Thus, Comte sees clearly that the complexity of the chemical phenomena prevents us from expressing their relations in a form which allows of the use of mathematical analysis. But none the less, in this science as in the preceding ones, he persists in making the experimental method a mere starting-point. The experimental method furnishes the data which it alone can supply. But these data are afterwards elaborated without its intervention. The scientific ideal in chemistry, as in physics and in astronomy, is _to substitute as much as possible rational prevision to experimental verification_. Science always seeks to _deduce_ the greatest number of consequences from the smallest number of data, and the smallest number of data in this case are the properties of simple bodies. Deduction will establish _a priori_ what the properties of a given combination of two simple bodies, or of two complex bodies will be. In the name of this scientific ideal, Comte reproaches the chemists with the superabundance of their analytical work. In default of a rational conception of chemistry they do not make their work bear upon the necessary points. What is the use of studying such and such a body, placed in such and such conditions, in an arbitrary way and according to the fancy of investigation? The progress of chemistry should consist far less in the acquisition of new materials than in the systematisation of those which we already possess. Chemistry is to-day as rich in details as it is imperfectly constituted as a science.[146] Its present state in no way gives an idea of what its normal state will be.