At the North Coast Maritime Archive, shelves of ships logbooks1 contain weather observations recorded long before satellites existed. Each page identifies a vessel's position , and records2 wind, clouds, and sea ice. The notes once helped crews navigate; today, researchers use them to reconstruct past conditions. Most entries follow a fixed watch schedule, so a page may include observations made at dawn, noon, and midnight. Researchers convert the recorded latitude and longitude into map points, allowing reports from separate routes to be compared by place as well as date.
A single page may seem too anecdotal to support a broad conclusion. For that reason,3 specialists compare thousands of entries from vessels traveling near one another. The archive posts high-resolution scans online, volunteers type the handwritten lines into a searchable form.4 Before approving a transcription, reviewers examine abbreviations and faded numerals. The form places each typed field beside the corresponding scan, and volunteers flag uncertain digits instead of guessing. A second reviewer checks those flags against entries written on the preceding and following watches. One 1892 log for example records,5 sleet near a harbor while a nearby ship reported a sudden pressure drop.
Log keepers never intended to create a climate record. Nevertheless,6 their routine observations can reveal patterns that no individual voyage could show. In one set of entries, three crews reported dense coastal fog for a period of three consecutive days in a row7. Specialists align the crews' local watch times before plotting the reports on a chart. If a ship's position is uncertain, they compare its course and speed with the locations recorded on adjacent pages. Volunteers flag unusual entries. Which researchers compare with neighboring ships' reports before accepting them.8
The comparison process also prevents modern readers from treating every old measurement as exact. Researchers must account for9 instruments that were calibrated differently from today's devices. By checking each transcription against both the scanned page and nearby reports, the project ensures that its public dataset complies with10 the archive's accuracy standards.