Methods for correlating experimentally observed data to the known physical, chemical, or molecular properties of a substance or a mixture of substances are discussed. These correlations, once drawn, can be used to estimate values not experimentally obtainable. The correlation methods may be theoretically, semitheoretically, or empirically based. The theoretically based correlations, most often rooted in thermodynamics, are usually preferred.
Data correlation may result from the empirical use of basic mathematical and numerical techniques, nomographs, or reference substances. Alternatively, the methodology may incorporate group contributions, structure‐activity relationships, molecular connectivity indexes, and graph theory. A significant amount of investigation is devoted to the assessment of the degree to which mathematical constructs faithfully mimic real systems.