Coins are not reliable check weights.
They only have to be close enough to work in vending machines and parking meters.
Wear, new sandwich coins vis real silver coins, and variations through the years make it impossible to use coins as reliable check weights.
The best Field expedient check weight I know of is factory jacketed bullets.
If Hornady or Sierra or Speer sells you a box of 55 grain .224" bullets?
You can go to the bank it weighs close enough to 55 grains for GooberMint work.
The only problem is, accurate at 55 grains doesn't mean it is accurate at 3.5 or 5.0 or 23.5 grains.
You really need a set of check weights if you really want to worry about RCBS's calibration being 100% right at all ranges of the scale.
rc
RCModel, I think you mis-read the OP (or I did).
He weighed each of three coins. No matter if they weighed the same as each other or not. They weight what they weight.
He then added the three numbers together
Then he weighed all three coins together (same coins)
The combined weight did not equal the sum of the three weights.
That adds up to a scale that needs attention of some kind in my book.
You (very correctly) stated
The only problem is, accurate at 55 grains doesn't mean it is accurate at 3.5 or 5.0 or 23.5 grains.
and that seems to me to be the crux of the O.P.'s question.
On the other hand, .3 grains is not a large variation when you consider it as a percentage of the measured weight. (Coins are a lot heavier than most powder charges).
If I were the O.P. I would get a few toothpicks or dried beans or something close to the weight of my anticipated powder charge and perform the same calculation. That way I would be testing the consistency of the scale through the range in which it will be performing for me.
Note: These toothpicks or beans would not be check weights over any period of time. I would not expect them to refrain from absorbing water from the air or getting eaten or used.
Lost Sheep