Consider I have a function:
$PV_m^3 - PbV_m^2 +aV_m+ab=RT$
In this example I have measured the uncertainties in $P$ and $T$ experimentally and the errors in $a$ and $b$ can be assumed to be zero. How would I go about estimating the error in $V_m$?
Normally, I use the variance route:
$s_a = \sqrt {s_b^2(\partial f/\partial b)^2 + s_c^2(\partial f/\partial c)^2+...}$
where
$a = f(b,c,...)$
Unfortunately, I'm not sure how to approach this with $V_m$ as I don't know what $V_m=f(P,T)$ looks like.
