A temperature record is not a thermometer.
It is a chain of custody, and every link in it is a decision somebody made about how to take the reading.
Change the method and the number changes, even when the water does not.
That is a manageable problem when the method is documented, because the offset can be estimated and removed.
It becomes a serious one when the method changed for reasons that had nothing to do with science.
Wartime is exactly that case.
A bucket and an engine room disagree
There are two ways to take the temperature of the sea from a moving ship, and they do not give the same answer.
The old way is a canvas bucket thrown over the side and hauled back up, with a thermometer put into the water on deck. The sample cools on the way up, so the reading runs low.
The other way is to read the thermometer already fitted where seawater enters the engine room for cooling, which is easier, faster and reads too warm.
The intake sits deep in a hot hull, the pipe run is warm, and the machinery around the gauge is warmer still.
Neither method is wrong on its own terms. The trouble starts when a fleet switches from one to the other without the record saying so.
The ocean then appears to warm overnight.
Ruled columns filled in by hand every hour
A warship’s deck log is a ruled ledger, and the weather sits in a band of columns across the page.
Dry bulb and wet bulb temperature, barometric pressure, wind direction and force, visibility, sea state, the temperature of the water and a coded description of the weather.
The watch officer filled in a row for every hour, in pen, in whatever hand he happened to write, on a ship that was often moving fast and sometimes being shot at.
Those pages were then bundled, shipped home and filed, and they have sat in a national archive ever since.
They were never lost and never secret. They were simply written in cursive, which no machine could read reliably.
Unreadable and perfectly preserved.
Four thousand people and a year and a half
The project put scanned pages onto a citizen science platform and asked the public to type out what they saw.
It ran about eighteen months and drew 4,050 volunteers, who worked through more than 28,000 logbook images.
Out of those came over 630,000 records containing roughly 3.7 million individual meteorological observations, covering 19 ships between 1941 and 1945.
The vessels include a battleship and a destroyer, chosen partly because they were at Pearl Harbor in 1941, in the Aleutians the year after, or caught in a typhoon in December of 1944.
In some parts of the Pacific the rescued readings more than doubled the observations available for that period.
The gap was never in the ocean. It was in the transcription.
Where the war sailed is not a sampling plan
The obvious limitation is that a navy goes where a war is, which is not where a climatologist would put a buoy.
Coverage follows convoy routes, invasion beaches and fleet anchorages, so some patches of the Pacific are now very well described and the rest are as empty as before.
The observations also carry the same problems the original had. A wet bulb left dry, a gauge nobody recalibrated, a blackout watch reading by shaded light, in the way that a widely quoted price or a published weight gets repeated without a source.
What transcription buys is not accuracy. It is enough overlapping readings from enough ships that the biases can be compared against each other and estimated.
The reporting on the effort framed it as a data rescue rather than a discovery, which is the right description of what happened.
The pages are evidence, not answers.
The baseline is the whole argument
Warming is measured against a starting point, so the starting point has to be trustworthy or the size of the change is guesswork.
A warm blip sitting in the middle of the twentieth century distorts that, and if it is an artifact of blackout and engine rooms it needs to come out.
The lead researcher put it plainly, saying that without certainty about the historical temperature there can be no certainty about how much the climate has changed since.
Getting that right also feeds the smaller questions, including a strong warm episode in the tropical Pacific during the war and the typhoon that caught a fleet at sea, as the laboratory behind the work set out when it began.
The logbook pages did not change the climate record.
They gave it something to check against.
