Short answer
Textiles are toleranced with an ellipsoidal formula, historically CMC(l:c) and increasingly CIEDE2000, against a buyer's approved standard, with the pass limit written into the buyer's specification and not into any standard. The harder part is measurement: fabric is translucent, directional and hygroscopic, so the reading depends on how many layers are folded, how the piece is presented and how long it has sat in the room. A textile tolerance is only as good as the presentation and conditioning procedure written beside it.
A dye house may match hundreds of shades a season for one buyer, and writing a separate tolerance for each would be unworkable. The textile standard for colour difference was written to avoid that: it calculates ΔEcmc(l:c) so that one maximum value can serve for every colour, set by how close a match the end use needs. The ratio l:c is part of the method. At 2:1, the usual acceptability setting, the lightness axis of the acceptance volume is doubled; Colourwise's calculation puts it at a little over two CIELAB units for a limit of 1.0 at mid-lightness, against roughly one unit under CIEDE2000. That generosity towards lightness is a judgement about how fabric is bought, built into the arithmetic.
CMC is no longer the only textile formula. The American textile chemists' procedure for small colour differences covers both CMC and CIEDE2000, describes both as ellipsoidal equations that build an acceptance volume from the reference colour and a tolerance level, and recommends CIEDE2000 as the primary method because it agrees slightly better with visual evaluation. For a mill this is a practical hazard more than a theoretical one. Buyers are at different stages of the change, the two formulas accept different batches at the same number, and a limit agreed in one cannot be assumed in the other. The formula and its parameters need checking on every buyer's specification, not once for the mill.
A single layer of most fabrics lets light through to whatever is behind it, so the reading includes the backing. Guidance from one instrument maker is to fold to two to four layers, with four often preferred, to use the largest aperture the sample allows, and to take the piece off the instrument, refold and reposition it between readings so that weave direction and local unevenness are averaged, not repeated. The same guidance gives a way to decide how many readings are enough: measure eight, then see how few can be averaged while keeping the variation between repeat measurements below the maker's own target of 0.15 ΔE CMC. Pile and bulky materials are read behind a glass plate or in a holder that stops fibres entering the port.
Textile fibres take up water from the air, and their colour shifts as they do. Testing therefore starts from a standard atmosphere: ISO 139 defines one for conditioning textiles, and one instrument maker's guidance cites the 21 °C and 65 per cent relative humidity of the equivalent ASTM practice. One instrument maker reports an experiment on dyed cotton held at constant temperature and varied humidity, measured against the same shade conditioned at the standard atmosphere. The differences were small but uneven across shades: a bright blue moved by 0.23 to 0.36 ΔE CMC(2:1) and a dark yellow by 0.03 to 0.08. Against a commercial limit those shifts can use up a meaningful share of the allowance before the dyeing itself is judged, and a lab dip measured in one climate is compared with bulk measured in another.
Approval in textiles runs in stages: a small laboratory dyeing is approved against the buyer's standard, and bulk production is then approved against it again. Each stage is measured and viewed. The visual procedure for textiles standardises how the specimen and standard are compared, and the second-light check matters more here than in most industries, because fibre blends and substituted dyes make metameric matches common and garments are bought under shop lighting. Buyers increasingly issue the standard as spectral data, which lets a mill check a recipe for metamerism before dyeing anything, but the physical standard still governs handle, lustre and surface, none of which a reflectance curve records.
| Comparison | Standard or scheme | Method | Condition | Figure, as the source states it | Caveat |
|---|---|---|---|---|---|
| Dyed or printed fabric batch against the approved standard | ISO 105-J03:2009 (2nd edition, confirmed 2025) | ΔEcmc(l:c) | Two specimens of the same material measured under the same conditions; the illuminant, observer and geometry are for the parties to fix | None verified; agreed by the parties | The standard's public abstract gives no value. The limit is a commercial decision per buyer and product, and is meaningless without the l:c ratio. |
| Small colour differences between a textile specimen and its reference | AATCC EP14-2021e2 | CIEDE2000 recommended as primary; CMC(l:c) also covered | Instrumental measurement as set out in AATCC EP6; conditions fixed by the user | None verified; agreed by the parties | No tolerance level is published in the procedure's public description. A CMC limit and a CIEDE2000 limit of the same number do not accept the same batches. |
Why: The lab dip and the bulk were measured at different moisture contents, or presented with different numbers of layers.
Fix: Condition both for the same time in the same atmosphere and fix the folding and aperture in the procedure.
Why: Too few layers, a small aperture on a coarse weave, or rotating in place without refolding.
Fix: Find the number of repositioned readings needed by trial, and average that many every time.
Why: Each mill matched the standard with different dyes, so the two are metameric to each other.
Fix: Specify a metamerism limit under the shop illuminant, and where it matters a restricted dye set.
Each statement is labelled by kind — established fact, a standard’s requirement, observed market data, a convention, or Colourwise’s own interpretation or analysis — with the strength of the evidence behind it.
StandardStrong evidence
ISO 105-J03:2009 provides a method for calculating ΔEcmc(l:c) between two specimens of the same material measured under the same conditions, permitting one maximum tolerance value that depends on the end use and not on the colour, and provides a means of establishing the ratio of differences in lightness to chroma and hue.
Caveat: From the public abstract of the second edition; no numeric limit is stated there.
Source: ISO 105-J03:2009 Textiles — Tests for colour fastness — Part J03: Calculation of colour differences
StandardStrong evidence
AATCC EP14 covers CIEDE2000 and CMC(l:c) as ellipsoidal colour-difference equations for textiles and recommends CIEDE2000 as the primary method for instrumental evaluation.
Source: AATCC EP14-2021e2 Evaluation Procedure for Small Color Differences
ConventionModerate evidence
An instrument maker's guidance for fabric is to fold to two to four layers for opacity, use the largest practical aperture, reposition and refold between readings, and choose the number of averaged readings that keeps measurement variation below 0.15 ΔE CMC.
Caveat: The 0.15 figure is one manufacturer's target for measurement repeatability. It is not a product tolerance and not part of any standard.
FactLimited evidence
In one instrument maker's experiment on dyed cotton at 21 °C, the colour difference from the same shade conditioned at the standard atmosphere, as humidity was varied, ranged from 0.03–0.08 ΔE CMC(2:1) for a dark yellow to 0.23–0.36 for a bright blue (D65, 10° observer).
Method: Dyed cotton standards held at constant temperature and varied relative humidity, measured against the conditioned standard.
Caveat: A small experiment reported by a company that sells conditioning cabinets, on cotton only; the article does not give the humidity levels in its public summary.
StandardStrong evidence
ISO 139:2005 defines a standard atmosphere for conditioning and testing textiles, and an alternative atmosphere that may be used if the parties agree.
Source: ISO 139:2005 Textiles — Standard atmospheres for conditioning and testing
Colourwise analysisStrong evidence
For a limit of 1.0 at a mid-lightness neutral, CMC(2:1) accepts a lightness difference of about 2.2 CIELAB units, roughly twice what CIEDE2000 accepts.
Based on: Calculated by Colourwise by bisection along the lightness axis with its implementations of both formulas; the full table is on the page about how a tolerance is written.
Caveat: For one illustrative standard at L* 50; the ratio varies with lightness.
Source: Modification to the JPC79 colour-difference formula; The CIEDE2000 color-difference formula: implementation notes, supplementary test data, and mathematical observations
Reviewed 6 October 2026. Colourwise summarises its sources in its own words and does not reproduce standards text or proprietary colour data. Spotted an error? Tell us.