It's funny you should ask, because I just got done doing some experiments with my scale; someone had suggested that I make sure my inconsistent powder throws are real, and not just my 10-10 scale misbehaving.
I checked 3 things:
- Accuracy (how close to the bullseye)
- Repeatability (the scale equivalent of group size)
- Resolution (the scale equivalent of how wide each scoring ring is)
For accuracy, I used my set of RCBS check weights. I weighed each check weight, and several combinations of check weights, recording the scale's indicated weight vs. the check weight's claimed weight. For example, the 0.5gr weight indicated 0.53 on my scale, so my scale is off +0.03 at 0.5gr, and the the 10gr weight indicated 10.10, so my scale is off +0.10 at 10gr.
The worst error at any weight was 0.10, so accuracy is +=0.10gr.
For repeatability, I measured the same weight many times, starting by rezeroing the scale. In five weighings of the 50gr weight, the spread of indicated weights was 0.05, so repeatability is OK. (I think repeatability is usually measured with standard deviation, but I didn't want to take the 20 measurements it takes to get enough samples for a reliable standard deviation).
For resolution, I made some 0.05gr check weights out of pieces of paper (my RCBS check weight set only goes down to 0.5gr). I cut a strip of paper about 1/8" long and of uniform width, weighed it, and did the math to figure out what lengths of paper to cut to get 0.5 and 1.0gr pieces. The pieces of paper will change mass with humidity, so they're only good for tonight's experiment, but that's alright. Maybe next time I'll use some magnet wire so I can add the weights to my set.
I then put 10.1gr of check weights on the scale (the 10gr RCBS weight and two of my 0.05gr weights), set the scale to 10.1gr, and then turned the zeroing knob until the balance indicator was centered. Then I removed a piece of 0.05gr paper with tweezers and saw if the balance moved noticably, then put it back on and saw if it went back to zero. Then I added an extra piece of 0.05gr paper to see if the balance moved noticably, then put it back on and saw the beam move back to zero.
I repeated the resolution experiment at different check weights to see if the resolution of the scale differs at different weights.
I found that differences of 0.05gr are noticable at all weights weights, but the beam doesn't move much. The beam moves just about the same amount up or down for a 0.05gr change in weight no matter how much weight is on the pan. I say "just about" because for very large masses, the beam does move less and its movement becomes harder to pick out.
Since I can reliabily tell a difference in weight of 0.05gr throughout the scale's range, but I don't think I'd be able to tell any smaller difference in weight, the scale's resolution is 0.05gr.
I hope this all makes some sense.