# Calculating/estimating volatiles based on other representative samples

I am currently working on a set of geochemical data representing calc-alkaline igneous intrusions from central Canada. The samples were analyzed in ~2013-2014. These samples have been tested for LOI wt%, but not H2O, CO2...etc. Historical sampling from the same area representing the same suite of rocks (but different sample locations) have results for H2O and CO2 wt%. I need H2O wt% values for my samples in order to complete a calculation for viscosity of the intrusions, but I won't be doing any more testing. Am I able to estimate H2O in my samples using the historical data? And if so, how would I go about doing that? And how would I correct for (if possible) the difference in quality of the measurements due to more advanced and accurate/precise instrumentation?

If you are sure that the samples for which you have H$$_2$$O data are from the same intrusion than yours, then I don't see a problem for using these values for viscosity modelling, as long as it is clearly mentioned in the paper. After all, a sample is supposed to representative of the rock unit, and science is done by standing on the shoulders of giants!
If you have several historical samples with different H$$_2$$O contents, this gives you a range of water contents. Do you intend to use a model like Giordano et al. (2008) ? If so, you also need a temperature estimate, which you may have by geothermometry. These temperature estimates have an uncertainty, see Putirka (2008) for a review. For example, if you use the two-feldspar thermometer, it gives you $$T$$ to $$\pm$$30 °C.
So, what you could do is to model two scenarios: one "cool and dry", with the lowest $$T$$ and H$$_2$$O content, which gives you the highest viscosity; and one "hot and wet", with the highest $$T$$ and H$$_2$$O content, which gives you the lowest viscosity. This way you can be quite confident that the viscosity was somewhere between these two values. Usually a $$\pm$$30 °C interval will change your result by half a log unit, and a $$\pm$$0.2 wt% H$$_2$$O interval will change it by one log unit, so you'd get something like $$\log \eta = X \pm 0.75$$ Pa s, which is not that bad.