In addition to the well-known guides, publications from the Thermodynamics Research Center of the NIST can be a useful source of information on the uncertainty of thermodynamic property data. See, particularly, Q Dong, et al. Article Uncertainty Reporting for Experimental Thermodynamic Properties†
J. Chem. Eng. Data 2005, 50, 546-550
The uncertainty caused by (im)purity of a material is simply evaluated from the reagent specification. The purity is often specified as “no less than the p %”, e.g., ≥ 98 %, which means that the actual purity is somewhere between 0.98 and 1.00. Since, as usually the case, there is no additional information about the shape of the distribution between the bounds, a uniform (rectangular) probability distribution is assumed. Then the standard uncertainty contribution u(p) is calculated by dividing the half-width of the purity interval, ap= (1.00 – 0.98)/2 = 0.01, by the conversion factor equal to 31/2 for the rectangular distribution: u(p) = ap/(3)1/2 = 0.01/1.73 = 0.0058 = 0.6 %. By root-sum-squaring, this uncertainty component is added to the other components.