What Actually Makes a Second Reorder Match the First?
A second run matches the first when four documents exist before booking: a retained reference, a component register, a tolerance table and a first-piece sign-off. This article shows what each one contains, who signs it, and how the match is proved at goods-in rather than argued about afterwards.
['MOQ 500 pcs per style and colourway; 300 pcs open to discussion on a repeat.', 'Sampling 6-10 working days from a signed-off tech pack.', 'Bulk production 35-50 days from written sample approval.', 'Inspection at AQL 2.5, ISO 2859-1, General Inspection Level II.', 'T/T 30/70, FOB Xiamen; CIF and DDP available.', 'Consistency is a document set, not a promise.', 'Four documents decide whether run two matches run one.']
Buyers ask for consistency as if it were a property a vendor either has or does not. It is not. Consistency is the output of four documents that exist before a run is booked, and it fails when any one of them is missing. The good news is that all four are clerical, and all four are reused on every subsequent booking.
Those documents are the retained reference, the component register, the tolerance table and the first-piece sign-off record. The first says how run one looked and felt. The second says what goes into the bag. The third says how far a measured property may move. The fourth proves the line was set correctly before volume began.
The order matters. The reference comes first because the tolerance table is written against it, and the component register comes before the tolerance table because a property that has no registered input cannot be bounded. Programmes that start with the tolerance table usually end up writing bands against a sample nobody can produce, which is why the bands get argued about at inspection instead of being read.
None of this requires new equipment or a new approval process. It requires someone to spend an afternoon at sample approval writing down what is already known, and someone at booking to check that the file travels with the order rather than staying in an email thread from the previous season.
What Has to Be True Before a Second Run Can Match?
Three conditions have to hold. Something physical from run one must still exist and still be trustworthy. Every input must be named precisely enough that a substitute can be detected. And every property that matters must have a number attached, with a method for taking it.
If the physical reference is missing, the comparison has no baseline. If the inputs are named loosely - black webbing, number five zipper - then a substitute is compliant. If the properties have no number, then every difference is arguable. Most failed reorders have failed on condition two, because naming inputs precisely is dull work that nobody does until it matters.
Which Four Documents Control the Outcome?
The retained reference is a sealed sample from the approved run plus cut panels of the body fabric, lining, webbing and any printed panel. The component register is a line list naming each trim item by brand, material, finish and code. The tolerance table is the six to ten measurable properties with bands. The first-piece record is the sheet signed before volume starts.
| Document | What it fixes | Who signs it | When it is made |
|---|---|---|---|
| Retained reference | How the approved run looked and felt | Both parties, sealed and dated | At sample approval |
| Component register | Every input by brand, material, finish and code | Both parties, on the spec file | At sample approval |
| Tolerance table | How far each measured property may move | Both parties, before booking | Before run two |
| First-piece record | That the line was set correctly | Line supervisor and QC | Before volume starts |
| Comparison sheet | What this run measured against the bands | Inspector, at goods-in | At final inspection |
| Revision log | Which changes were authorised and when | Both parties, per change | Whenever something moves |
How Does a Written Tolerance Change the Conversation?
A written tolerance moves a disagreement out of adjectives and into arithmetic. Without one, the conversation is whether the shade is noticeably different, which depends on who is holding the bag and what light they are standing under. With one, the question is whether the reading is inside the band, which has one answer.
The band also changes behaviour before the run. When a plant knows the body shade has a number and the number is checked against a retained panel, the dyehouse is asked for a shade match against the panel rather than against a recipe name. That single change prevents more disputes than any inspection that happens later.
What Is a First-Piece Sign-Off and Why Does It Matter?
A first-piece sign-off is the check performed on the first units off the line after the machine is set, before volume runs. On a new style this happens naturally because everyone is uncertain. On a repeat it is skipped, because everyone is certain, and that certainty is exactly what lets the previous setting survive into the new run.
The sign-off should cover the seam that carries load, the panel that sets the shape, the artwork placement, and any assembly that is hard to reverse. Five units is enough for most styles. The output is one sheet with the measurements, signed by the line supervisor and the QC technician, filed with the run record.
How Many Units Should Come off the Line First?
Enough to cover every station, because the point is to catch a machine or an operator, not a unit. On a seven-station line, five units means every station has produced at least one piece that was checked. Where a style has a difficult assembly - a welded seam, a bound corner, a padded divider - take units until that station has produced three acceptable pieces in a row.
The units are then either retained as part of the run record or scrapped. They should not be packed into the shipment, because they were produced while the machine was still being adjusted and they are the least representative units of the whole run.
