In my opinion requiring limits on Peak Gaussian Factor, or Tailing Factor, ect. are holdovers from 30 years ago when peak integration software was not as robust as it is now. Current integration software is able to handle just about any peak shape if you get the settings adjusted.
Another thing that is somewhat of a holdover is requiring tight average response factor limits for calibration curves or 0.995 or higher r squared values for calibrations lines and frowning on the use of a quadratic fit. The EPA with the newest Drinking Water methods have began to accept calibration fits that were once questionable including curve weighting, with the requirement of tighter requant values. The new requirement makes more sense where after calibration you must recalculate all the points on the calibration and the point at or below the Minimum Reporting level must be within +/-50% and all other points within +/- normal calibration check limits (anywhere from 30% to 10% depending on the method) with the caveat that nothing is reported outside the upper and lower bounds of the calibration standards. This I believe gives good quality data. Before we just had the requirement that you do a linear calibration with a 0.995 or better fit and call it good, but I have seen a calibration of 0.5ppb to 500ppb give a curve fit of 0.998 but when recalculating the low standard the calculated value was 2.0ppb or 400% deviation. This may work well when you have mostly samples with detects near the mid point of the calibration but if you have most samples that are non-detect, just how sure can you be it is not really there?
As long as you get a good calibration and recalculation is accurate and reproducible, then peak shapes should not matter unless it leads to co-elutions, but that should be evident as a problem in the recalculation test.
