Short answer
It should be at the same temperature and moisture content as the standard, opaque or over a stated backing, presented the same way every time, and read often enough that the average is stable. Each of those is a real, reversible source of colour difference that has nothing to do with the batch: a saturated red can move by a whole CIELAB unit over 10 °C, cotton shifts with humidity, a thin film takes colour from its substrate, and a textured or woven surface reads differently at each placement.
Thermochromism is a reversible change of colour with temperature, distinct from the permanent change caused by overheating. Its size is rarely published for commercial materials, but it is published for the ceramic tiles used to verify instruments, and the table shows the pattern. For a 10 °C rise the four grey tiles move by 0.06 ΔE*ab or less. The red and orange tiles move by about 1.0, the yellow and greens by about 0.7, and the deep blue hardly at all. If the shift is taken as linear, a difference of only 3 °C between standard and sample is worth about 0.3 on the red. The practical rule is one measurement temperature for standards and samples at every site, and for anything coming hot off a process, a cooling time established by measuring it at intervals until the reading stops moving.
Cotton and other hygroscopic fibres exchange water with the air, and their measured colour changes as their moisture content does. Textile testing therefore conditions specimens in a standard atmosphere before measurement, defined in an ISO standard that also allows an alternative atmosphere by agreement. The effect on colour is modest but not uniform. In one instrument maker's experiment on dyed cotton, differences from the conditioned standard ranged from under 0.1 ΔE CMC(2:1) for some shades to more than 0.3 for a bright blue as humidity varied at constant temperature. Samples need time to reach equilibrium, so the conditioning time belongs in the procedure, and a standard should be re-measured after shipping or a change of storage instead of being assumed unchanged.
An instrument measures whatever light comes back, including light that passed through the sample and returned from behind it. Fabric is folded until another layer no longer changes the reading, typically two to four layers; a sheer fabric that never becomes opaque is measured over a specified white tile. A coating is checked for hiding by applying it over black and white and comparing the two areas, and until those agree its colour belongs partly to the substrate. Translucent plastics are measured as flat specimens of uniform thickness over a stated backing, and print is measured over a backing the graphic-arts measurement standard specifies. In every case the backing is part of the measuring condition and must be written down, because white and black backings give different numbers for the same piece.
A reading samples one small area in one orientation. On a smooth, uniform panel that is enough. On a woven or knitted fabric, a grained moulding, a brushed metal or a mottled print it is not, because weave direction, texture and local unevenness all change the result from one placement to the next. The remedy is the same everywhere: the largest aperture the sample allows, several readings with the sample taken off and repositioned between them, and for directional materials a fixed set of orientations. Cleanliness is part of presentation. A fingerprint changes local gloss, and guidance for both physical standards and samples is to avoid touching the measured surface at all. Pile and loose fibre need a glass plate or a compression cell at constant pressure.
The number of readings to average is something to determine, not assume. A method given by one instrument maker is to measure a representative sample eight times with full repositioning, take the average as the reference, and then test how few readings can be averaged while the result stays within a chosen spread of it; that maker's own target for fabric is a variation below 0.15 ΔE CMC. The answer depends on the material: one reading may do for a gloss panel and six may be needed for a coarse weave. Two things follow. The count belongs in the written procedure for each product, and the measurement spread it leaves is a fixed cost that the product tolerance has to be wide enough to absorb.
| Tile | ΔL* | Δa* | Δb* | ΔE*ab for 10 °C | ΔE*ab for 3 °C, if linear |
|---|---|---|---|---|---|
| Pale grey | -0.03 | -0.02 | 0.02 | 0.04 | 0.01 |
| Mid grey | -0.03 | -0.02 | 0.04 | 0.05 | 0.02 |
| Difference grey | -0.04 | 0.04 | 0.03 | 0.06 | 0.02 |
| Deep grey | 0.00 | 0.01 | 0.00 | 0.01 | 0.00 |
| Deep pink | -0.10 | -0.44 | -0.19 | 0.49 | 0.15 |
| Red | -0.37 | -0.71 | -0.61 | 1.01 | 0.30 |
| Orange | -0.45 | 0.56 | -0.66 | 0.98 | 0.29 |
| Yellow | -0.27 | 0.70 | -0.11 | 0.76 | 0.23 |
| Green | -0.18 | 0.66 | -0.04 | 0.69 | 0.21 |
| Difference green | -0.18 | 0.69 | -0.05 | 0.71 | 0.21 |
| Cyan | -0.10 | 0.31 | 0.01 | 0.33 | 0.10 |
| Deep blue | 0.00 | -0.04 | 0.05 | 0.06 | 0.02 |
Why: The sample was measured warm, or the standard cold; saturated warm colours are strongly thermochromic.
Fix: Bring both to the measuring room's temperature and establish a minimum cooling time by trial.
Why: Layers, orientation, pressure or backing are not specified, so each operator chooses.
Fix: Write the presentation into the procedure and check operators against each other on one sample.
Why: Neither conditions its samples, so each measures a different moisture content.
Fix: Condition to the same atmosphere for a stated time at both sites, in a room or a cabinet.
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.
Colourwise analysisStrong evidence
The published thermochromic shift of the Ceramic Colour Standards tiles for a 10 °C rise corresponds to about 1.0 ΔE*ab for the red and orange tiles, about 0.7 for the yellow and green tiles, and 0.06 or less for the grey and deep blue tiles.
Based on: Colourwise calculated the Euclidean CIELAB distance from the maker's published ΔL*, Δa* and Δb* for each tile (specular included, D65, 10° observer); see the table on this page.
Caveat: Glazed ceramics only. The column for a 3 °C rise assumes the shift is linear with temperature, which the source does not state.
FactModerate evidence
Thermochromism is a reversible change of colour with temperature and is distinct from thermal degradation, which is irreversible; guidance is to set one baseline measurement temperature for samples and standards across all sites.
Source: Temperature Effects on Color Measurement: Maintaining Accuracy in Extreme Conditions; Thermochromism: The Effects of Temperature on Pigments and Dyes
StandardStrong evidence
ISO 139:2005 defines the standard atmosphere for conditioning and testing textiles and a standard alternative atmosphere that may be used by agreement between parties.
Caveat: The standard is written for physical and mechanical testing; colour laboratories adopt its atmosphere by convention.
Source: ISO 139:2005 Textiles — Standard atmospheres for conditioning and testing
FactLimited evidence
In one experiment on dyed cotton at constant temperature, varying humidity produced colour differences from the conditioned standard of 0.03 to 0.36 ΔE CMC(2:1) depending on the shade, with blues and violets among the most sensitive.
Caveat: A manufacturer's own small experiment on cotton; not independently published.
ConventionModerate evidence
Guidance for textiles is that two to four layers are sufficient to make most woven and knitted fabrics opaque for measurement, that sheer materials may need a white ceramic backing, and that the number of averaged readings should be established by measuring eight times and testing progressively fewer.
StandardStrong evidence
ISO 13655:2017 defines the measurement conditions, including backing and ultraviolet content, under which graphic-arts materials are to be measured, because brightened papers read differently as the ultraviolet content of the illumination changes.
Source: ISO 13655:2017 Graphic technology — Spectral measurement and colorimetric computation for graphic arts images; ISO 13655 measurement modes M0–M3
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.