Thank you for your good book and elucidations on isogeochem.
This is a stimulating discussion.
Uncertainty propagation is a good practice in chemometric (and isotope
analysis) but you'll know that "propagation" is automatically included if
you make a least squares regression (or any inferential statistic test) with
ALL standard measurements (repetitions)...ALL?...and the 2sigma rejected
Moreover, what is the sense of error propagation if you don't consider other
variables (e.g. drifts on mass spectrometer or on EA-devices) during long
Other than, I think that the real (and crucial) problem is the comparison
between "off " and "on" line methods on the variuos materials (NBS-127,
By the way, the "exact" value of NBS-127 is really 9.3+-0.3 permil VSMOW?
Published on-line mean value is 8.6+-0.2. And "error propagation"? My point
of view is that the results are significatively differents.
With Kindest Regards.
University of Parma.