman city by studying its rich necropoleis. Modern geoarchaeological research, based on non- or little-invasive methods, provide data on the environmental history of the site. The story that we hope to write is one of natural and anthropic changes of Ainos’ connection points with Thrace and the Aegean, mainly through the lagoons that served as potential harboring sites, fisheries, and salines. The result would not be just another reflection on the interdependency between landscape and humans, but also a lesson of economic and cultural prosperity; the current city could thus learn from its past.