by
lmh » Wed Mar 24, 2021 4:45 pm
You need to document whether your limits are inclusive or exclusive. If 0.3 is the limit, is a peak of 0.3 excluded (just) or included (just)?
The motivation for excluding tiny peaks is a historical feeling, dating back to the days of wobbly baselines, that it's wrong to reject something based on the sum of a large number of very tiny bumps, which might just be noise. This is a good justification, but it was never intended as a justification for ignoring definite contaminants.
If your motivation for worrying about the extra 0.3% is that 1% summed contaminants is the threshold at which disaster strikes, then the answer to the rounding question is this:
If you have only three, well-defined contaminant peaks that sum to 1.25% of the total, are you excited about the idea of explaining to a regulatory body why 1.25% < 1.0% based on what might easily look like creative use of rounding?