Artificial intelligence has helped historians turn about 100,000 pages of handwritten port books into a database and interactive map of Britain’s maritime trade. The project reveals a surprising detail: in 1696–97, Europe accounted for around 80% of England’s recorded overseas trade value.
Watch Desk analysis
What happened
The University of Southampton project used specialist software and AI to extract information from port records dating from 1565 onwards. The documents record voyages, ships, ports, shipmasters and cargoes. Researchers say they have assembled information on hundreds of thousands of voyages and more than 3,000 ports.
The team says the AI was taught to handle the distinctive handwriting of different port officials and changes in spelling over time. Historian Craig Lambert said that extracting and organising the information this way would otherwise have taken years. The results still needed manual checking.
The University of Southampton’s project account describes an interactive map for exploring trade by period, port, ship, route and goods. The data also records more than 101,000 cargo entries for coal, its most frequently recorded commodity.
Why it matters
This is AI doing useful historical work at a scale that would have been painfully slow by hand. Instead of leaving a vast archive as thousands of difficult-to-read pages, the project makes records searchable and allows researchers and the public to explore patterns across time and place.
The results also give a more textured picture of Britain’s trade. Despite the growth of the empire, Europe remained a major trading region: the project’s figures put its share at around 80% of recorded overseas trade value in 1696–97. That is a finding from this collection and period, not a summary of every year or every kind of commerce.
Our read
The impressive part is not simply that AI can read old handwriting. It is that a painstaking transcription task has become a tool for asking new historical questions, with people still checking the extracted information. AI gets the archive moving; historians supply the judgement. Sensible division of labour, for once.
The map is worth attention both as public history and as a practical demonstration of AI-assisted research. The next question is how well the approach handles difficult records and whether other archives can use it without losing the messy details that make them valuable.
What to watch
- How the project documents transcription accuracy and the manual checking process.
- Whether the map and underlying data expand to include further records.
- What new historical patterns researchers find by exploring the database.
Discussion spark: When AI makes huge archives searchable, should historians prioritise expanding the collection or publishing detailed evidence of how accurately it was transcribed?
Sources and evidence
- Source update (19 October 2026, 00:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.