Towards the compatibility of stable isotope data: review of related concepts, recent progress with reference materials and current challenges.

Comparing and combining stable isotope datasets from different laboratories and different years is essential for many research areas, such as isotope hydrology, greenhouse gas observations, food studies, isotope forensics, palaeo-reconstructions, etc. Data compatibility (i.e. the ability to combine data) is related to the data quality. The prerequisite for data comparability is data normalization to a common stable isotope scale (often referred to as calibration) based on reliable reference materials (RMs) with accurately assigned values and uncertainties. Still, that does not guarantee the data compatibility (mutual agreement). Albeit metrological concepts related to data compatibility and measurement uncertainty have been developed and applied to analytical chemistry in general, these concepts have not yet been fully applied to stable isotope research. This can affect daily calibrations, analytical data and, therefore, data compatibility. In addition, IRMS users often prepare different laboratory standards themselves. Thereafter, users should then understand the contemporary concepts used for assigning RM value and uncertainty, as well as the limitations and potential problems associated with RMs. The history of RMs, preparation reports and also some problems in the past provide lessons to be learned. These include the C drift of LSVEC (the second anchor on the C scale before 2017), revisions to the value assignment principles, the introduction of replacements for LSVEC, related disputes and the potential underestimation of uncertainties for secondary RMs. The review describes metrological concepts related to isotopic scales, RMs and calibration hierarchies and data compatibility. The main RMs and their uncertainties are reviewed through the lens of metrology concepts. Additional focus is given to the VPDB scale for C and issues of scale discontinuity, which can significantly reduce data compatibility in C. The given examples of value and uncertainty assignment for RMs should be viewed as an example of value and uncertainty calculation in daily practice.