Thus not only are moral laws with their principles essentially distinguished from every other kind of practical knowledge in which there is anything empirical, but all moral philosophy rests wholly on its pure part. When applied to man, it does not borrow the least thing from the knowledge of man himself (anthropology), but gives laws a priori to him as a rational being. No doubt these laws require a judgement sharpened by experience, in order on the one hand to distinguish in what cases they are applicable, and on the other to procure for them access to the will of the man and effectual influence on conduct; since man is acted on by so many inclinations that, though capable of the idea of a practical pure reason, he is not so easily able to make it effective in concreto in his life.

A metaphysic of morals is therefore indispensably necessary, not merely for speculative reasons, in order to investigate the sources of the practical principles which are to be found a priori in our reason, but also because morals themselves are liable to all sorts of corruption, as long as we are without that clue and supreme canon by which to estimate them correctly. For in order that an action should be morally good, it is not enough that it conform to the moral law, but it must also be done for the sake of the law, otherwise that conformity is only very contingent and uncertain; since a principle which is not moral, although it may now and then produce actions conformable to the law, will also often produce actions which contradict it. Now it is only a pure philosophy that we can look for the moral law in its purity and genuineness (and, in a practical matter, this is of the utmost consequence): we must, therefore, begin with pure philosophy (metaphysic), and without it there cannot be any moral philosophy at all. That which mingles these pure principles with the empirical does not deserve the name of philosophy (for what distinguishes philosophy from common rational knowledge is that it treats in separate sciences what the latter only comprehends confusedly); much less does it deserve that of moral philosophy, since by this confusion it even spoils the purity of morals themselves, and counteracts its own end.