Lagrange and the ideal of purity of methods