B16 · باب الترجيح · Preferring one proof over another: what counts as extra weight, and what does not (v2/249–265)

In short

This chapter asks how one proof outweighs another of the same kind [Overview]. Preference is an added quality, not an added proof [B16.01]. It has four grounds, the first being stronger effect and the weakest the ruling's absence with the cause's absence [B16.02]. More likenesses, wider scope or fewer features do not count [B16.04].

Because an extra witness is of the same kind as the proof itself, four witnesses do not outweigh two, and even one extra witness does not [B16.01].

In plain words: When two proofs of the same kind conflict, what makes one outweigh the other? Sarakhsī says preference is extra weight "as a quality, not as substance", like the grain that tips a scale, not an extra coin. So what could ground a ruling on its own cannot tip the balance. The grounds are four. Stronger effect; a firmer hold on the ruling, carrying it wherever the cause recurs; more roots; and, weakest, the ruling vanishing when the cause vanishes. A ground in the thing itself beats a ground in its state. More likenesses, wider scope and fewer features do not count.
Example: Four witnesses for one claimant do not beat two for the other. But two upright witnesses win over two whose character is unexamined.
Who disagrees: al-Shāfiʿī makes usufructs (the use of a thing, like living in a house) liable when destroyed, as under a contract. He claims his cause holds firmer. Sarakhsī replies that firmness lies with the Ḥanafī view, that usufructs are not liable: liability for destruction requires a like, and a contract's liability is not built on likeness.
Deeper: v1 B16.01; B16.02; B16.03; B16.04