Still haven't scared up a scale, but decided to measure a bunch of pellets today. And yes, the eyes and nerves both survived :-).
First, I changed the methodology from what you did. I went with a set gap, but was interested in a pass/fail only. So for the 4.50's, I set the gap at 4.49. With the pellet sitting on my desk head up, I slipped the caliper under the head at the waist, and lifted. If it passed through, reorient 90 degrees, and try again. If I lifted it off my desk with the calipers, it passed, if not, it failed. I didn't bother to retest the failed pellets to see what they actually measured, all I was interested in was getting a group of consistent size in each head size. That's the first cut. When I come up with a scale, weight will be the second cut. Not sure when that will happen, this may have to go on hold until I can order a scale.
Here's the new data. An interesting trend arose as I was measureing the pellets. As the nominal head size increased, the incidence of undersize pellets also increased. I'm going to use the Match Diabolo S100's as a control, so I measured a bunch of them also. What I have a 4.50's. Both these and the 4.50 Exacts were very good, I didn't actually count, but I'd say only 10% failed. The next step, 4.51, jumped to something like 25% failed. 4.52 was over 50%, and the 4.53's were terrible, around 90% were undersized. I also think I had at least one tin that was mis-marked, the open tin of 4.52's I started with had a near 100% failure rate. I opened a new tin, and got the previous listed result. I'm still not certain about the measureing method, but regardless, I think it's pretty apparent that JSB's quality control is slipping.
Guess that's it for now. I still have one more possible source for a scale tomorrow, but if that doesn't pan out, I'm definitely waiting until I can get one. Do you think 1/10th gr. resolution is going to be good enough? I'll look for better, but cost is certainly a factor. Later.
Dave