If you consider that your partial extractions data are representative for a given composition of a geologic material, you should also treat them as the data which are subject to rules of CDA. This means that you could interpret the ionic compositions of your extractions (solutions) as closed data sets in the same way as we do this in hydrochemistry. The issue of representativeness of extracts for the composition of the rocks, however, is a matter of rock-forming minerals reaction kinetics. because extraction is not proceeding over geologic time.

Thanks Krysztof - I agree that the solutions themselves have a closed composition (not the we usually include the water content!), but when calculated on the basis of regolith mass (i.e a ratio, albeit a simple one), the concentrations are not necessarily constant-sum. I'm not a good enough mathematician to figure out the degree to which closure might be relevant for such data, or whether they are geometrically the same as total compositions.