Even if we suppose that syllogistic reasoning does not depend on analogical reasoning, and that knowledge of universal propositions does not depend at all on knowledge of any particular instance, one still cannot say that knowing a universal, together with knowing that some of its instances exist in external reality, is less complete than knowing it without knowledge of that particular instance. Knowledge of the particular only makes that knowledge more complete. Thus, the form of inference they reject is more complete than the one they affirm.

Know that in “metaphysical logic”—indeed, even in “natural logic”—they altered some of what Aristotle had stated. Yet their additions in metaphysics are better than Aristotle’s own statements, for I have examined both accounts. In metaphysics, Aristotle and his followers are far, far more ignorant than the Jews and Christians. In natural philosophy, however, most of what he says is sound. As for logic, his statements on it are better than his statements on metaphysics.

Their disparagement of “analogical reasoning” comes only from their later thinkers, who observed that jurists commonly employed it. Jurists frequently use it with content that yields only probable knowledge. Here, however, the probability arises from the content, not from the form of the inference. Had they cast that content in syllogistic form, it would likewise have yielded only probable knowledge. These thinkers, however, supposed that the weakness lay in the form. They therefore deemed their own form of inference productive of certainty, and the jurists’ form productive only of probability. They illustrated this with examples drawn from speculative theology, seeking to establish their contention: “The speculative theologians argue against us using merely probabilistic inferences.” One example they gave of an argument used against them was that the celestial sphere is a composite body and therefore originated in time, by analogy with a human being and other generated beings. They then proceeded to discredit this inference. Yet it was only