Digital Precision Scale (20g Max / 0.001g Resolution)

I own a small set of calibration weights with weights ranging from 10 mg (1/100 g) to 100 g. It's a very cheap set that is not certified in in any way and I don't expect them to be up to the standards of professional calibration weights. I had never had the opportunity to test these weights before I bought this digital scale, so it was just as much a test of the weights as it was of the digital scale. I figured that the scale with a 0,001 resolution and the wide range of weights would give me a fairly good idea on the accurracy of both.
I used the 10g weight that came with the scale to calibrate the scale before I started the test. I used the weights in the 10 mg-20g range as that is the max load for the scale. Below are the results I got. First column is the weight used and the second is what the scale read.
10 mg - 0,010
20 mg - 0,020 I've got 2x 20 mg weights
20 mg - 0,020
50 mg - 0,052
100 mg - 0,100
200 mg - 0,200
200 mg - 0,200
500 mg - 0,500
OMG. Apart from the 50 mg weight the results from the mg weights were spot on. I did the test several times with all the weights in different order. A few times the read on the scale fluctuated a little above or below the supposed weight, but never more than +/- 0,002. As i repeated the test several times, the results for the most part turned out to be correct to the third decimal. I also weighed all the mg weights together and the result for the combined weight of these were 1,101, just +0,001 above what it should be. Finally I again weghed the 10g weight that was included and the scale stiil read a perfect 10,000.
The results from the 1g-20g were a little more off but it's hard for me to judge if that's because of the weights used or because of the scale.
1 g - 0,985
2 g - 2,001 I lost the other 2g.
5 g - 4,991
10 g - 9,978
20 g - 19,961 2x 20g also
20 g - 19,962