Hi Clemence,
We would be very keen to show you the code for the error propagation as we hate the idea of iolite being a black box. However, the code is spread between a bunch of C++ files, which makes it more difficult but not impossible. That said, the process is written up in Paton et al. (2010) and explained in more detail here: https://notes.iolite.xyz/posts/excessuncertexplorer/
The python of that plugin has the same calculations (except that there was a minor error that was recently brought to our attention and we're currently working on a fix).
So, all that is to say that we're very keen that people understand how iolite is propagating the uncertainty. However, the Rb-Sr DRS doesn't use the full error propagation routine of iolite (at least in the version available on GitHub... people may have edited their own copies of this DRS). Instead, it uses a combination of individual point uncertainties (calculated using the combination of Poisson uncertainty and Johnson-Nyquist noise, as described in Cruz-Uribe et al 2023) and the stats of the values within the each selection.
We're keen to improve the uncertainty propagation for this sort of analysis, so any suggestions you have for us would be greatly appreciated. If you need any help coding up a solution, and would prefer to discuss it off the forum, please feel free to email iolite's support email address.
Kind regards,
Bence