After the wash, and the defect an inspector cannot see

One tolerance, three components, three directions

Aubrette's general specification carries a single rule for dimensional stability, which means how far a garment moves when it is washed: no point of measure to move more than 3.0% after five domestic washes. It is a sensible rule. It was written for jersey, and it is meaningless on BR-6140.

Thalna washed twelve bras five times at 40 degrees Celsius and line dried them:

Three components of one garment moved in three different directions, and behind them are three different mechanisms.

• The moulded cup is a shape set into a foam laminate with heat. Heat and water let that set relax, so the cup loses depth and spreads. It has not shrunk. It has partly forgotten its shape.
• The wire channel lengthened because the cup spread underneath it and the casing tape followed.
• The band shortened because the elastic recovered. That is the elastic working, not failing.

One percentage applied to all three at once fails the cup at 3.2%, passes a wire channel that grew, and passes a band whose movement is the one you want. On a moulded garment, "shrinkage" is not one thing, and a single stability tolerance is not a specification.

What means something is a named list, with a direction and a limit on each line. Cup depth may lose up to a stated amount and must not gain. Wire channel length may gain up to a stated amount. The relaxed band may shorten within a stated range and must not lengthen. Course 4.4 argues that the tolerance decides pass or fail, not the garment. Here it also has to decide which way.

Two more readings make the point. Tumble dried on a hot cycle instead of line dried, the same cup goes from 12.6 cm to 11.5 cm, which is minus 8.7%, and the bra is finished. That is not a fault in the garment. It is the care label doing structural work. In this category the care instruction is a design decision taken with the fit, not a labelling job done at the end. Courses 4.6 and 6.5 cover care labelling as a document. What is specific to this category is that the instruction is load-bearing.

The wire, and the standard that does not exist

An underwire that works its way out through the end of its casing is the failure this category is remembered for. It is a safety matter rather than a commercial one. A wire end is a sharp object against skin.

There is no harmonised product standard for wire retention in a bra. That does not mean there is no obligation. It means the obligation is the general one: the producer must place a safe product on the market and be able to show the assessment behind it. A certificate that a lot passed a visual inspection is not that assessment.

So the assessment is the manufacturer's own, and it has to be written down. Here is Thalna's, agreed with Aubrette. It is this pair's own protocol, not anybody's standard.

• The wire tip is capped, and the casing is closed with a bartack at least 6 mm beyond the tip. A bartack is a short block of dense stitching.
• A sample from each shade and each of the nine volumes is flexed on a cyclic rig for 5,000 cycles. After that, no wire end may be sticking out of its casing.
• Any migration found in the finishing room is a process stop, not a repair. The bartack position is checked on the machine before the line restarts.

The season-one bulk carried a bartack at 3 mm on one volume, half the specified distance. It passed the final inspection, because a bartack 3 mm from a wire tip looks exactly like a bartack 6 mm from a wire tip on an inspection table. It was found on the cyclic rig.

The defect the wearer feels

Course 6.2 sets out AQL sampling. AQL is the acceptable quality limit, the defect rate a buyer will still accept. At 9,000 pieces, general inspection level II and an AQL of 2.5, the inspector draws 200 pieces and may accept up to 10 defects. None of that changes here, and none of it is the problem.

The problem is that AQL samples for defects an inspector can find on a table in a few seconds, and this garment's typical failures are not that shape:

Every one of those is a process finding. Each is caused by a setting, a tension or a guide, and each affects a run rather than a piece. A sampling plan is the wrong instrument for a run-wide setting error. It is built to decide about a lot, and by the time it decides, the lot exists.

What answers it is a small wear check, run before the AQL inspection and used as a process input rather than as a decision about the lot. Thalna's version is twelve bras, three from each of four volume groups, worn for two hours by a fit panel at the start of bulk and again after the first full day. It costs EUR 340 a style.

In its first season the panel found three things the final inspection did not: the 3 mm bartack, an elastic tension setting drifting on one of the two elastic-attach machines, and a strap slider that rotated in wear because its coating had made it slightly narrower than the elastic. All three were machine or component fixes made inside a day.

A caution about the number that follows. Aubrette's comfort-related return rate on this style fell from 11.2% to 7.8% between the two seasons. The panel was introduced in that gap. So were a revised gore height and a different strap slider. The fall fits with the panel working, and it does not prove it, because nothing in that comparison separates the three changes. The honest claim is the one the panel can support on its own: it found three defects the AQL inspection could not have found, and it costs EUR 340.

Normal here, a reject elsewhere, and the reverse

A needle mark is a reject on powernet. A puncture in a warp knit or a fine tricot does not close. It becomes a ladder that opens in wear. On a woven shirt a needle mark is nothing and no inspector writes it up. On this garment it is a major defect. It is the reason a bra line changes needles on a schedule rather than on breakage, and uses a ball point of the specified size and nothing else.

A visible shade difference between the cup and the wing is accepted. The moulded cup laminate and the powernet frame are different constructions in different fibre blends. They take dye differently, and no dyehouse can bring them to an identical reading under every light. So the shade standard for this style is not one swatch. It is a two-part standard, one for the cup and one for the frame, with a stated maximum difference between them, judged under a stated pair of lights. A merchandiser who sends this back as a shade fault is sending back a garment that cannot be made any other way. The same problem of two different materials lay behind the hardware chip in lesson 1.

The category, in six lines

1. Twenty-four components means twenty-four clocks and twenty-four minimums, and the finish is the latest of them.
2. A size grid is two axes. The buy comes from the cells, the tooling from the diagonals.
3. An elastic measurement is a length at a stated force, taken from sewn assemblies.
4. Cloth is the minority line. Trims and minutes are where the money and the negotiation are.
5. In swim, test the property that fails, which is retained power, and test the trims and not only the body.
6. Some defects are felt rather than seen. Those are process findings, and sampling will not reach them.

Act as a lingerie technologist writing a measurement specification that two parties can be held to. Style facts: [STYLE], [DESCRIPTION], size range [LIST], base size [SIZE]. Paste my current points of measure with their values and tolerances: [PASTE]. The elastics in the garment, by position, width, construction and supplier: [PASTE]. My buyer's inspection manual says elastic points are taken [RELAXED OR EXTENDED, AND TO WHAT]. Do the following. First, sort every point of measure into three groups: rigid points, elastic points, and points on a moulded or bonded component — and say why each one is in its group. Second, for every elastic point, rewrite it as a value plus a CONDITION plus a statement of which condition governs acceptance, and tell me what has to be measured on SEWN assemblies before the extended figure can be written at all. Say plainly that a figure taken from an elastic tape's own data sheet is not a figure about the assembled garment. Third, tell me how many sewn samples to measure, what to record besides the mean, and how to word the sentence that settles a conflict between the two figures. Fourth, build the after-wash set: name each point that must be measured after washing, give it a DIRECTION as well as a limit, and say which points are expected to grow, which to shrink, and which must not move. Do not give me one percentage for the whole garment. Fifth, tell me which wash and dry cycle the specification has to name and what changes if the care label allows tumble drying. Sixth, list the points that cannot be judged by eye at inspection speed and belong on an in-line gauge check instead. Ask me for anything missing rather than assuming a value.