ABSTRACT
18Three rules, the Titius–Bode, the Dermott rules, and a classical linear molecular quantitative structure–property relationship, are revisited discussing some of their main characteristics and revealing up to which level the models have real predictive power or simply a descriptive one. A careful choice of the experimental values to be included in the model seems to be essential to the usefulness of a relationship. Furthermore, a predictive relationship is not always free from flaws.
