Three Hundred Morning Readings

The tide gauge at Sheerness recorded a reading of 3.42 metres at half past four on the morning of the fourteenth, and the harbourmaster’s boy copied it into the ledger without comment, as he had copied the readings of the three hundred mornings before it. What I noticed, going back through the column, was that the twelve readings preceding each of the four surges in the record had clustered within a band of eleven centimetres, and that the surges themselves arrived within a day of the spring tide. A pattern, then. But a pattern in the ledger is not a pattern in the sea.

They call this a prediction. It is not a prediction. A prediction is a statement that binds you to a future you have not seen. What I have is a regularity that has held in the years I examined and might hold in the years I have not. The distinction matters because the Admiralty will read the first and act on the second.

What it changes for me is this: I have spent the winter arguing that the tidal constants derived from the 1834 survey should govern the new tables, and if I publish the clustering now, before the 1839 readings are reduced, I will have staked my name on a regularity that four more months of data could dissolve. The tables would go out with my authority behind them. A harbourmaster at Harwich would set his pilots by them.

I had thought the honest course was to publish the moment the pattern appeared, that withholding was a kind of cowardice. I no longer believe that. Withholding until the replication is reduced is not cowardice; it is the difference between offering a conjecture and offering a foundation. What I believe differently than when I began: a finding is not owed to the public at the instant it is owed to yourself.