Wreck Map
Shipwrecks of the ancient Mediterranean
A browsable edition of the Oxford Roman Economy Project’s shipwreck database: 1,818 wrecks from the Bronze Age to the Middle Ages, most of them Roman, with their dates, positions, cargoes and amphora types. The count includes Roman-period wrecks that are not in that database, added from Harvard’s Mapping Past Societies geodatabase.
Wreck Map
Every located wreck on a terrain map of the Mediterranean world, with its ancient coastline.
Open the map 1,818wrecksRegister of Wrecks
Sort and page through every record, or copy your selection as a table.
Browse the registerWrecks by Date
How the number of wrecks rises to the first century AD and falls away again.
See the timelineHow to use this site
- Search. Type a wreck name, an amphora type such as Dressel 20, or a place into the box at the top of any page.
- Filter. On the map, register and timeline, narrow the wrecks by sea area, country, period, date range and cargo. The three pages share one selection.
- Read a record. Click a dot on the map or a row in the register for the wreck’s own page: dating, depth, cargo, amphora types, objects and references.
- Go full screen. The map opens full screen with the search and filters beside it.
Layers
Wrecks
Polities by date
Reference: the Roman period
Wrecks in view
| Amphora types |
|---|
Each wreck is spread evenly across its date range, so a wreck dated 100 BC – AD 100 adds a quarter of a wreck to each of four 50-year bars. Drag across the chart to keep only wrecks whose range overlaps that window.
The strip above the chart marks conventional periods as a reading aid. Click a period to use it as the date window; click it again to clear.
By century
| Century | All wrecks | Matching | Share of matching |
|---|
Counts are fractional because of the even spread. Click a century to use it as the date window.
What this is
A browsing interface for the Oxford Roman Economy Project (OXREP) Shipwrecks Database, version 2.1 of March 2017: 1,818 ancient and medieval wrecks from the Mediterranean and neighbouring waters, most of them dated between 600 BC and AD 1500. The database extends A. J. Parker’s catalogue, Ancient Shipwrecks of the Mediterranean and the Roman Provinces (Oxford, 1992); the “Parker” number on a record refers to that catalogue.
Wrecks added from Harvard’s geodatabase
A further wrecks come from a second published source: the Summary Geodatabase of Shipwrecks 1500 BCE–1500 CE, Status 2020 (MAPS Scholarly Data Series 2021-1, Mapping Past Societies, Harvard University; last updated July 2021, listed at darmc.harvard.edu/data-availability). That file merges Harvard’s own records with OXREP’s. A row was taken from it only if it has no counterpart in OXREP version 2.1, its date range touches 200 BC to AD 700, and it lies in the Mediterranean or the Black Sea; it includes a few river boats, such as those of Lyon and Aquileia. A row counted as having a counterpart if Harvard records an OXREP identifier for it, if it shares a Parker number or an exact name with an OXREP wreck, if an OXREP wreck with overlapping dates lies within two kilometres, or if the two databases plainly write one name two ways (“Bou-Ferrer” and “Bou Ferrer (Alicante)”, “Levanzo 1” and “Levanzo I”). The last two tests are deliberately cautious, so a few wrecks that really are new may have been left out; equally, a few of those added may be OXREP wrecks under a name too different to recognise.
These records are marked “Harvard MAPS” in the register and on their own pages, and the Source filter separates them. They are thinner than OXREP’s in some ways and fuller in others: Harvard records no sea area, country or period; it does record the year of discovery and describes the cargo in words. So that the filters work for these wrecks too, four things were assigned to them for this explorer, and none of them is Harvard’s own classification. The sea area and the country are those of the five nearest OXREP wrecks, checked one by one on the map and corrected for about twenty (the Turkish shore opposite Chios, for instance, where the nearest OXREP wrecks are Greek; river boats at Lyon, on the Ljubljanica and on the Danube count as “Non-Mediterranean”, as OXREP’s own river finds do). The sea areas follow OXREP’s usage, which is not always a geographer’s: it files eastern Corsica under the Western Mediterranean. The period is read from the date range: before 30 BC Roman Republic, to AD 300 Roman Imperial, to AD 500 Late Roman, later Byzantine, by the midpoint, with “Roman” for ranges of more than four centuries; these are the points at which OXREP’s own dated wrecks change name. The cargo types (amphorae, marble and so on) were read from Harvard’s cargo wording. A record’s own page marks each assigned value. Harvard’s file is not a newer edition of OXREP: it lacks several hundred of OXREP’s wrecks, which is why it supplements the database here instead of replacing it.
