The Harbour of Naukratis, 'Mistress of Ships' The British Museum Naukratis Project s third fieldwork season at Kom Ge if, Egypt (Beheira MSA site no.100253) April-May 2014 Ross Thomas and Alexandra Villing, with Ben Pennington, Kristian Strutt, Aurélia Masson and Astrid Lindenlauf The port of Naukratis was the earliest, and for a period the only, Greek port in Egypt, functioning as the main Mediterranean port of Egypt during the 7 th to 4 th centuries BC. Established in the late 7 th century BC as a base for Greek (and Cypriot) traders and the port of the royal Pharaonic city of Sais, it was an important hub for trade and cross-cultural exchange in the ancient world long before the foundation of Alexandria. Previous fieldwork by Petrie and Gardner (1884-1886), Hogarth (1899-1903) and Coulson and Leonard (1970s-1980s) concentrated on the central areas of the town, and further research is required to understand this key archaeological site. Our previous two seasons (October 2012, April 2013) encompassed geophysical prospection, the creation of a topographic survey map using GPS technology, surface pottery collection, excavation and geological work with a hand auger (Thomas and Villing 2013). In the third season, a grant from the Honor Frost Foundation enabled our team to dedicate resources specifically towards improving our understanding of the site s harbour, its relation to the river and canal networks, and the infrastructure that enabled Naukratis to operate effectively as a major entrepôt of the ancient world. Bizarrely, despite the fact that the significance and prosperity of Naukratis derived entirely from its maritime role, this aspect had not previously been explored. Our aims for this discrete aspect of work were therefore to investigate the course and extent of the Canopic branch of the Nile and the location and nature of harbour, canal and quay facilities (including the ritual quay for the great Egyptian temple of Amun Ra); to gain a better understanding of their development from the site s foundation to its decline in the 7th century AD (the later history of the site previously having been much neglected: Thomas 2014); and to assess the potential for more comprehensive future investigations of the port of Naukratis and the Western Nile Delta port system. Fieldwork is set alongside a study of maritime aspects of and artefacts from earlier fieldwork, forming part of our wider research on ancient Naukratis and Greek-Egyptian relations. Fieldwork was carried out at Naukratis over a period of four weeks in April and May 2014 by a team consisting of field director Ross Thomas, Alexandra Villing and Aurélia Masson (British Museum), Ben Pennigton and Kristian Strutt (Southampton University), Astrid Lindenlauf (Bryn Mawr College), Mohamed Magdy El Deeb, Mohamed Roshdy Gomaa Soliman and Ashraf Salah El Din Mohamed (Ministry of State for Antiquities [MSA], Egypt). Fieldwork was funded by the Honor Frost Foundation, Christian Levett and the Mougins Museum of Classical Art and the Institute of Classical Studies, London. It comprised six distinct aspects: 1. Magnetometry geophysical prospection of available fields in the area in and around the ancient settlement (Figs 1-4). 2. Topographical survey using RTK-GPS equipment. 3. Electronic Resistance Tomography (ERT) sections of the site and its surroundings (Fig. 5). 1
4. Geological auger coring in the area in and around the site (Fig 6). 5. Excavations in the northern part of the site, in the area identified by Hogarth as the Greek Hellenion sanctuary, now at the bottom of a largely dried-up lake (Figs 2, 7, 8). 6. Training of MSA inspectors in the use of survey equipment and excavation (Fig. 4). The generous and proficient help of our MSA inspectors from Beheira Inspectorate in Damanhur, the guards at the site in Kom Ge if, and the collaboration of the farmers of the villages around Naukratis contributed greatly to the success of this season. Magnetometry The Naukratis project used a type of magnetometer called a fluxgate gradiometer to survey the available fields. This instrumentation, which has previously proved very successful on sedimentary geology (such as the Nile Delta) in locating sub-surface archaeological structures and features, works by measuring minor changes in the earth s magnetic field, thus facilitating the detection of many different types of archaeological features including kilns, hearths, ovens, ditches, pits and walls or where anthropogenic deposits have been used to infill pits, postholes or other features. However, the results of a magnetometry survey can be adversely affected by areas of modern disturbance and the presence of ferrous material. The equipment used was a Bartington Grad601-2 Dual Array Twin Fluxgate Gradiometers instrument (Fig. 4). Readings were taken at 0.25m intervals along traverses every 0.5m within the 30 x 30m grids, providing a high-resolution survey for detecting potential archaeological remains. A resolution of 0.1 Nano-tesla (nt) was used for each instrument, with all traverses walked in a zig-zag fashion. The data were recorded using a Grad-01 data logger. At the end of each day data were downloaded from the instrument using GRAD601 programme. The data was processed using the programme Geoplot 3.0 to remove any environmental disturbances or variations produced in the course of the survey. The results were then checked to see if any grids needed re-doing and to identify archaeological traces to inform the selection of other areas to work. Detailed post-processing was undertaken by Kris Strutt (Southampton University) using Geoplot 3.0 software, though further work is required to process the results and to interpret them thoroughly. The map included here (Figs 1-3) shows the results