Short answer
A metamerism index is a colour difference with a condition attached: how far apart two samples are under a test light, given that they match under a reference light. It is reported in ΔE units for a named pair of illuminants, observer and formula, and means nothing without them. On pairs constructed from a measured grey, the index from D65 to a three-band fluorescent lamp ran from 1.0 to 5.4 ΔE*ab as the spectral disagreement grew, while the same pairs scored much lower from D65 to tungsten. An index ranks risk for stated lights; it does not say a match is safe under all of them.
ΔE under daylight tells you whether a batch matches its standard under daylight. A metamerism index asks something else: supposing it does match there, how big is the mismatch somewhere else? The CIE's special metamerism index for change in illuminant is defined exactly so, as the colour difference between two specimens under a test illuminant when they have identical tristimulus values under the reference illuminant. The result is a ΔE, so it carries all the usual qualifications: which formula, which observer, and that a ΔE*ab of 2 in a saturated blue is not the same visual step as 2 in a grey. Reference and test illuminants are a choice, commonly D65 against tungsten light and a store-lighting type.
Production pairs never match exactly, and the definition has no answer for them: a pair two units apart in daylight and two and a half apart in tungsten light is mostly just a mismatch. Practical indices therefore remove the reference-light difference first. Datacolor's technical note lists the DIN 6172 index as the test-light difference adjusted for an exact match under the reference, and ISO 18314-4 gives a formal calculation of the same kind for solid colours. The spectral explorer on this site uses a multiplicative adjustment, scaling the second sample's tristimulus values so that the pair matches under the reference, and reports the residual it removed. The larger that residual, the less the index means. Past a few ΔE units it describes two different colours, not a metameric pair.
The same construction works with the light held constant and the observer changed. CIE 80 defines a special metamerism index for change in observer, substituting a defined deviate observer for the standard one, and presents it as a basis for tolerances because some disagreement between real observers is unavoidable. A cruder check is to recalculate with the 2° and 10° observers. Other indices skip colorimetry altogether and score the difference between the two reflectance curves, unweighted or weighted by the colour-matching functions; formulation software uses these, and root-mean-square ΔE over several conditions, to rank candidate recipes before anything is made. They are quick and comparative, and do not predict a visible difference under any particular lamp.
The table builds pairs with a known answer. Each starts from a measured reflectance curve and adds a spectral variation that leaves X, Y and Z untouched under D65 for the 2° observer, at a stated fraction of the largest such variation that keeps the curve physical. Three things show. The index grows in proportion to the variation: for weathered asphalt it is 1.0, 2.6 and 5.4 ΔE*ab towards FL11 at 10%, 25% and 50%. The test illuminant matters as much as the pair: the same asphalt pairs score 0.3, 0.7 and 1.4 towards illuminant A, a ranking that belongs to this particular variation, not to tungsten light in general. And the last column shows the pairs separating under unchanged D65 when only the observer becomes 10°.
| Pair: measured curve and its constructed metamer | Largest gap between the curves | Index, D65 → A | Index, D65 → FL11 | Index, D65 → LED-B3 | ΔE*ab under D65 at 10° |
|---|---|---|---|---|---|
| Weathered road asphalt, variation at 10% of the largest possible | 1.0 points | 0.3 | 1.0 | 0.8 | 0.3 |
| Weathered road asphalt, variation at 25% of the largest possible | 2.5 points | 0.7 | 2.6 | 2.1 | 0.7 |
| Weathered road asphalt, variation at 50% of the largest possible | 5.0 points | 1.4 | 5.4 | 4.2 | 1.4 |
| Fresh pine plywood, variation at 10% of the largest possible | 3.6 points | 0.5 | 1.5 | 1.4 | 0.4 |
| Golden dry grass, variation at 25% of the largest possible | 4.3 points | 0.9 | 2.7 | 2.9 | 0.8 |
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
The CIE classifies the degree of colour mismatch caused by substituting a test illuminant for the reference illuminant as the special metamerism index for change in illuminant, formulated in its Colorimetry publication, and defines a separate special metamerism index for change in observer in CIE 80.
Source: CIE 080-1989 Special Metamerism Index: Change in Observer; CIE 015:2018 Colorimetry, 4th edition
FactModerate evidence
Colour-control software offers several indices of illuminant metamerism: the CIE special index, which assumes an exact match under the reference illuminant; the DIN 6172 index, which adjusts for pairs that do not match exactly there; the Nimeroff indices, which score differences between the spectral curves; and a root-mean-square of colour differences over several illuminant and observer combinations.
Caveat: An instrument maker's summary of the indices in its own software; the DIN standard itself was not read.
Source: Metamerism
StandardStrong evidence
ISO 18314-4 gives a formal calculation of illuminant metamerism for pairs of solid surface colours, expressed as a colour difference and intended as a quality measure when a match must hold under different illumination; ISO/CIE 23603 uses the special metamerism index for change of illuminant on virtual specimen pairs to grade daylight simulators.
Caveat: From the standards' catalogue abstracts.
Source: ISO 18314-4:2024 Analytical colorimetry — Part 4: Metamerism index for pairs of samples for change of illuminant; ISO 23603:2005 / CIE S 012/E:2004 Standard method of assessing the spectral quality of daylight simulators for visual appraisal and measurement of colour
Colourwise analysisModerate evidence
For pairs constructed from a measured weathered-asphalt spectrum to match exactly under D65 for the 2° observer, the change-of-illuminant index is 1.0, 2.6 and 5.4 ΔE*ab towards FL11 and 0.3, 0.7 and 1.4 towards illuminant A at 10%, 25% and 50% of the largest physically possible spectral variation.
Based on: Colourwise added a spectral variation with zero X, Y and Z under D65 (2° observer) to USGS splib05a record 9737, then computed CIELAB for both curves under each test illuminant relative to that illuminant's white and took ΔE*ab, in the manner of the CIE special metamerism index. Data are 10 nm band means over 380–730 nm.
Caveat: The second curve of each pair is constructed, not measured, and one particular variation was used; real colourant pairs disagree in other spectral regions and would rank the illuminants differently. Colourwise's calculation, not a value published by any standard.
Source: USGS Digital Spectral Library splib05a (Open-File Report 03-395); CIE datasets (colour-matching functions, illuminants); CIE 015:2018 Colorimetry, 4th edition
Colourwise interpretationModerate evidence
A metamerism index is only informative when the pair's difference under the reference illuminant is small; applied to a pair that plainly differs there, it reports the remainder of an ordinary colour difference.
Based on: The CIE index is defined for specimens with identical tristimulus values under the reference illuminant, and the correction used by practical indices exists only to remove a small residual.
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.