where the shared attribute is self-evident, there is no need to mention the original case.

Analogy, then, takes one of two forms: either (1) by stating the common factor, or (2) by dismissing any relevant difference. The common factor is either the very cause (ʿillah) itself or the proof that indicates it. In the second form—analogy by canceling the differentiator—this cancellation functions as the middle term of the syllogism.

Thus, when one says, “This is equal to that, and whatever equals an equal is itself equal,” the notion of equality is the middle term, and “canceling the differentiator” is simply another name for that equality. Likewise, when someone declares, “There is no difference between the branch and the original except for such-and-such, and that difference is inconsequential,” it is tantamount to saying, “This equals that, and the ruling that applies to one equal applies to the other.”

They further contend, “Every line of evidence that seeks to show that the shared attribute is a cause of the ruling is merely conjectural.” We reject this claim, for it is a sweeping generalization offered without proof. Moreover, the very evidence used to establish the causal efficacy of the shared attribute is identical to the evidence that establishes the truth of the major premise. Whatever proves the major premise in the syllogism of inclusion likewise proves the causality of the shared attribute in representative analogy, whether the proof yields certainty or only probability. In representative analogy, the shared attribute is the middle term: the attachment of the ruling to it parallels the attachment of the major term to the middle, and the attachment of the middle to the minor parallels the linkage of the shared attribute to the branch—that is, the establishment of the cause within the branch.

Accordingly, when the shared attribute—called variously the common factor, the cause, the proof of the cause, the operative factor (manāt), and so on—is firmly present in the branch, this guarantees the truth of the minor premise; and when the ruling is inseparably attached to that attribute, this guarantees the truth of the major premise. Mention of the original case is employed only to substantiate one of these two premises.

If the analogy proceeds by canceling the differentiator, a definite original case is indispensable, for the point held in common is the equivalence of the two—i.e., the negation of any distinguishing factor—and that equivalence itself is the middle term. But if