Do what Curtis did with the chain tooth walk going up, meaning, pull the teeth lock out far enough so it slides up and down and no further. Move the toothed bar up till it stops, but has to lock in straight on, not push up one more lock... is not the starting point.
The next move is to move it down one tooth notch. That's the easy lift/load slack you want. Measuring is moot. Couple of more clues. Go look at a frozen shot of a drag bike chain's lower rung under load, or a road racing chain under load and the belly loop of the bottom rung of the chain. They are not bow-string'd in other words. That ever so slight belly bowl at the top rung is close enough. I just did a clutch and before I took it apart, I could not spin the clutch center. I setup the center screw, the cable slack, all that shit and no joy. Found the plates were hardly worn, no warp of the steels, shuffled the plates around, reassembled, set the screw/cable/lever pull, and could spin the clutch center by hand. It has nothing to do with finding N is the primary slack, but plate breakaway. You want that sag so you can slide the fork with less drag between chain pulls [lift/load]... being that fast on the shift is that fast on the sag is that helper.
Make sense? Not crossbow tight. That's insurance you are not pre-loading on the outer basket, but more has float; no load at the bearings; no chain high-spot coming around to load the bearings. The Curtis move is to swing the tape backwards 2 turns, not forward 2 turns or you have a new measurement. It would take 50 or more spins to get back to the same tooth to link again.