Both before and after Kant, philosophers have promised to “put metaphysics on the sure path of a science” or drop it altogether. What they meant by this is anything but clear. Contemporary historians and philosophers of science have shown that the models of science on which those promises seem to be based have little to do with the actual practices of science, which are laden with value in form, method and goals. Though every serious study of science is in agreement about this, it hasn’t seemed to stop attempts to rid science of the normative. I argue that the older notion of humanities, rather than the younger one of Geisteswissenschaften, would be helpful against such tendencies. I discuss some differences between humanities and Geisteswissenschaften, and argue that Kant’s best discussions of science are anything but scientistic.