The record arrives as a single number. Fourteen points per hundred square yards. Underneath it, a line that says PASS, and a signature you cannot read. The buyer approves the lot. The goods ship.

Six weeks later your cutter is standing over an open roll with a chalk line and a problem, and the number that was supposed to protect you is sitting in a PDF doing nothing at all.

That is not because the record was faked. It is because almost everyone reads the wrong part of it.

THE SCALE BEHIND THE SCORE Defect length (either direction) up to 7.5 cm 1 point counted, but a lot of them adds up fast 7.5 – 15 cm 2 points watch the position, not only the count 15 – 23 cm 3 points usually a cut-out, so it costs metres over 23 cm, or any hole 4 points four of these on 100 m² ends the lot
The scale is fixed; what is negotiable is where a defect sits. A 4-point defect in the middle of a panel wastes a garment — the same defect on the selvedge wastes nothing.

First, what the 4-point system is not

ASTM D5430 is a scoring convention. It tells an inspector how to penalise a defect by its length along the warp: a fault up to three inches takes one point, three to six inches takes two, six to nine takes three, anything longer takes four, and a hole takes four whatever its size. Total the penalties, divide by the area inspected in hundreds of square yards, and you have a score.

Notice what that procedure contains: a method for counting. What it does not contain is a pass mark. D5430 will tell you how to score a roll; it will not tell you whether the roll is acceptable. That decision belongs to the buyer, expressed as a threshold.

Which is why the two sentences "we inspect to 4-point" and "we pass 4-point" get used as though they mean the same thing. The first is a description of a technique. The second is a claim about a limit nobody has named. A supplier can be entirely honest saying the first and have said nothing useful at all.

Three numbers nobody forwards

The score at the top is a ratio, and a ratio without its inputs is a rumour. Ask for these three and the record starts to mean something.

Coverage. How many metres or rolls were actually inspected, against how many shipped. A lot average drawn from a fraction of the lot describes that fraction. Twenty points measured across a small sample and twenty points measured across everything are not the same finding, and the report usually presents them identically.

The denominator. Because the score is normalised to area, width and roll length are baked into it. A mill running a wider fabric earns more area per defect than a narrow one; two reports of eighteen points on "the same" fabric may describe quite different severity. Get the raw defect count and the inspected yardage, and do the division yourself once. It takes a minute and it changes how you read every record afterwards.

Length reconciliation. Inspected length against shipped length. This one is quietly important: if a lot failed an inspection, was cut at the defects, and was then re-measured, the second record can be clean and the roll short. A clean report on a roll that lost metres is not a good outcome. It is the same defect, relocated into your yield.

Points measure density. Your cutter cares about position

A 4-point score tells you how much defect is in the fabric. It tells you nothing about where. Cutting rooms do not fail on how much. They fail on where.

Consider two lots, both scoring twelve. In the first, the points come from forty small faults sitting within an inch of the selvedge. They trim off with the edge, and the marker never knows they existed. In the second, four faults of four points fall down the middle of a 150 cm width, one every few metres. Same number. One lot makes the order; the other destroys the marker and the delivery date with it.

This is the single largest gap between what inspection records measure and what garment makers need, and it is a gap of geometry, not quality. A record with only a total has thrown away the one field that predicts your yield. A record with a position column, or a sketch of the roll showing where the penalties sit, has kept it.

The part the trade doesn't say out loud

Defect classification in the grey zone is a judgment call, and the judgment decides the total. A slub, a thick place, a faint machine stop mark: is that a one-point fault or a three? Two competent inspectors can score the same roll and arrive at figures far apart, both defensible. The score is therefore not a physical constant. It is a decision, recorded in a table, made by someone whose name is at the bottom of the page.

Which means the question "what did you score" matters less than "who scored it, and when." In-house inspection is self-assessment. That is not an accusation of dishonesty; it is a statement about incentives. And timing matters just as much as authorship. Fabric measured straight off a finishing line, still under tension, is not the same object as fabric that has relaxed for a day. Inspect the wrong object and you get a precise number about something you did not buy.

The third thing nobody prints: a record you cannot re-check is not a record. It is a claim with a table around it. If the sampling positions, the roll identities and the retained cuttings do not exist, the number is unfalsifiable, and unfalsifiable numbers are the easiest kind to be relaxed about.

Eight checks before you accept a lot

Print these against the next record you receive. Most records fail at least three.

  1. Is the standard and the acceptance threshold in the same document? "ASTM D5430" without a limit tells you the method, not the outcome.
  2. Inspected quantity versus shipped quantity — in metres and in rolls. If only a percentage is inspected, whose percentage, and chosen how?
  3. Roll-by-roll results, not a lot average. An average is where one bad roll goes to hide.
  4. Defect position, not just defect count. Distance from the selvedge, or length along the roll. Without it you cannot forecast cut yield.
  5. Classification notes for anything in the three- and four-point band. Those are the entries where judgment was exercised, so those are the entries worth reading.
  6. First inspection or re-inspection. If re-inspection, ask for the length reconciliation, and be suspicious of a spotless second report.
  7. Who inspected, and on what date relative to finishing. In-house or third party, before or after relaxation.
  8. Can you re-inspect? Retained samples, roll numbering, a person who will answer a question about a specific roll by number.

None of these require a certification. They require a supplier who kept the working papers, which is a different and much rarer thing than a supplier who owns a light box.

How we handle it

Every production lot we ship is inspected on the 4-point system to ASTM D5430, and the record goes to the buyer rather than into a drawer. Where an order calls for independent verification, we arrange third-party testing per order with SGS, Intertek or BV, and the report is issued in your company's name and travels with the goods — so that when your customer asks the question, you are holding the answer.

Then the check that sits ahead of the inspection report: when bulk comes off the machine we cut one to two metres — or whatever length you specify — from the actual production lot and send it to you. You approve the real fabric before anything ships. If the swatch is wrong, we are having that conversation while the goods are still in our warehouse, which is the only moment it can be had by courier instead of by claim.

The rest of the chain is arranged so the record is answerable. Our greige comes from named partner mills and is traceable. Lab dips run five to seven days from your colour standard; bulk runs fifteen to thirty-five days. Minimum order is fifty metres per colour, so a small run gets the same inspection as a large one rather than a lighter version of it.

The purpose of an inspection record is not to prove the fabric is good. It is to move the argument about the fabric to a point where a swatch can still settle it. Records that arrive after the container has sailed are not quality control. They are paperwork about the past.