Where the first pieces fail, the fix is almost always a setting rather than a material. Tension, presser-foot pressure, needle size and feed timing account for most first-piece failures on a repeat, and each can be corrected in minutes. That is the second reason the sign-off is worth doing: it catches a setting before it has been applied to a thousand units rather than after.
Where the first pieces fail on a material property - shade, thickness, hand feel - the run should stop until the material question is answered, because no setting will correct it. Distinguishing those two failure types at the first-piece stage is the single highest-value judgement on a repeat line, and it is the reason the sheet is signed by someone who can stop the line.
What Does a Comparison Station Need on the Table?
A comparison station is a table, a light booth or a fixed lamp with a known illuminant, a steel rule, a tape, a calliper, a magnifier and the retained reference. It is not a laboratory and does not need to be. It needs to be the same table every time, with the same lamp, so that two runs are compared under the same conditions.
The station also needs the sheet. A comparison performed from memory - this looks the same as last time - has no record and cannot be reviewed by anyone who was not there. A comparison performed on a sheet can be attached to the inspection report, sent with the shipment documents, and read by the buyer at goods-in.
Why Does Lighting Decide the Colour Argument?
Two panels can match under one lamp and differ under another, because the illuminant changes which wavelengths reach the eye. This is not a trick; it is why instrumental colour reading names the illuminant and the observer angle, and why a visual check without a named light source is not evidence of anything.
Practical rule: compare under one named light source, at a fixed distance, with both panels oriented the same way. Where the two sides still disagree, fall back to a gray scale reading, which grades the difference rather than describing it, and to an instrument reading taken by a laboratory working to a recognised competence standard.
| Comparison condition | Why it changes the answer | Setting worth writing down |
|---|---|---|
| Light source | Different illuminants reveal different wavelength differences | One named source, named in the spec file |
| Viewing distance | Small differences disappear at arm length and appear at close range | A fixed distance stated in centimetres |
| Panel orientation | Nap and weave direction change the apparent shade | Both panels oriented the same way, always |
| Background | A bright surround pulls the perceived shade | A neutral grey background behind both panels |
| Observer | Two people grade a borderline difference differently | One named grader, or an instrument instead |
| Time of day | Daylight shifts in colour temperature across the day | A booth or a fixed lamp, never a window |
How Do You Judge a Match on Touch and Hand Feel?
Hand feel is genuinely hard to number, and pretending otherwise produces false precision. The workable method is a retained hand panel: a cut piece of the approved fabric kept with the reference, compared by touch at the station. It will not produce a number, but it will produce agreement, which is the actual goal.
Where hand feel must be instrumented, the proxies are fabric mass per unit area, thickness under a fixed pressure, and bend length. Those are conventional textile measurements with published methods, and they catch the two causes that matter: a different substrate and a different coating or laminating weight.
- Body shade
- Instrumental difference against the retained panel, named illuminant
- Fabric mass
- Mass per unit area, measured on a cut panel from the run
- Foam thickness
- Thickness at named points, in millimetres, uncompressed
- Overall dimensions
- Length, width, depth measured flat and unstuffed
- Stitch density
- Stitches counted over a fixed length at named seam positions
- Hardware code
- Brand, material, finish and code from the component register
- Artwork position
- Distance from two fixed edges on a retained printed reference
What Happens When a Component Is Genuinely Unavailable?
It happens: a zipper code is discontinued, a plating bath closes, a webbing weave is dropped by the mill. The process answer is that a substitution is permitted only with a written change note issued before the affected units are made, and supported by a physical comparison against the registered item.
The note should state what changed, why, what the substitute is, and what the buyer's options are: accept with a revised reference, accept this run only, re-sample, or pause the run. A substitution discovered at goods-in is not a substitution; it is a claim, and claims are resolved slowly.
| Option | When it fits | What it costs | What it leaves behind |
|---|---|---|---|
| Accept with a revised reference | Substitute matches on every property that has a band | A new sealed reference and an updated register | The programme continues without a gap |
| Accept this run only | Difference is visible but the season cannot wait | A written exception on one run number | A note that the next run returns to the register |
| Re-sample before volume | Substitute changes fit, assembly or function | Sampling 6-10 working days, plus a new approval | A second gold sample and a revised tolerance table |
| Pause the run | Substitute affects a regulated or tested property | Calendar time and possibly a second material purchase | A decision that protects the tested claim |
| Source the registered item | Substitute is available but the original is only delayed | Calendar time, quoted openly | No change to the file at all |
How Is the Match Written Down So Both Sides Agree?
One comparison sheet per run, with the run number, the lot size, the sampling plan, the reference values, the measured values, the band for each property and a pass or fail line per property. Both sides sign it. It travels with the inspection report and is filed against the style for the next booking.
