Is there any way to test the accuracy of small scales using common household items? I have two electronic scales. One weighs 0 – 7000 grams in 1 gram increments. The other weighs 0 - 100 grams in .01 gram increments. Is there any way to test the accuracy of these units to within 1% without purchasing a set of calibration weights? The two things that come to me off the top of my head which could be used are coins and water. However, I can’t think of any way to precisely measure water. I assume that the weights of US minted coins must be extremely invariant—would this be the best way to do it? Would I have to use freshly-minted coins? Do worn coins weigh measurably less than new ones?
Aluminum foil is of quite consistant thickness and purity. You’d need to measure your roll with a tenth mic (micrometer with .0001" resolution) then you could make your own weight set by measuring out the fairly large areas.
Coins would be perfect. If you’re using coins dated in the last ten years or so, the wear would be minimal. Optimally, use the newest coins you can. Weigh more than one to see if there is a variance.
Cents(since 1983) weigh 2.5 g
Nickels, 5 g
Dimes, 2.27 g
Quarters, 5.67 g
Halves, 11.34 g
Hey—I think I found my answer. According to the Treasury Dept., a 50-State quarter weighs 5.670 grams. When I multiplied this by 40 and divided by 28.35, it came out to exactly 8.000. This can’t be a coincidence. So a roll of 50-State quarters should weigh exactly half a pound, plus the weight of the wrapper. But this begs the obvious question—is there any variance whatsoever in coin weights for a given coin?
ETA: Looks like Samclem beat me to it while I was still typing. 
I know samclem already answered, but I’ll add I don’t think you’ll see much variance between coins. My brother sells and services scales and weighing equipment, and one of the types of scales he works with are counting scales. These are made to count small items by weight. You weigh one part, dump a few hundred in the bin on the scale and it tells you how many pieces are in the bin.
My understanding is that they have a bunch in these in service in Las Vegas, where they are routinely used to count coins. If there was much variance from coin to coin I can’t imagine that would be the case.
US paper currency weighs 1 gram per bill.
Thirding coins. Years ago, my company used scales to count coins, and many of us still use scales on the registers when buying back rolls of change from customers.
Roll of quarters: .5 lb
Roll of dimes: .25 lb
Roll of nickels: .44 lb
Roll of pennies: .29 lb
There is some variance due to coin wear and to the use of different style rolls, but these are pretty consistently correct for our purposes. It’s a quick way to find a roll of quarters that some idiot has shoved nickels into while thinking that they’ve come up with a new way to scam us.
Coins, YES! and more…if you get a piece of wire, and calibrate that against your coin standard, you can chop up the wire into fragments of whatever amount of mass. Yeah, cumulative error comes into play. As mentioned upthread, this can be ameliorated by doing a bunch at once and taking an average.
Your other choice is to take random items of metal and weigh them somewhere that uses a calibrated scale and then use the items for checking your scale. So long as you don’t need ISO traceability , or to maintain legal saleability by weight you’re OK. Don’t forget that scales often vary over different weight ranges by non linear amounts. Checking the far limits of the scale may lead you to think your scale is good but it may contain an unacceptable variance. Calibrating in weight increments of 10g you may see something like this on your scale.
Linear 10 20 31 41 52 62 73 83
Non Linear 10 20 31 40 51 60 67 80