In such a case it is impossible for the additional attributes that accompany the original instance to be effective causes of the ruling, while the other attributes are absent; those additional attributes are peculiar to the original. Had they been causative, the ruling could never be found outside that original, so their absence is of no consequence, whereas in the parallel instance the presence of the shared attribute may well be affirmed.

As for the possibility that some other independent attribute might serve as the cause, this must be ruled out—either by a decisive proof, by an evident indication, or by the presumption that what is not proven does not exist. The analogy will then be certain or, at least, probable.

The discussion concerns a situation in which the possible categories of the effective cause have been exhaustively enumerated and causation has been denied for all of them save one. Such denial may either reject their causal role from the outset or negate causation altogether. The first sort still requires the second in the specific instance, whereas the second covers that instance and any other besides.

They say:

‘Even if nothing else could serve as a cause, it is still conceivable that this attribute should be a cause only for the specific case, not for the class in general.’

We reply: This is essentially their first objection, which assumes that the ruling is established by the very locus itself, not by an external factor—that is, by a narrowly confined cause limited to the text at hand.

They further contend:

‘To demonstrate that this attribute entails the ruling and that the ruling follows from the attribute’s universal nature—even though that may seem far-fetched—dispenses with the need for analogy.’

We answer: There is nothing far-fetched about it. Whenever it is shown that the middle term entails the major term, that middle term may be taken as the common description shared by the original and the subsidiary case, and the ruling necessarily follows.

As for their statement that ‘analogy can be dispensed with,’ we concede that it can; yet analogy is cited here for clarification, so that the branch case may be pictured alongside its counterpart. A universal exists as a universal only in the mind, not in the external world; once its realization in the external world is known, it becomes easier to find a counterpart. Moreover, the example may facilitate the establishment of causation—indeed, at times it cannot be established without it. When the other methods for proving causation, such as concomitance, suitability, and the like, are combined with the technique of elimination and classification, the analogy may attain the level of certainty.