of the 2014 season alongside those of previous seasons, cumulatively covering 26 hectares. A number of archaeological structures can be seen within the results that relate to likely Late Period, Ptolemaic and Roman buildings. The majority are domestic tower and courtyard houses, although structural elements likely to belong to the large religions structure identified by Hogarth as the Hellenion (Fig. 2) can also be seen, as well as the likely main temple and other buildings of the Great Temenos, the Egyptian sanctuary of Amun-Ra (Fig. 3). The edge of the settlement and the river front of the now long dried-up Canopic Nile river branch is visible in the west, with adjacent likely harbour (quay?) and warehouse structures (see also Thomas and Villing 2013). The ERT and auger surveys provide data on the depth and dating of these features, confirming successive Late Period, Ptolemaic, Roman and Byzantine phases. Important additional information was supplied by Sabri Choucri, whose rescue excavations for the Ministry of State for Antiquities some years ago revealed parts of a massive stone quay to the west of the Egyptian sanctuary precinct s monumental gate, probably part of the sacred quay serving the temple of Amun-Ra. 2
Figure 1. Map of Kom Ge if with magnetometry results from 2012, 2013 and 2014 ( The Naukratis Project, British Museum). 3
Figure 2 Map of northern Kom Ge if (including the area of the Hellenion, the Greek sanctuary complex) with magnetometry results from 2012, 2013 and 2014; excavation area of 2014 outlined in red ( The Naukratis Project, British Museum). 4
Figure 3 Map of southern Kom Ge if (including the area of the Great Temenos, the Egyptian temple complex of Amun-Ra) with magnetometry results from 2012, 2013 and 2014 ( The Naukratis Project, British Museum). 5
Figure 4 Kris Strutt training MSA inspector Mohamed Roshdy Gomaa Soliman in magnetometry in the area east of the Great Temenos ( The Naukratis Project, British Museum). Survey All magnetometry grids, trenches, structures, features, auger holes and spot finds were located in real-world coordinates using a Real Time Kinematic (RTK) Global Positioning System (GPS), in WGS84 (with ellipsoid heights). This ensures that survey data are fully geo-referenced and facilitates the incorporation of other survey methodologies into a realworld co-ordinate system. The survey was conducted using two RTK GPS units a (Leica GX1230) as a reference station, and a rover unit. The reference station occupies a fixed point and receives positioning data. It then broadcasts its location to the rover unit, allowing the rover to more accurately resolve its own position. Data was downloaded and viewed in ArcMap to check coverage. This was crucial for locating accurately finds, trenches and the position of geophysics, but particularly for recording accurately the relationship between river heights and the settlement during different periods. Additional archival research allowed us to integrate the stratigraphic levels recorded during Petrie, Hogarth, Leonard and Coulson s previous work with the current lay of the land, so as to predict the level of features selected for excavation in 2014, a method we will also use in future seasons. 6
Electronic Resistance Tomography (ERT) An ERT section was undertaken by Kris Strutt with a Tigre ERT. Using an array of electric probes, this devise measures the resistance faced by electrical currents passing through the ground at different depths, thus revealing a range of high resistance solid barriers and low resistance waterlogged deposits. This provided a section over 15 meters deep running nearly 900 meters East-West across the southern part of the site and Canopic branch of the Nile (Fig. 5). A second high-resolution (1 meter spacing) section was collected in the central part of the site, in the area of the likely location of the main harbour of the site. The aim was to better understand the geological relationship between the river and the settlement, though results may also identify walls and other structures within the main area of the ancient settlement. Further work is required to interpret the data gathered this year, including adjusting the data to the site s topography, though the deep and wide Canopic river channel (in yellow-orange) is already clearly visible on the left (west) of Fig. 5 now. The data is important moreover for interpreting of the depth and extent of deposits represented in geological auger cores, providing crucial dimensions for understanding the full extent of the sand body deposited by the Canopic river branch at different levels (at different times), Together they provide a better understanding of the size and form of the river in antiquity and how this changed over time. Figure 5 ERT 800 meter section of southern Kom Ge if, cutting through the Canopic branch of the Nile in the west (left) to the area of the Great Temenos in the east (right) ( The Naukratis Project, British Museum). Geological auger study The geological auger study conducted by geologist Ben Pennington in 2013 and 2014 (Fig. 6) helped us to better understand the ancient landscape of Naukratis by refining our understanding of the Canopic branch of the Nile, the location of which was indicated for the first time in 2013 through geophysical and geological work, and to understand the geology and site formation processes in the area of the ancient settlement and its surroundings. An Eijkelkamp hand auger was used to recover subsurface material from points (surveyed via RTK-DGPS) located to the west, east and within the ancient settlement. An 7
