How do I minimize the standard deviation of an error?

suppose that, weight scale 'A' has a mean of error equal to 0%(unbias) and standard deviation of error equal to 1% of the weigth that scaled

weight scale 'B' has a mean of error equal to 0%(unbias) and standard deviation of error equal to 2% of the weight that scaled

if I want to scale 1 kilogram, how to divide the weight for each weight scale in order to reduce the standard deviation of error to the minimum?

note: standard deviation of error define by percentage of the weight scaled. that means, the heavy the weight is, the bigger standard deviation of error's became.