Overall measurement uncertainty will depend on the nature of the sample. For a highly homogeneous material like a gas or liquid, measurement uncertainty will be limited only by instrument parameters. For d13C it should be better than 0.1 o/oo. Measurement uncertainty of actual samples will depend on sample homogeneity and may therefore be higher, for example, 0.15 o/oo. For d15N, overall sample measurement uncertainty tends be higher than that for d13C. 1. N2 is not as easily ionised as CO2. 2. There is typically less N than C in organic conpounds.

That's perfect professor Meier-Augenstein but these limitation are for five replications. I repeat each of my analysis for three times and for this number of replication, I want to know the exact range if you can help me.

Thank you very much

There are no fixed values for measurement uncertainty other than instrument specfications for repeat analyses of homogenous gases.

To repeat what I said before, overall measurement uncertainty depends on the nature of your samples and performance of your instrument from EA to IRMS. Under optimal conditions, measurement uncertainty for analyses in replicates of at least three should be 0.1 and 0.15 o/oo (or better) for d13C and d15N, respectively.

Measurement uncertainty given as RSD should always be 1 % or better. A d13C value of -100 o/oo (hypothetical example) should have a measurement uncertainty of 1 o/oo (or better) while for a d13C value of -10 o/oo it should be 0.1 o/oo or better.

Minor point - the standard deviation of three replicate analyses by EA-IRMS will give you an indication of the precision of the measurement not the overall measurement uncertainty. Measurement uncertainty will include other factors beyond precision - e.g. the uncertainty introduced by the use of reference materials for normalisation of results to the reporting scales. The Good Practice Guide for IRMS from the FIRMS Network (freely available from http://www.forensic-isotopes.org/gpg.html) summarises potential sources of measurement uncertainty for EA-IRMS.

The values given by Wolfram are reasonable as the standard deviation of replicate analyses of C and N isotope ratio by EA-IRMS - and as he's said, the exact numbers will depend on the properties of the sample as well as on the performance of your specific instruments.

A very important point Philip. When I mentioned overall uncertainty that encompasses instrument uncertainty and sample uncertainty, I took it for granted it was understood that RM uncertainty would of course also have to be accounted for once measured delta-values had been scale normalised on the basis of the contemporaneously analysed RMs serving as the two scale anchors.

Arndt Schimmelmann and I have recently published an article on the subject of proper use of RMs.

Article A guide for proper utilisation of stable isotope reference materials