A foundry's testing wing reports: 'No casting of ours has ever failed in service. A casting is passed for service only if the wing is satisfied it will not fail, and castings the wing is not satisfied of are broken up in the yard.' The report's reasoning is most open to criticism because it

Whether a casting has failed in service is settled out in the field, by users; the wing instead reports its own intention at the moment of passing — that it passes nothing it thinks will fail — which is a claim about the wing's confidence, not about any casting's behaviour afterwards. Since the wing's confidence being sound is exactly what is in question, the reason and the conclusion are the same statement wearing two coats. The number broken up would bear on how strict the wing is; whether those would really have failed is a question about the rejects and not about the ones in service; and a longer record would strengthen a properly gathered claim without repairing this one.