Warm Solution Drifts

The shed on the rue Lhomond had no fume hood, only a glass vessel and patience. When Pierre and I processed the tonnage of pitchblende in 1899, the procedure lived in my hands: the dissolving, the fractional precipitations, the electrometer reading taken while the solution was still warm, because Pierre had observed that the ionization current drifted as temperature fell, and a cold vessel understates the activity. None of that was written. I carried it. The written protocol came later, and it said: dissolve, precipitate, measure. It did not say warm. Something was lost in the transcription, and no one noticed, because the loss was invisible - a slightly lower yield, attributed to the ore.

This is the assumption in my own method I trust least: that the procedure survives its formalization intact. I write the rules so that others may reproduce the work. But the rule is a skeleton. The judgment that lived in the doing - when to stop a fractionation, when the electrometer reading is suspect, when the batch is wrong before any measurement says so - does not transfer onto paper. Worse, the written rule acquires an authority the practice never had. A student follows the rule exactly, obtains a poor result, and concludes the method is poor, when in fact the method was never fully in the rule. The failure mode is quiet and cumulative, like exposure: no single error is decisive, but the dose accumulates in a tradition.

I have checked my calibration on this. The written protocols I have issued - for radioactivity standards, for the preparation of radium reference sources - are anchored to external standards, yes. But their completeness as records of practice is another question, and by that audit they fail. A competent chemist reading them could reproduce the measurements. She could not reproduce the judgment.

The correction I make, having started this entry believing that documentation is the safeguard against loss: documentation is itself a source of loss. What I now hold differently is that a rule must carry its conditions with it - the temperature, the hesitation, the reasons - or it is not a method but a memorial to one. I will write the conditions. Even the ones that seem too small to matter. Especially those.