J. C. Gower introduces a general coefficient to measure the similarity between two sampling units, which is shown to be positive semi-definite (PSD) when there are no missing values. This property is crucial for representing the sample in multidimensional Euclidean space and establishing inequalities among similarities for three individuals. The coefficient is extended to handle hierarchies of characters. The paper discusses the calculation of similarity scores for dichotomous, qualitative, and quantitative characters, and explores the relationship with other similarity coefficients. It also addresses the issue of weighting character scores and the handling of hierarchic systems of characters, providing methods to ensure that primary characters are given more weight than secondary characters. The PSD property is essential for numerical methods and Euclidean-based analyses, and the paper includes an appendix detailing the properties of PSD matrices.J. C. Gower introduces a general coefficient to measure the similarity between two sampling units, which is shown to be positive semi-definite (PSD) when there are no missing values. This property is crucial for representing the sample in multidimensional Euclidean space and establishing inequalities among similarities for three individuals. The coefficient is extended to handle hierarchies of characters. The paper discusses the calculation of similarity scores for dichotomous, qualitative, and quantitative characters, and explores the relationship with other similarity coefficients. It also addresses the issue of weighting character scores and the handling of hierarchic systems of characters, providing methods to ensure that primary characters are given more weight than secondary characters. The PSD property is essential for numerical methods and Euclidean-based analyses, and the paper includes an appendix detailing the properties of PSD matrices.