The discipline is that the sheet is written at inspection, not at dispute. A sheet written during a dispute is advocacy. A sheet written at inspection, before anyone knew there was a problem, is evidence, and evidence settles things in an email instead of a month.
Most repeat programmes do not fail on capability. They fail because the component register was never written, so a substitution cannot be called a substitution.
Send us the four documents
Send the reference, the component register and the tolerance table you hold today. We coordinate production against them, bulk in 35-50 days from written sample approval, inspected at AQL 2.5, and we will tell you which field is missing.
Frequently Asked Questions
What makes a second reorder match the first?
Four documents that exist before booking: a sealed retained reference from the approved run, a component register naming every input by brand and code, a tolerance table with bands for the measured properties, and a first-piece sign-off taken before volume starts.
Do I need a laboratory to check a reorder?
Not for most properties. A table, a named light source, a steel rule, a calliper and the retained reference cover dimensions, stitch density, hardware codes and artwork position. A laboratory is worth using when shade or a chemical property is disputed, because a laboratory report names its method.
What is a first-piece sign-off?
A check of the first units produced after the machine is set, before volume runs, covering the load-bearing seam, the shape-setting panel, the artwork placement and any hard-to-reverse assembly. Five units is usually enough, and the sheet is signed by the line supervisor and QC.
Why do two panels match under one lamp and not another?
Because the illuminant changes which wavelengths reach the eye. That is why instrumental colour reading names both the illuminant and the observer angle, and why a visual check without a named light source proves nothing either way.
How is hand feel controlled if it cannot be measured?
With a retained hand panel kept alongside the reference sample and compared by touch at the station. Where a number is genuinely required, the proxies are fabric mass per unit area, thickness under fixed pressure, and bend length, all of which have published test methods.
What should a component register contain?
One line per trim item naming the brand, the material, the finish and the code, plus a retained physical reference attached to the spec file. Loose descriptions such as black webbing or number five zipper invite substitution and cannot be enforced later.
Can a substitution ever be accepted?
Yes, when it is proposed in writing before the affected units are made and supported by a physical comparison against the registered item. The note should state what changed, why, and whether the buyer accepts a revised reference, accepts one run, re-samples, or pauses the run.
How many properties should have a tolerance?
Six to ten. Programmes that write thirty tolerances usually check none of them, and programmes that write two leave the visible differences unbounded. Body shade, trim shade, foam thickness, dimensions, stitch density and artwork position cover most bag styles.
Who should sign the comparison sheet?
The inspector who took the measurements, plus the QC lead on the production side. Two signatures matter because the sheet is the document that settles a later disagreement, and a sheet with one signature reads as one side describing its own work.
Does a written tolerance make the run more expensive?
Sometimes, and the change should be quoted rather than assumed. Tighter bands mean more sorting and occasionally a dedicated material purchase. Set that against the cost of rework, the markdown on a mismatched shade, and the smaller order a buyer places next time.
What if the retained reference has faded?
Replace it. A reference that has faded, been handled daily or sat in daylight no longer describes the approved run, and comparing against it produces a false pass. Retain two references from the same run and keep the working one in a drawer away from light.
How long should a reorder record be kept?
For as long as the style is active plus one reorder cycle. The record is what lets run five be compared against run one rather than only against run four, which is the only way a slow drift that nobody noticed becomes visible.
People Also Ask
How do you make a repeat order identical to the first?
You cannot make it identical, but you can make it predictable: seal a retained reference from the approved run, register every component by brand and code, agree tolerance bands for six to ten measured properties, and sign off the first pieces before volume starts.
What documents prevent reorder problems?
The retained reference, the component register, the tolerance table and the first-piece sign-off, supported by a comparison sheet written at inspection and a revision log that records every authorised change.
What is a gold sample in manufacturing?
The physical unit approved by the buyer and retained by both parties as the reference for later runs. It should be sealed, dated, stored away from light, and accompanied by cut panels of fabric, lining, webbing and printed material.
How do you check if a reorder matches the original?
Pull a sample sized from the ISO 2859-1 tables, compare each unit against the retained reference at a fixed station under a named light source, record dimensions, shade, hardware codes and stitch density, and mark each property inside or outside its band.
What tolerance should I set for bag dimensions?
There is no universal figure. Set a band the process can actually hold, write it into the spec file with the measuring method and the named points, and tighten it only after two runs have shown the plant can hold the tighter band consistently.
Sources and further reading
Every figure on this page is either our own programme data or attributed below. Open the originals before you write them into a specification.
- ISO 2859-1:1999 — sampling procedures for inspection by attributesInternational Organization for Standardization
- AATCC test methods and evaluation proceduresAmerican Association of Textile Chemists and Colorists
- ISO 105 series — tests for colour fastness of textilesInternational Organization for Standardization