The two seedling trays that looked the same, until we measured them
The nursery bench was still wet from last night’s rain. I lined up two trays of young trees: our trusted stock, and a new delivery with missing labels. I couldn’t match seedlings one by one, so I needed a tray-to-tray check. Like a hospital comparing two piles of scans without pairing each picture.
I tried the lazy way first. I stood back and judged the trays by looks: green, tidy, fine. But close up, one tray had thinner stems and odd leaf texture, like it grew under different lights. That’s like using photo-style “looks similar” scores on medical scans and missing the bits doctors care about.
So I stopped doing a single eyeball verdict. I wrote down the same simple measurements for every seedling: size, shape, surface feel, light-and-dark patterns. With scans, people can take the same kind of shared, repeatable measurements from each image. Takeaway: compare what you can name, not a vague hunch.
Then I tightened it up. I checked each seedling under a few filter sheets that made fine edges pop, not just normal light. I also stopped letting one freakishly tall seedling set the scale for the whole tray. And I used the trusted tray as the ruler, so both trays were judged against the same reference.
Now each tray became a cloud of numbers on my clipboard. The question turned into: how far apart are these two clouds, in their middle and in how wide they spread? If the gap got huge, I squeezed it into an easier-to-handle score. In scan terms, it’s one distance between two whole collections, even when nothing is paired.
At the loading dock, that distance worked like a warning light. Most familiar deliveries sat close to the trusted tray; odd ones sat far away, so we could flag them early. The same idea helps with medical scans: it can spot when an incoming batch is from a different machine or setup, even if the change is subtle.
It even behaved when there were only a few seedlings on the bench. A quick “looks right” judgement swings about with small trays, but the measurements stayed steadier. When the trays didn’t match, I could point to what shifted most, like stem thickness or leaf texture, instead of shrugging at a mystery.