Struck through twice before letting it stand

The log for the seventh of January, 1610, carries a blot of ink at the lower right corner of the page, and beside it a number I struck through twice before letting it stand. I had recorded three fixed stars near the belt of Orion, and then, in the margin, a figure for the distance between the second and third. The fair copy in the Sidereus Nuncius gives the position cleanly. The margin gives the argument I had with myself about whether the air that night was steady enough to trust the micrometer. I decided it was not, and wrote the figure anyway, and then wrote it again in the fair copy as though the doubt had never occurred.

This is the assumption I trust least in my own method: that a measurement entered twice, once in doubt and once in confidence, is the same measurement. The instrument is not mistaken. The lens grinds true, the mount holds, the eye at the eyepiece is patient. But the hand that writes the second figure has forgotten the hand that wrote the first, and the forgetting is not recorded. The margin remembers the quarrel. The finished text remembers only the verdict.

I have been auditing the Jupiter observations of the same month for this reason. The satellites are there in every night’s entry. What is not there is the weather, the hour, the tremor in my own wrist after four hours at the tube. Those conditions live in the margins or nowhere. When I compare the nights against one another, I am comparing verdicts, not quarrels, and the verdicts agree more than the quarrels would.

So the next thing I will do is this: I will copy the raw margin notes for the fourteenth through the twenty-second of January into a separate sheet, unedited, with the struck-through figures left struck through, and I will run the satellite positions from those raw entries alone, before consulting the fair copy. If the two sets disagree, the disagreement is the finding.