Short answer
With a physical standard, a measured target in CIELAB under a stated illuminant and observer, a tolerance in a named ΔE formula that is realistic for recycled base material, and an agreed sampling plan across lots. A HEX value alone cannot specify it, because it describes screen output, not how a material with its own base colour, gloss and haze reflects light.
A HEX code is a screen instruction in sRGB. A moulded part, a sheet of board or a dyed yarn is a reflecting surface whose colour depends on the light it is viewed under, its gloss and texture, and — for recycled material — a base colour that differs from lot to lot. A HEX value says nothing about illuminant, geometry or tolerance. A usable specification names a physical reference (a plaque, drawdown or yarn card), a measured CIELAB value with the illuminant, observer and measurement geometry stated, and a tolerance. The screen colour can stay as a digital proxy for marketing materials, but it should never be what a supplier is asked to hit.
Tolerances are written as a colour difference, ΔE, and there are several formulas. The same pair of samples can give quite different numbers in ΔE*ab (the 1976 Euclidean distance) and ΔE00 (CIEDE2000), because CIEDE2000 compresses differences in chroma — such as a yellow shift in b* — much more than differences in lightness. The table on this page recomputes a published reprocessing series both ways. Where darkening dominates, the two formulas stay within roughly 10–20% of each other; for the step between the third and fourth cycles, which was mostly a jump in b*, ΔE00 is only about 60% of ΔE*ab. A tolerance of 'ΔE 2' is therefore meaningless unless the formula is stated, and a supplier's report in one formula cannot be checked against a specification written in another.
Virgin-material tolerances are often set tight because the base is constant and the only variable is the pigment recipe. With recycled base material the base itself moves, so the same tolerance can reject a large share of perfectly functional lots. Realistic specifications separate the components: a tighter tolerance on hue for brand-critical parts, a looser one on lightness and yellowness where the base drifts, and perhaps a separate haze limit for transparent parts. They also say how many lots are sampled and whether acceptance is per lot or against a running average. Visual approval under a named light source remains the final check, because formulas do not capture texture or speckle.
Recycled plastics are usually coloured with a concentrated masterbatch blended into the base at a set letdown ratio. When the base colour moves, holding the ratio fixed lets the final colour move with it; holding the colour fixed means the compounder adjusts the recipe or ratio per lot, which costs more but gives steadier colour. The specification should say which of the two is expected. For transparent parts, a small amount of blue or violet toner is sometimes used to cancel yellowness, at the cost of slightly lower lightness — another reason to specify L* and b* separately rather than a single ΔE.
| Extrusion cycles | ΔE*ab vs cycle 0 | ΔE00 vs cycle 0 | ΔE*ab vs previous cycle | ΔE00 vs previous cycle |
|---|---|---|---|---|
| 1 | 3.6 | 3.5 | 3.6 | 3.5 |
| 2 | 11.5 | 10.5 | 11.0 | 9.2 |
| 3 | 13.2 | 12.0 | 2.0 | 1.6 |
| 4 | 14.7 | 12.0 | 7.3 | 4.4 |
| 5 | 22.9 | 19.2 | 10.6 | 9.3 |
| 6 | 18.2 | 14.7 | 5.0 | 4.7 |
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
Recomputed from published CIELAB values for PLA after repeated extrusion, ΔE00 stays within roughly 10–20% of ΔE*ab where the change is mainly darkening, but only about 60% of it for the third-to-fourth-cycle step, where the change was mainly a rise in b*.
Based on: Colourwise computed ΔE*ab and ΔE00 from the L*, a*, b* values in Agüero et al. 2019, Table 5, using the site's colour-difference functions; see the table on this page.
Caveat: The ratio depends on which colour component changes and where in colour space the samples sit.
FactStrong evidence
The same study reported ΔE*ab of 3.6 after a single reprocessing cycle, which its authors class as a difference an inexperienced observer notices.
Colourwise interpretationModerate evidence
Specifications for recycled materials should state the ΔE formula, illuminant and observer, and set separate lightness and yellowness limits, because a single ΔE value hides which component the recycled base is shifting.
Based on: Follows from the formula comparison on this page and the lightness and b* drift reported in the cited recycled PET and PLA studies.
Reviewed 29 September 2026. Colourwise summarises its sources in its own words and does not reproduce standards text or proprietary colour data. Spotted an error? Tell us.