Mercury Lies in the Log

The barometer readings from the Orinoco, March of 1800, sit before me tonight with their corrections half-written, and I find myself unable to finish the column. Not because the mercury lied. Because I did.

The blank page of the log book is honest in a way the filled page never is. It asks nothing, assumes nothing, records nothing. Every figure I enter passes first through my own expectation, and I have discovered - auditing these old notebooks like a stranger’s work - that my hand drifts toward agreement. Where the instrument read 712.3 millimetres and my theory of the trade winds wanted 712.5, the ledger shows 712.4, split like a merchant splitting a difference. Small enough to hide. Systematic enough to matter, because a hundred such reconciliations do not average away; they compound, the way a slight current bends a ship’s track over a whole crossing.

Here is the assumption I trust least: that my instruments were read under conditions identical to those under which they were calibrated. The barometer was standardised in Paris, at a temperature and a gravity that the llanos never once reproduced. I corrected for temperature, yes, and for latitude. But the leather of the packing case, the damp of the river nights, the vibration of the canoe - I recorded none of these, because I assumed they were beneath the dignity of measurement. The indigenous boatmen on the Casiquiare knew the river’s behaviour without a single figure; they were not wrong about the water, only uncredited in my tables. I trusted my calibration more than their weather sense, and I cannot now say the ledger was the better for it.

The chain runs: calibration in Paris, transport across the Atlantic, reading in the tropics, correction in my study - and at every link a small unmeasured difference enters, each one defended at the time as negligible. Draw the diagram of these negligibles and the whole picture appears at once: my error is not random. It points, consistently, toward the answer I expected to find.

The decision: tomorrow I re-enter every Orinoco barometric reading from the raw field notes only, striking my interpolated corrections entirely, and I will publish the discrepancy between the two columns alongside the results. The blank page will judge which version I trust.