Method paper discussion

15 Apr 2016

Dear all,

You have probably all heard about the arXiv, where some scientific communities (physicists, mostly) upload and discuss pre-print manuscripts. In my opinion, there are major benefits to this way of sharing scientific results, and I wonder what would happen if something like that became the default method for communicating new findings in our community.

< gets off soapbox >

So, in the spirit of open discussion, I'd like to share a paper we just submitted to GCA. It's a Δ47 method paper where we try to address issues of inter-lab discrepancies, so I think it can be of interest to many of you. I'm very interested in any comments you might want to share, publicly or privately.



PS: I want to publicly thank Andy Schauer, whose talk at the latest clumpy workshop provided the original motivation for this work.


Supplemental material:

Cedric's picture
Joined: Feb 6 2011
Hi Mathieu, Thanks for

Hi Mathieu,

Thanks for sharing this MS, I just went through it. Very interesting! It feels like you may have cracked the inter-lab reproducibility issue here, at least to some extent. And it reinforces my personal opinion that we should all settle on the same parameters, and use them systematically, the way we use Wang et al 2004. This is what Easotope was built to do. I appreciate that you put some recommendations for what parameters to use in your paper, and if we all agree on them, then I think we can easily change the ones in Easotope to match those.

Actually, thinking out loud, the mathematical effects for sample processing that you describe probably also nicely explain why different labs obtain different acid correction factors between 25-90˚C. Maybe something to consider in your paper?


Mathieu Daëron's picture
Mathieu Daëron
Joined: Sep 17 2014
Re: acid correction

Re: acid correction factors

I'm not sure. Unless I'm mistaken, acid correction effects have mostly been measured by analyzing (often within the same session) the same materials at different temperatures. In that case, the difference in bulk composition is going to be minimal (about 2 permil in δ18O), which should produce only a small effect (0 to 2.7 ppm according to figure 5 in the manuscript).

- Mathieu

Kate Huntington's picture
Kate Huntington
Joined: Sep 21 2011
Thanks to Mathieu for the

Thanks to Mathieu for the kind words and for sharing the submitted manuscript.  It is great he was interested in Andy’s presentation and followed up.

We had come to essentially the same conclusion regarding the explanation for Andy’s observations and had already started incorporating it into our data processing and submitted papers. One such submission is a synthetic carbonate manuscript by my PhD student Julia Kelson, which processes data using Brand et al. lambda value and shows that 58 carbonates precipitated using many different methods (including raft carbonates) all conform to a single temperature-D47 relationship (same sensitivity for both 90°C and 25°C acid). This strongly suggests that once we adopt the same data reduction methods as a community all of our calibration woes will go away; we are so close!

Since the workshop, Andy added substantially to his dataset demonstrating the influence of the choice of 17O correction scheme on D47 values. We have also shown the effect of this data processing issue on calibrations using Julia’s synthetic carbonate data in the context of previously published calibration data.

These findings are reported in Schauer et al. (in prep for RCMS; same authors as Schauer et al. 2016 clumps workshop abstract).

We note that our idea to look at 17O correction to explain Andy’s observations was inspired by Gerry Olack and Albert Colman, who observed a strong effect on D47 in 18O spiked samples (spike does not affect 17O) (AGU 2015,

As a natural extension of the work we have done independently so far, Mathieu and we are enthusiastic to move forward together collaboratively. The UW group has also been talking about this with Gerry and Albert over the past few months, and we look forward to coordinating our future efforts with them. Stay tuned!

All the best,


Kate Huntington, nee Ruhl
Associate Professor, Dept. of Earth and Space Sciences
University of Washington, Seattle, WA;

Joined: Dec 21 2011
Hi Cedric and others, I think

Hi Cedric and others,

I think that the acid fractionation is more involved than simply change in the calculation.

You should note a couple of points: first, it is not a matter of agreeing on parameters for calculation but on identifying the correct ones. Even if we all use the Santrock correction as is, Mathieu shows that the D47 we get will depend on the exact scenario of standards we each use, and therefore get different interlab (and sometimes intralab) D47 values. second, intuition doesn't work well here and one has to actually go through the full calculation ro see what changes and how.

because we standardize using both a range of d47 to correct for non linearity and of D47 to correct for scale compression, there is an endless number of posible standardization scenarios (namely, combonation of standards used in terms of their d13C and d18O). as a result, variations can occur in unexpected places.

It is likely that this calculation will account  for interlab variations in cases when different labs use different isotopic parameters in the calculation. But what happens if we use the same (wrong) parameters and different set of standards - we may or may not be different, depending on the specific set of standards.

With respect to calibration and acid fractionation: it is likely that when measured in one lab within one measurement session the calculation will be self consistent but different measurements session may or may not be, again depending on the standards. this may explain Julia's result. But acid fractionation is typically determined within one session and the change in bulk composition between 25C and 90C is small so my intuition is that it will not explain it (and unlikely to explain my results on variable 90-25c offset that varies with bulk composition when measured within one session). but as I said above, intuition is not great here and we need to do the actual calculation to check.