A shared Pantone number is a target, not a standard. When two factories each measure their own production against that target and each stays inside their own tolerance, both can report a pass and the two halves can still read as two different colours on the shelf. The fix is not a tighter number. It is a basis pack that both factories receive and both are measured against: one sealed master, one measurement sheet, one tolerance that carries a direction, one base cloth declaration, and one batch code that travels to the carton. Send the same pack to both sites, keep the arbitration copy, and the comparison stops depending on whose instrument was used.
Why a Shared Colour Number Is Not a Shared Standard
A Pantone reference names a pigment target. It does not name the cloth underneath, the dye route, the ink film thickness, or the instrument that will decide whether the result passes. On a dyed fabric pouch, the base cloth and the dyeing route decide most of what the eye sees. On a printed pouch, the ink, the mesh and the substrate brightness decide it. Two factories with different equipment can hit the same target with errors that point in opposite directions, and both can be honest about the result.
The second problem is arithmetic. Colour difference is a magnitude, and a magnitude has no sign. A batch that is slightly red and a batch that is slightly green can sit at the same distance from the master and pass the same tolerance, then look wrong the moment they are placed next to each other. The annex described below therefore asks for a tolerance with a direction, not a circle around the target.
The third problem is the instrument. A spectrophotometer reading taken at 45/0 geometry and one taken with an integrating sphere are not the same measurement, and neither is a reading taken on a flat laminated card the same as a reading taken on a sewn panel with pile. Two factories can both be right and still produce numbers that cannot be compared. The measurement sheet exists to remove that possibility before the first bulk run, not after the first complaint.
Symptom, Cause, and the Fix That Belongs in the Spec
Most two-supplier colour failures follow one of five patterns. Each one has a cause that can be written out of the process, and each fix belongs in a document both factories sign rather than in an email thread.
| What you see | What usually caused it | What to write into the colour annex |
|---|---|---|
| Both batches pass their own check, and the two halves are visibly different side by side | The tolerance is a circle with no direction, so each site drifted the opposite way | A one-sided tolerance with a named hue direction, so both sites must land on the same side of the master |
| Matches under office light, splits under the shop’s lighting | Metamerism, and no light source named in the approval | Measurement under a named illuminant, plus a visual check under the actual retail lamp |
| One half reads duller or greyer overall, with no clear hue shift | Different base cloth lot, or a different substrate brightness | A cloth lot line, or a base brightness reference measured on the undecorated panel |
| The second delivery drifts even though the first was approved | No retained master at the buyer, so each site drifted from its own working copy | A sealed master held by the buyer, measured at every goods-in against both sites |
| The numbers agree but the panels do not look alike | Different instrument geometry, aperture, or reading position | One geometry, one aperture, one illuminant and observer pair, and named reading locations |
Read the third row carefully, because it is the one buyers miss. If the two factories buy their base cloth from different mills, or from the same mill in different lots, the undecorated panel already differs before any print or dye is applied. No ink formula can correct a base that is a shade off, and a tolerance measured against a Pantone chip will not reveal it. That is why the annex carries a cloth line as well as a colour line.
What Each Factory Receives on Day One
The pack is small enough to send by courier and specific enough to remove most of the argument. It has three parts, and all three have to reach both sites on the same day, because a factory that receives the master a month after the other one has already started will have made its own reference in the meantime.
The sealed master and its working copies
One master panel is sealed, numbered, dated, signed across the seal, and kept by you in the dark. Each factory receives a working copy cut from the same approved piece, sealed the same way, with the same number. The copies are for handling; the master is for arbitration and is never opened for inspection. A pouch panel is a better master than a Pantone chip, because it carries the cloth, the finish and the construction in one object, and it is what the customer will actually see.
Keep the master out of daylight and out of the sample case your sales team carries around. A faded master is worse than no master, because it silently moves the target and every future batch is measured against a colour that no longer exists.
The measurement sheet
The sheet states how a number is produced, not just what number is expected. It names the instrument model, the geometry, the aperture, the illuminant and observer pair, how many readings are averaged, and exactly where on the pouch those readings are taken. On a sewn bag that last point matters more than on a flat card: the reading belongs on the flat area of the front panel, away from the seam allowance, away from the edge of the decoration, and away from the drawstring channel where the fabric is doubled.