interpretation of the paleo-landscape was accomplished through the description and analysis of the reconstructed sedimentary sequences. Local workmen were trained and employed to drill the auger, and cores were brought up in ~20cm chunks for immediate analysis and description. The sediment was recorded in terms of grain size, clast percentage and composition, organic content and form, as well as sorting, rounding and mineralogy where appropriate. A general description of the sediment was also made. Sherd material was removed and bagged for identification to provide a chronology for the stratigraphy. The preliminary interpretation of the data suggests the presence of a (c. >3 meters) deep and (>250 meters) wide river channel located to the west of the site, with the river flowing south to north. The data support a lateral migration from east to west. This was the Canopic branch of the Nile. More analysis is required in order to ascertain the exact relationship between this dynamic riverine landscape and the development of the settlement, with different environments produced by human intervention caused by the construction of quays and cutting of inlets. Auger cores within the harbour area revealed Late Period, Ptolemaic, Roman and Byzantine surfaces relating to developments of the river front at these times, which will inform the planning of excavations of the harbour in future fieldwork seasons. An auger core in the harbour area also revealed waterlogged fragments of worked wood (Fig. 6 inset), a positive indication of the good preservation levels in the harbour area of this important port. Figure 6 Geologist Ben Pennington drilling auger cores in the area of the excavations; inset: finds of worked wood from auger cores in the harbour area ( The Naukratis Project, British Museum). 8
Excavation Excavations in 2014 focused on a small area in the north of the dried-up lake of Kom Ge if (Figs 2, 7 and 8). The prime objective was to test magnetometry results from the area that revealed traces of a large wall, thought to be a large mud-brick wall excavated by David Hogarth in 1899 and 1903 and possibly part of a temenos wall for the Hellenion, a large sanctuary and administrative complex established by Greek traders operating in Naukratis. Our RTK DGPS topography and architecture survey had located a door jamb marked in Hogarth s maps of 1899 and 1903, further supporting our hypothesis that the large mud-brick wall was to be found in this area. The area s topography and ground water level in combination with research on Hogarth s excavation from publications and excavation notes meant we could predict that the sanctuary wall was preserved and accessible in this area, with structures expected preserved within c.0.5 to 1 meter of the surface. We were not certain of its precise location, as the accuracy of Hogarth s 1899 and 1903 maps could not be trusted and we could not be sure to what extent the wall might have been damaged or destroyed after Hogarth s excavations, either by locals (sebbakhin) removing soil from the site or by salt and water from the deep lake that had covered this part of the site for over 100 years. Figure 7 Excavations in the northern part of the site, in the area of the Hellenion, Trenches 1, 2 and 3 (from right to left), view from the north ( The Naukratis Project, British Museum). 9
Figure 8 Excavations in the northern part of the site, in the area of the Hellenion, Trench 1: western edge of mudbrick wall, view from the south ( The Naukratis Project, British Museum). Excavations started in a 10 by 10m area (Trench 1) which revealed the western edge of a large wall at the eastern end of the trench (Figs 7 and 8). Two new trenches (Trenches 2 and 3) were opened to the east to investigate the eastern edge of the wall (Fig. 7). Trench supervision and finds processing were carried out by Ross Thomas, Aurélia Masson, Astrid Lindenlauf and Alexandra Villing. Archaeological deposits were found between c. 0.2 and 1.3 meters below the surface. The wall was built directly upon the basal mud as described by Hogarth. The basal mud was cut away from the external edge of the wall, probably to raise the height of the wall and to provide building material for its construction. This left a basal mud platform 30cm high, over which the wall of which 0.86 meters were still extant was built. Individual mud-bricks were not usually identifiable, with the exception of a limited number of sandy mortar joins between courses. The bricks were entirely clean of artifacts because they were built on and of virgin soil not containing any residual pottery material. Geologically the wall was slightly sandier than the basal mud. Numerous layers of mud mortar, slightly sandier than the wall, were added to the outside of the wall and this was applied more thickly at the base, producing a slightly sloping wall. Against the wall, 1.1m of stratified archaeological deposits produced an abundance of material, mainly pottery, produced locally in Egypt as well as imported from the Levant, Cyprus, mainland Greece and the East Greek cities in Western Anatolia, dating to the 6 th century BC. A large number of animal bones, particularly fragments of jaw and cranium, were found. The animals were sheep/goat, cattle and pig. Above was a layer of collapsed mud-brick and mixed material with a series of pits and fills, including mud-brick material 10
