The people keeping the tide ledgers were not recording the weather. They were recording the water. But a person who walks to a harbour wall every morning for forty years will, occasionally, write down why the walk was unpleasant.
Marginalia as data
“Heavy rain from first light.” “Could not read, spray.” “Wind NE, strong, since night.”
There are roughly four thousand such notes across the forty-one volumes. They are not systematic, they are not calibrated, and they were never intended to be compared with one another. What they are is dated.
What that buys you
Onset. Not intensity, not total rainfall, not anything a meteorologist would accept as a measurement, but the first day of sustained rain, in this specific place, in most years since 1912. That is a thin signal and a long one, and long thin signals are exactly what the instrumental record is short of.
The obvious objection
Different people wrote these notes, with different thresholds for what counts as “heavy”. A reader in 1930 who mentions rain three times a year and a reader in 1970 who mentions it weekly are not measuring the same thing.
This is real and it is not fatal. The comparison that survives is within a single reader’s tenure, and tenures were long: the median is nineteen years. Stitching them together requires care and an honest error bar, and it is the kind of work where saying “we do not know” about the joins is more useful than smoothing over them.
The point
The ledgers were kept for one purpose and turn out to serve another. That is an argument for keeping records carefully and for keeping them at all, including the parts that seem beside the point at the time.