If the two sites use different instruments, ask both to measure the same sealed master and send the readings before bulk. Two instruments that disagree on the master will disagree on production, and it is far cheaper to discover that on one panel than on two thousand pouches.
The cloth lot line
The annex names the base cloth, its mill, and the lot or a base brightness reference. If a single lot can be allocated to both factories, do it, even at a small premium, because it removes the variable that no amount of colour management can fix. If lot allocation is impossible, say so in the annex and require both sites to submit an undecorated panel for approval before decoration starts.
The Swap Test at Goods-In
The test that catches a two-supplier mismatch is not a measurement of each delivery. It is a comparison of the two deliveries against each other, done in one place, under two light sources, by someone who did not approve the samples.
Pull ten pouches from each factory’s carton and lay them alternately on a neutral grey surface. Look at the row under a daylight-matching lamp, then under the lamp your retail customer actually uses. Record a pass or fail for the pair, not for each side, because a pair that fails is a pair that fails even when both sides are individually inside tolerance. Photograph the layout with both batch codes visible in the frame, then measure five panels from each side against the master with the geometry named on the sheet.
Two outcomes are worth writing down separately. If one side is off the master, the annex did its job and the conversation is with one factory. If both sides are inside tolerance but pointing in opposite directions, the annex failed, and the fix is a one-sided tolerance rather than a tighter one.
The Parameters That Belong in the Colour Annex
The annex is one page. Every line on it should be something a third party could check without asking either factory what it meant.
| Paramètre | What to write | Why it has to be explicit |
|---|---|---|
| Master identity | Sample number, date, seal signature, and which site holds a copy | Points every future argument at one physical object |
| Target and tolerance | The colour reference, the colour-difference formula by name, the magnitude, and the allowed direction | Two formulas give different numbers for the same pair, and a direction stops opposite drift |
| Measurement conditions | Instrument, geometry, aperture, illuminant and observer, number of readings | Makes two sites’ numbers comparable before bulk |
| Reading locations | Front panel flat area, distance from the seam, distance from the decoration edge | Removes the difference between measuring a bag and measuring a convenient flat spot |
| Base cloth | Mill, lot code, or an undecorated brightness reference | Separates a cloth problem from a colour problem |
| Batch identity | Code printed on the polybag and the carton, recorded on the packing list | Traces a complaint back to one lot and one site |
| Acceptance and remedy | What happens on a fail, who measures, who pays for the recheck | Turns a disagreement into a procedure |
The colour-difference formula line is not pedantry. The same pair of panels produces a different number under different formulas, so an annex that says only “within tolerance” has not said anything a second site can act on. The method for calculating colour differences is standardised in ISO 105-J03, and naming the edition removes the ambiguity in one line.
When the Two Halves Still Disagree
Sometimes the pack is in place and the two deliveries still split. The first move is to measure both against the sealed master with the geometry on the sheet, not against each other, and to record the direction of each difference rather than its size alone. That single step usually names the site that is out.
If both sites are inside tolerance and the direction differs, the problem is the annex and the remedy is a commercial decision rather than a technical one. You can reallocate the base cloth, require one site to re-run against a one-sided target, sort the two deliveries into separate SKUs so no shelf ever holds both, or accept the split with a written note. What you should not do is average the two and call it a match, because the customer sees the pair, not the average.
Keep the measurement records with the batch codes. A colour conversation that starts six weeks after delivery, with no retained master and no readings, is decided by whoever is more confident, and that is not a basis for a reorder. The same logic runs through which document settles a colour dispute on a reorder and through the lighting checks in metamerism and colour approval on pouch fabric, and both are worth reading before the second factory starts, not after.
Conclusion
Two factories will not match because they were given the same colour number. They match when they are given the same object, measured the same way, against the same cloth, with a tolerance that sends both of them to the same side of the target, and a batch code that lets you find the one that drifted. Build the pack once, send it to both sites, keep the master sealed and dark, and run the swap test at every goods-in. It is a page of work that replaces a season of arguments.
One habit makes the pack last: any change to the cloth, the decoration or the site goes through the same change control that governs an approved pouch specification, so a new lot never enters production on a verbal approval. That is the difference between a colour standard and a colour hope.
