The Retest That Passed
You lead receiving inspection at a manufacturer of hydraulic assemblies. A supplier ships 12,000 machined pins; the certificate of conformity states a minimum tensile strength of 520 MPa. Your lab's first sample of 8 pins gives a mean of 512 MPa, with two individual results below the specification. The supplier's quality engineer responds within hours: your lab used the wrong grip geometry, their method is the one in the drawing note, and they request a retest. A retest using the supplier's stated method on a fresh sample of 8 gives a mean of 531 MPa, all individual results in specification. Production stops in three days. There is no other qualified source, and an expedited reorder is six weeks. Your procedure states that a failed lot is rejected, and that any retest requires quality manager approval supported by a documented root cause for the original failure. The quality manager is on annual leave and her deputy is covering two other sites.
1. Do you accept the lot on the strength of the retest? Work through what the retest does and does not establish.
Read the worked answer
The retest establishes that the pins pass under one method. It does not, on its own, establish which method applies, and that is the whole question.
Start with the drawing note. If the note genuinely specifies the supplier's grip geometry, your first test was non-conforming and the retest is the valid measurement - in which case the correct action is to accept the lot and open a corrective action on your own test method, not to treat the lot as suspect. If the note is silent or the supplier's reading of it is generous, then you have two defensible methods producing results on opposite sides of a limit that is only 8 MPa wide relative to a 20 MPa spread between samples. That gap is within the noise of method variation, which means the specification margin is thin regardless of who is right.
So: accept only after the method question is settled by the drawing, not by the supplier's assertion. Meanwhile, do not release to production on the strength of the second sample alone.
2. Your procedure requires quality manager approval for the retest. The deputy is unavailable. What do you do about authority?
Read the worked answer
The procedure is doing exactly what it was written to do - stopping a retest from becoming a quiet way to reverse an inconvenient result. Work around the availability problem, not around the control.
Options in order. First, reach the quality manager on leave for a documented decision; a five-minute call with a written record is legitimate and is not an imposition when a line stoppage and a six-week reorder are at stake. Second, if she is genuinely unreachable, escalate to the site operations manager and ask them to authorise a named deputy in writing for this decision only. Third, if neither is available, the lot stays in quarantine and you escalate on the record that the control is blocking a commercial deadline.
What you do not do is sign it yourself because the process is inconvenient, or book the retest as routine verification so that no approval was needed. The distinction between a retest and routine sampling is precisely the one your procedure draws, and an auditor will ask which it was.
3. Assume the method question is resolved in the supplier's favour and the lot is accepted. What documentation and follow-up should exist?
Read the worked answer
The acceptance decision has to be reconstructable a year later by someone who was not in the room.
The record should contain: the original failing results with the test method and grip geometry used; the drawing note and the interpretation that resolved it, quoted, not summarised; both sample sizes and individual results, not just means; the approval trail, including any delegated authority invoked and by whom; and a clear statement that the lot was accepted because the first method was non-conforming, not because the second result was more convenient.
Follow-up, in order of importance. Correct your own test method and add the grip geometry to the inspection plan so this cannot recur. Ask the supplier why their certificate gave no method detail - a certificate that cannot be reproduced is a weak certificate, and the next dispute will not resolve this cleanly. And review the specification margin: a 520 MPa minimum on a process that scatters by 20 MPa invites exactly this argument, and a conversation about capability is cheaper now than at the next receiving inspection.