to make a long story short:
My comment is strictly correct only for calcite using NBS 19 as a reference material, of course.
Not for other carbonates and not for mixtures.
the long version:
Indeed my comment
> You can run your carbonates at any temperature you want as long as your temperature is _constant_ (in case of 18O analysis) for your standards and your samples.
can be misunderstood in a way that this refers to any carbonate (calcite, dolomite, ankerite etc.) or even carbonate mixtures. That, of course, is not the case. The original question from Andrew did not point out what carbonate should be analyzed. However, in most cases labs deal with calcite and since NBS 19 is also a calcite this simplifies things to the IT principle and constant temperature. Willis answer already pointed out the problem with 18O and carbonate mixtures in more detail. Running mixtures you have to take care about the different fractionation factors at different temperatures for the minerals in the sample and have a look into the already cited Rosenbaum and Sheppard (1986) or Sharma and Clayton (1965). In case of such mixtures you should also think about methods to react your minerals separately and you pointed out this in detail. Thanks.
What I am by far more concerned about is the fact that the real composition is sometimes not known in a lot of cases. As long as samples come from limestones or fossils you will have an idea about calcite, dolomite or aragonite but for other sediments you have to run XRD on your material before any isotope analysis. Obviously it should be the first step to know what is analyzed but I have seem a couple of projects being not aware of this fact. Just to keep this in mind.
Thanks for the concrete references.
Are you aware of any fractionation studies of minerals associated with portland cement reaction type?