The first is that they retract before the judgment is issued; it is not permissible to issue judgment based on it according to the general consensus of the scholars. It is reported from Abu Thawr that he held an isolated view against the scholars, saying: It is valid to judge based on it because the testimony has already been delivered, and it does not become void by the retraction of those who testified to it, just as if they had retracted after the judgment. This is corrupt (invalid), because testimony is a condition for the judgment, so if it ceases before it, the judgment is not permissible, just as if they had become immoral (fasiq); and because their retraction manifests their falsehood, so it is not permissible to judge by it, just as if they had testified to the killing of a man, then he is known to be alive; and because the judge's certainty regarding the validity of what was testified to has ceased, it is not permissible for him to judge by it, just as if his ijtihad (legal reasoning) had changed, and it differs from [the situation] after the judgment, for then it has been completed with its condition (1). And because doubt does not invalidate what has been judged, just as if his ijtihad had changed. The second condition is that they retract after the judgment and before the implementation (istifa'). One must look: if the subject of the judgment is a punishment (uqubah), such as the prescribed punishment (hadd) or retaliation (qisas), it is not permissible to implement it, because prescribed punishments are warded off by doubts, and their retraction is among the greatest of doubts; and because the subject of the judgment is a punishment, and no [certainty remains] (2) of its entitlement, and there is no way to provide restitution for it, so it is not permissible to implement it (3), just as if they had retracted before the judgment. It differs from property, as it is possible to provide restitution for it by obligating the two witnesses to pay its substitute, whereas the prescribed punishment and retaliation are not rectified by making a similar [punishment] obligatory upon the witnesses, because that is not restitution, and it is not obtained for the one who was entitled to it as a substitute; rather, it was legislated for deterrence, venting one's grievance, and retribution, not for restitution. If it is said: You have stated that if judgment was issued for retaliation, then the two witnesses became immoral (fasiq), it is implemented, according to one of the two views. We say: Retraction is greater in terms of doubt than the occurrence of immorality; for they acknowledge that their testimony is false and that they were immoral when they testified and when the judge issued judgment based on their testimony. Whereas the one whose immorality occurred later, it is not established that his testimony was a lie, nor that he was immoral when he delivered the testimony, nor when the judgment was issued based on it; and that is why if he becomes immoral after the implementation, nothing is required of him, whereas the two retractors are liable for the indemnity of what they testified to, thus they differ. If the matter testified to is property, it is implemented and the judgment is not overturned (4), according to the opinion of the jurists of the various lands. It is reported from Sa'id ibn al-Musayyib and al-Awza'i that they said: The judgment is overturned even if the right has been implemented; [because the right] (5) was established by their testimony, and when they retract, that by which it was established ceases.