The attempt to construct a fundamental taxonomy of the sciences encounters great difficulties. For instance, logicians and mathematicians disagree among themselves about the objects of their respective research; there is no agreement on the relation of theoretical physics to empirical knowledge and to mathematics. The so-called social sciences are particularly difficult to classify. They have not been demarcated by systematic considerations. In Duhem and Poincarè, general considerations are not only exemplified, but the origins of the concepts and of the problems are traced right from the initial observation of facts if at all possible.