Second‐order regression of ¹³C substituent chemical shifts with Taft's sigma constants

In addition to Taft's dual substituent parameter (DSP) and DSP–non–linear resonance (DSP–NLR) models, and Charton's triple substituent parameter (TSP) equation, two second‐order regression models were tested using published data for 13 C substituent chemical shifts of the C‐4 atom in 1,4‐disubstituted benzenes. The complete data matrix consisted of 14 series with a fixed Y (C‐4) substituent, each with 14 variable X (C‐1) substituents. Second‐order (in the σ parameter or in a combination of the σ 1 and σ constants) models gave results as good as, and in some cases even better than, the DSP–NLR equation; the results of the TSP model were always worse. The whole data matrix was submitted to principal component analysis. Target testing has shown vectors of σ , (σ ) 2 and σ I constants as ‘real’ principal components.