How much do colour values change between the 2° and 10° observers?

What switching the observer changes in the sum

Nothing about the sample or the light changes. The reflected spectrum is weighted by a different set of three colour-matching functions, and the white the result is judged against is recalculated with the same functions. ISO/CIE 11664-1 describes the 1931 set as representing fields of about 1° to 4° and the 1964 set fields larger than about 4°. The 1964 functions give more weight to short wavelengths in ȳ and have a differently shaped x̄ and z̄, so any spectrum whose energy sits in the blue or blue-green is reweighted most. A flat spectrum is reweighted too, but so is the white, by the same proportion, and the two cancel.

Integrated from the 10 nm band means used on this site, D65 white is X 95.04, Z 108.81 for the 2° observer and X 94.81, Z 107.22 for the 10° observer (Y = 100 for both).

Neutrals stay put; blues and greens move

The first table recalculates each of the site's twenty measured reflectance curves under D65 with both observers and gives the CIEDE2000 distance between the two results. Snow, gypsum, galvanised steel and white vinyl move by less than half a unit. Haematite, cinnabar and the plywood move by about one. The blue tarpaulin, lazurite, azurite and both living leaves move by three to a little over five: 6 of the twenty move by more than two. In the large cases it is a* that changes most, by roughly three to eight units, while L* moves by two at most. That ΔE is not a colour difference anyone could see. It is the size of the mistake made by reading a 10° value against a 2° standard.

Differences between two samples move much less

Tolerances are set on the difference between a standard and a batch, not on absolute values, so the second table matters more than the first. For pairs of measured materials with broadly similar curve shapes, ΔE00 under D65 changes by less than a unit and a half between observers even when the pair is twenty units apart. The exception is a metameric pair. The last row is a curve constructed to match weathered asphalt exactly for the 2° observer under D65: for the 10° observer the same two curves are 0.9 ΔE00 apart. A match that exists only because two different spectra happen to give equal sums for one set of weighting functions does not survive a change of functions. This is the instrumental form of observer metamerism.

Why there is no 2°-to-10° conversion

Three numbers cannot be turned back into the 36 or more from which they were summed. Many different spectra share one XYZ for the 2° observer, and they do not share one XYZ for the 10° observer, so no formula from L*a*b* (2°) to L*a*b* (10°) can be right for all of them. Software that appears to convert is recalculating from a stored spectrum. If a specification, a supplier's certificate and a quality-control reading are to be compared, all three must state the same observer and illuminant, and the standard should be kept as spectral data so that it can be recalculated when a customer asks for the other convention.

The data

How to check it yourself

1. Open a stored standard in your quality-control software and note its L*a*b* under D65/10°.
2. Change only the observer to 2° and note the values again; the software is recalculating from the spectrum.
3. Do the same for a batch reading and compare the two ΔE values: if they differ by more than a few tenths, the pair is metameric to some degree.
4. Check that the certificate, the specification and the instrument all name the same illuminant and observer before comparing any numbers.

What the evidence says

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/CIE 11664-1:2019 specifies two sets of colour-matching functions: the CIE 1931 set, representative of fields of about 1° to about 4°, and the CIE 1964 set, representative of fields larger than about 4° at light levels and with spectra for which rod receptors are not expected to take part.

  Source: ISO/CIE 11664-1:2019 Colorimetry — Part 1: CIE standard colorimetric observers

• Colourwise analysisModerate evidence

  Recalculating twenty measured reflectance spectra under D65 with the 2° and then the 10° observer moves their CIELAB values by between ΔE00 0.04 (fresh snow) and 5.4 (blue woven plastic tarpaulin); 6 of the twenty move by more than 2.

  Based on: Colourwise integrated each USGS splib05a reflectance spectrum against CIE D65 and each observer's colour-matching functions, all as 10 nm band means over 380–730 nm, converted to CIELAB relative to the D65 white for that observer, and took the CIEDE2000 distance between the two results.

  Caveat: Twenty specimens chosen to be visually distinct, not a sample of industrial colours. 10 nm data over 380–730 nm; a 1 nm calculation over 360–830 nm would differ slightly. The shift is a difference between two calculations, not a visible difference.

  Source: USGS Digital Spectral Library splib05a (Open-File Report 03-395); CIE datasets (colour-matching functions, illuminants); CIE 1964 colour-matching functions, 10 degree observer (data table)

• Colourwise analysisModerate evidence

  A reflectance curve constructed to match measured weathered asphalt exactly under D65 for the 2° observer is 0.9 ΔE00 from it under D65 for the 10° observer, while the measured, non-metameric pairs in the table change by less than a unit and a half.

  Based on: The second curve is the measured asphalt spectrum plus a quarter of the largest spectral variation that has zero X, Y and Z under D65 for the 2° observer and keeps reflectance between 0 and 1. Both curves were then compared with each observer.

  Caveat: The metamer is arithmetic, not a measured material. Real metameric pairs come from different colourants and may be more or less sensitive to the observer than this construction.

  Source: USGS Digital Spectral Library splib05a (Open-File Report 03-395); CIE datasets (colour-matching functions, illuminants); CIE 1964 colour-matching functions, 10 degree observer (data table)

• FactModerate evidence

  Observer metamerism can arise in production when a match is formulated under one CIE standard observer and later evaluated under the other.

  Source: Metamerism

• Colourwise interpretationModerate evidence

  CIELAB values for one observer cannot be converted to the other observer without the underlying spectrum, because spectra that share tristimulus values for one set of colour-matching functions do not in general share them for another.

  Based on: Follows from the definition of tristimulus values as three weighted sums of a spectrum, and is shown by the constructed pair in the table, which is identical for one observer and different for the other.

Related questions

Which colours are most affected by choosing 10° instead of 2°?

Those whose reflectance is concentrated in the blue and blue-green, and greens with a narrow peak. In the measured set the blue tarpaulin, lazurite, azurite and green leaves moved most; whites, greys and blacks barely moved.

Can I convert a D65/10° L*a*b* value to D50/2° with a formula?

Not reliably. A chromatic adaptation transform can approximate the change of illuminant, but nothing recovers the change of observer. Recalculate from the reflectance spectrum, or re-measure and export spectral data.

Does the white tile calibration depend on the observer setting?

No. Calibration fixes the instrument's reflectance scale, wavelength by wavelength. The observer and illuminant are applied afterwards in software, which is why one stored measurement can be reported under any combination.

Related topics

Go further

Sources

• ISO/CIE 11664-1:2019 Colorimetry — Part 1: CIE standard colorimetric observers — CIE (International Commission on Illumination) with ISO/TC 274 (Official standard or government)
• USGS Digital Spectral Library splib05a (Open-File Report 03-395) — U.S. Geological Survey (Clark, Swayze, Wise, Livo, Hoefen, Kokaly, Sutley) (Official standard or government)
• CIE datasets (colour-matching functions, illuminants) — CIE (International Commission on Illumination) (Official standard or government)
• CIE 1964 colour-matching functions, 10 degree observer (data table) — CIE (International Commission on Illumination) (Official standard or government)
• Metamerism — Datacolor (Manufacturer technical documentation)

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.