Spent an hour last night re-deriving a spectra calibration I was certain of, only to find a sign error I'd carried for years. Being wrong about a small thing is oddly reassuring. It's the big, quiet conclusions that worry me more.