Epistemologia (ed. 2)

To unite divided parties it is best to proceed from the point wherein they agree. Philosophy is a science; in this all descriptions of philosophy are as agreed as they are divided in determining the object of this science. This division, may it not have arisen, because the conception of that science, which they unanimously asserted philosophy to be, had not yet been wholly developed? And may not the determination of this one characteristic, wherein they all agree, suffice to determine the conception of philosophy itself? A science has systematic form. All propositions in it are connected in one single fundamental proposition or principle, and unite with it to form a whole. This is universally conceded. But does this characteristic exhaust the conception of a science? Supposing somebody were to build up ever so systematic a natural history of certain spirits of the air, on the unproven and unprovable assumption, that such creatures exist in the air, with human passions, inclinations, and conceptions, should we call such a system a science, no matter how closely its several parts might be connected with each other into a whole? On the other hand, supposing somebody were to utter a single proposition—a mechanic, for instance, the proposition that a pillar erected on a horizontal base in a right angle stands perpendicular, and will not incline toward either side, however far you extend it into infinity—a proposition which he may have heard at some time and approved as true in experience: would not all men concede that such a person had a scientific knowledge of the proposition, although he should not be able to evolve the deduction of his proposition from the first fundamental principle of geometry? Now, why do we call the fixed system, which rests on an unproven and unprovable first principle, no science at all, and why do we assert the knowledge of the mechanic to be science, although it does not connect in his reason with a system?