from the wall that had likely been damaged by activities of archaeologists and sebbakhin in the 19 th and early 20 th centuries. The deposits were further damaged (melted) and complicated by the lake that had covered this part of the site for the past 100 years. Lake deposits and surface material covered the latest archaeological layers, having accumulated during recent years. Pitting in the early 20 th century had obscured the inside edge of the wall, though the top of the wall was located in Trench 3, confirming that this wall was indeed 7-9 meters wide as stated by Hogarth. The presence of ground water, over a meter higher to the east of the wall towards the sanctuary of the Dioskouroi excavated by Ernest Gardner in 1885/6 than to the west of it(with the wall and the basal mud layer acting as a barrier to the water draining in from the fields), meant that we could not progress beyond the more recent pits and fills, exposing only part of the top of the wall, but not any defined internal edge. It is possible that the inside has been obscured by collapse of the wall. The 6 th century BC material associated with this wall included many objects consistent with Greek and East Greek religious practices, notably Greek (especially East Greek) pottery, but also similar quantities of Cypriot and Phoenician pottery. At the same time, local Nile Delta wares account for nearly 80% of the pottery from this area; their abundance as well as the presence of an Egyptian figurine and a Cypriot mortarium with a Greek inscription suggests this area was more culturally mixed than previously thought. Further investigations are required to confirm if the picture is the same in other areas, or whether exclusively Greek material is to be found e.g. within temple precincts. Pottery and small finds Over 4000 diagnostic pottery sherds and small finds were collected and studied this season. All were identified and recorded, whilst a selection were washed, drawn, photographed and stored for future reference and study. The majority were recovered from the excavations of Trenches 1, 2 and 3. In addition, 83 small sherds were found in the auger holes drilled by Ben Pennington, helping to date the geological and archaeological sequences represented in these cores. A single Hellenistic stamped Koan amphora handle was collected during magnetometry south-east of Kom Hadid, a mound at the eastern of the ancient settlement where abundant survey pottery was collected in previous seasons (Thomas and Villing 2013). The remaining material came from the excavations, from two types of deposit. Firstly the pottery from the surface, including that from pits in the top of the temenos wall, included some 7 th to 5 th century BC material but consisted predominantly of 4 th century BC to 7 th century AD objects, with some modern material also. We believe that this material largely came from Hogarth s excavations in the area, being re-deposited between 1899 and today. All indicator sherds (rims, bases, handles) in this deposit were quantified on site, with a limited number of significant pieces retrieved for further study. The material was consistent with that observed across the site in previous seasons, with the usual variety of Hellenistic (particularly from Rhodes see stamped amphora handle Fig. 9, left Kos, Knidos and Thasos), Roman and Byzantine imports (from Cilicia, Cyprus, Turkey and Gaza), but predominantly local or Egyptian in origin. A single figurine fragment dating to the Late Period was found at the bottom of this disturbed deposit, near the undisturbed 6 th century BC layers below where it likely originated (Fig. 9, centre). 11
Within the sealed archaeological deposits underneath these disturbed layers a very different range of material could be recognized. Finds in the layers abutting the wall were consistently 6 th century BC in date, with the lowest layers dating to the first half of the 6 th century BC. In this area all sherds were quantified, with a representative selection of all material as well as any finds of particular significance retained for further study. Imports were identified from Athens (Fig. 9, right), Corinth, Cyprus, Lebanon, Miletos, Chios, Samos, Thasos, Lesbos, Rhodes and other Aegean and Ionian (East Greek) areas. For the first time, the excavations provide reliable representative data for the wide range of ceramic material and other finds present at the site in a 6 th century BC context and thus a valuable corrective for the biased selection of finds preserved from early fieldwork. Figure 9 From left: stamped Rhodian amphora handle, 2nd century BC; figurine head, 6 th -4th century BC; Attic black figure skyphos fragment, late 6th century BC ( The Naukratis Project, British Museum). Summary The work of the 2014 season produced significant new data on the layout of ancient Naukratis as a settlement and port city, its local environment and hinterland, and its relation to the river. We now have a much better idea of the extent of the ancient site, the location and development of the Canopic branch of the Nile, and a better understanding of the archaeology particularly of the northern part of the site. The settlement is much larger than previously through, exceeding 60 hectares. It was accessed from the Mediterranean via the broad and deep Canopic river branch, navigable by seagoing vessels. The main river harbour was located in the western part of the settlement, north of a limestone ritual quay that was connected with the temple of Amun-Ra via a processional way. Over a meter of archaeology is still preserved in places in the dried up lake, and much more in many other parts of the site. A large number of mud brick and stone structures particularly in the north and east of the site. A stratified assemblage from the area of the Hellenion for the first time provides insights into the full archaeological spectrum of the site as well as its religions practices in its earliest periods, indicating closer links with Cyprus, Phoenician and the Levant and a greater degree of cultural mixing between Greeks and Egyptians than previously thought. The results highlight the need for, and great potential of, further research and especially fieldwork, whilst providing significant new insights relevant to a range of disciplines. 12