A popular and widespread, though rarely ever scrutinised method of measuring a complex multi-dimensional construct y and consists of adding the scores x[subscript 1], . . . , x[subscript m] assigned to a unit in m "dimensions". These scores ("indicators") are supposed to represent the m relevant aspects of y. Interestingly the sum-of-scores method (SSM) is not only lacking a theoretical justification there is not even a clear understanding of the relevance of the correlation between the indicators for y. The paper therefore aims at relating the SSM to methods of multivariate analysis and tries to clarify the role of weights in the summation of scores.