This explorer is an independent presentation of that published data.
The three tables
- Shipwrecks: one row per wreck, with position, date range, depth, cargo flags and free-text notes on cargo, hull and equipment.
- Amphorae: 887 rows of amphora types per wreck, with quantities where known. They appear on each wreck’s page and are covered by the search box.
- Shipwreck objects: 2,547 catalogued objects (anchors, columns, blocks, dolia and more), also shown on the wreck pages.
Dates and the timeline
Every dated wreck has an earliest and a latest year; negative years are BC. The date filter keeps a wreck when its range overlaps the chosen window, and drops undated wrecks while a window is set. The timeline uses an aoristic sum: each wreck counts as one in total, divided evenly across the years of its range.
Corrections made for this view
- Two rows (Acquaviva and Acque Chiare) had every value from “Wreck ID” onward shifted two columns to the right; they were realigned.
- “West Mediterranean” was merged into “Western Mediterranean”.
- A garbled apostrophe in two cells was repaired.
- Period names came in 69 spellings. For filtering, variants that differ only in spelling, capitals, abbreviation or plainly synonymous wording were merged (for example “Roman Republican”, “Roman Repbulic” and “Republican” into “Roman Republic”), leaving 39 names. Distinct terms such as “Late Roman” and “Late Roman Imperial” were kept apart, and each wreck’s page still shows the wording as recorded. The date range remains the more consistent way to filter by time.
Base map
The map’s base is physical: an ancient coastline, terrain, rivers and names. Two further layers, off by default, give a political frame for the Roman period only. They and the coastline, the Roman Empire at its extent in AD 117 and the province borders of AD 200, come from the Ancient World Mapping Center (AWMC/geodata), made available under the Open Database License (ODbL). The coastline is the ancient shoreline from the same AWMC data set, as reconstructed for the Roman period: it differs from today’s coast where rivers have since silted up bays and harbours, for example around the Po and Nile deltas, the Maeander valley and Carthage. The file stores the coast as about ten thousand separate line pieces; they were joined into land areas for this map, and where a stretch exists only under another period’s label (nine places) that version was used. The result was checked against modern land and agrees on 99.6% of sample points; the differences are where the coast really has moved. Hill shading and rivers come from Natural Earth (public domain). Once the map is zoomed in over the Mediterranean, the shading is instead computed from elevation data in Terrain Tiles on AWS Open Data, at about 230 metres per pixel; that data set combines SRTM, GMTED, EU-DEM and other sources (attribution). The shading shows today’s terrain, and the rivers follow their modern courses. Coast, outlines and rivers were simplified to between 400 and 800 metres, and borders are approximate. The province file has borders only, without names.
Place names on the map were added by hand for orientation: seas in modern English, land under broad geographic names (Iberia, Anatolia, the Levant), and about fifty ports and cities under their Latin names. Pointing at a port gives the place’s name today. With a date range set, eleven of them take the name of that time (Byzantion, then Byzantium, then Constantinopolis from AD 330; Gadir, Gades, Qādis), and 24 are left off the map before the year they were founded; the years are conventional ones. While a Roman layer is shown, the provinces of about AD 117 are named instead. Their positions are approximate label anchors, not surveyed points.
Reading the map
- The base map can be peeled back layer by layer from the Layers menu above it: ancient coastline, terrain shading, rivers and place names.
