I have a doubt about uncertainty of a measurement. I have a sample of a specific material which is my reference material and it is calibrated and NIST traceable. This sentence is in its data-sheet:
The estimated overall uncertainty (at 95% confidence i.e. 2σ) is 0.02 + 0.2% of the reported value in the table.
Now suppose that one of these value is equal to 1.214. What I do not understand is the way I have to express my error according to the sentence:
1) 1.214 ± (0.02 + 0.002*1.214)
2) 1.214 ± sqrt[ 0.02^2 + (1.214×0.002)^2 ]
3) Choose for every reported value, the greatest between 0.02 and 0.2%*value
The second method returns a smaller uncertainty with respect to method 1, but still not sure what to do.