A data table of fugaeity coefficients for H20-CO 2 mixtures (Ryzhenko and Malinin, 1971) was analysed by least squares to study nonideality effects assuming various solution models after Thompson (1967). The dependent variable, D i ~ ~Ziex/ (1-Ni) ~, was calculated for 743 sets of selected data with a~ ~ 475.3 eals, and an estimate of the maximum error in D of 730 eals. The regular solution model yields; for the residuals of i D, a~ ~ 165.4 calories and WG ----3116.7 -k 9.3117 X 10-~P --3.5528 T where W G is in calories, P in bars and T in ~ The subregular solution model did not improve the fit significantly on the 50 % level of confidence.
Single terms of the nature of T 2, TP and T/P etc. were added to the regular solution model and the most significantly (99.9%) improved fit with four parameters was: aR ----133.3 calories, and Wa ~ 1243.44--1.68096 P--1.3783 T-]-1524.14 PIT. The relative large size of the maximum error in D, coupled with the size of a n such that a~ < dD/4, suggests that: (1) Other types of solution models with more than three parameters are propably not warranted e.g. quasi chemical model, subregular model, and higher order Margules expansions. (2) Because the type of model chosen controls the P-T dependence of nonideality, extrapolation would not be beneficial with this fitted three parameter regular solution model.