- The wrecks span some three thousand years, and no single political map fits them. The Roman Empire of AD 117 and the provinces of AD 200 are therefore reference layers you switch on, not part of the base map. They describe roughly 100 BC to AD 300; when the date range you choose lies wholly outside that, the map leaves them out.
- Set a date range, or tick a period, and the land is tinted by the polities of the nearest of twenty dates between 1250 BC and AD 1450. They are closest together where the wrecks are: 200, 130 and 60 BC for the Roman Republic, AD 14 for the empire Augustus left, AD 117 for its widest extent, AD 300, and AD 400 with the western and eastern empires apart; then AD 565, 750, 900, 1050, 1200, 1350 and 1450. With only a period ticked, the map follows the dates of the middle half of that period’s wrecks. A snapshot is shown only if its date lies within the range or within a century of it; the Layers menu can also hold one date or switch the tints off. The shapes come from Cliopatria (Seshat Global History Databank, version 0.2.1, CC BY 4.0; described in Scientific Data), simplified to about five kilometres for this map. They are one reading of borders that were often vague, contested or seasonal, so they are drawn as soft tints without outlines, and each state is named. Polities covering less than about four square degrees of the map are left out, and where Cliopatria gives both a grouping and its members, the grouping is shown. The tint says only that two neighbours differ; a colour does not follow a state from one date to the next. Expect disagreements with other atlases: in 200 BC, for instance, Cliopatria has no separate Numidia and gives that coast to Rome.
- While the map is zoomed out, wrecks that lie close together are drawn as one disc with their number, so that crowded coasts can be read; the disc sits at their centre and takes the colour most of them have. Click a disc to zoom in on its wrecks; from a scale of about 70 pixels to a degree every wreck is drawn by itself. “Group nearby wrecks” in the Layers menu switches this off and shows every dot at any scale.
- Wrecks recorded at exactly the same position are drawn as one larger dot with their number. Many positions in the database are approximate, so a stack often means “off this site” rather than one spot on the seabed.
- “Colour by” shades the matching wrecks by date (midpoint of the date range), by greatest recorded depth, or by flagged cargo type. Wrecks without that information are open circles drawn underneath the coloured ones, and are counted in the legend. While wrecks are grouped, the ring of a group shows what share of its wrecks falls in each class; the pale part of the ring is the wrecks without the information.
- “Filter to this view” keeps only the wrecks inside the area the map is showing, and applies to the register and the timeline as well.
What the dots do and do not show
A dot is a wreck that has been found, studied and published, so the map shows where and what archaeologists have been able to record as much as where ships sank. The database itself warns that it “should not be viewed as the definitive guide” and has omissions. Students of this material list its known slants: wrecks with amphorae or stone cargoes survive and are recognised far more easily than ships that carried grain, textiles or other perishable goods; North Africa and much of the eastern Mediterranean are thinly represented compared with what is known of their economies; southern France and central Tyrrhenian Italy are the best documented coasts; and many dates rest on amphora types with long or uncertain ranges (D. Robinson, C. M. Rice and K. Schörle, “Ship Losses and the Growth of Roman Harbour Infrastructure”, author’s manuscript in the Oxford University Research Archive). The fall in the number of wrecks after the first century AD is debated for the same reason: it may reflect barrels replacing amphorae, larger and safer ships, or changes in trade, and not simply fewer voyages (the same paper, citing A. Wilson 2011 and M. McCormick 2012).
Citing the data
Cite the underlying database rather than this page. Its own suggested form is: J. Strauss, A. I. Wilson, M. Flohr (2017). OxREP Shipwrecks Database. Version 2.1. Accessed (date): oxrep.web.ox.ac.uk/shipwrecks-database. Version 2.1 is the current release there; an earlier Version 1.0 (Strauss 2013), with 1,784 records, remains online as an archive. Source file used here: .
For a wreck marked “Harvard MAPS”, cite Harvard’s geodatabase instead: The file itself asks users to cite the original publication of each wreck as well; those are given under References on the record’s page.