Short answer
R9 is the CIE's special colour rendering index for a strong red test sample. It is not part of the general CRI (Ra), which averages only eight softer colours, so a lamp can post Ra 80 or more while rendering saturated reds — and with them skin, wood and meat — poorly. TM-30's red-bin fidelity and chroma shift (Rf,h1 and Rcs,h1) do the same job on a more consistent scale.
CIE 13.3 defines fourteen test-colour samples. The first eight, fairly desaturated, produce the general index Ra. Six more — strong red, strong yellow, strong green, strong blue, a complexion colour and a foliage green — produce special indices R9 to R14, reported separately if at all. R9 is the strong red. Because it sits outside the average, a spectrum can be weak in the deep red, where saturated reds reflect most strongly, without Ra noticing. This is the gap the site's colour rendering page warns about; this page explains why it is so common and what to ask for instead.
A basic phosphor-converted LED has little output beyond about 640 nm. A saturated red surface reflects mostly in exactly that region, so under such a lamp it has less light to return and looks darker and browner than under daylight or halogen. Adding a red phosphor fixes this but costs efficacy, because the eye is less sensitive there. The CRI scale then exaggerates the problem: the CIE 1964 U*V*W* space it uses is non-uniform in the reds, so R9 runs on a different scale from the other indices, drops steeply, and can even go negative — a possibility often wrongly described as impossible.
A lamp that boosts red saturation can also lose CRI points, because the index treats any departure from the reference as a fault — even a change most people like.
TM-30 splits colour space into 16 hue bins; bin 1 is the reds. Rf,h1 is the average fidelity of the samples in that bin, on the same 0–100 scale as Rf, and Rcs,h1 is their average change in chroma as a percentage: negative means reds are muted, positive means boosted. Together they separate two cases R9 lumps together — a lamp that desaturates reds and one that oversaturates them can have the same fidelity but opposite effects. Among the 99 individual samples, the one whose fidelity tracks CIE R9 most closely is sample 7, but TM-30 recommends the bin values because they average several similar colours and are more robust.
Deep red is not a niche colour. Skin gets much of its warmth from blood beneath it, timber and leather are red-brown, cured meat and many fruits depend on red, and red is among the commonest accent colours in textiles and branding. That is why a high-R9 or positive-Rcs,h1 lamp can make a room feel warmer and more alive than a lamp of identical CCT and Ra. The calculation below shows the size of the effect on one saturated red surface: seen by an eye adapted to each light, its chroma ranges from somewhat below its daylight value to about a quarter above it, and its hue swings by up to about 16°, depending on where each spectrum puts its red power.
| Illuminant | CIELAB chroma C*ab | Chroma vs daylight D65 | Hue shift vs D65 |
|---|---|---|---|
| Daylight, CIE D65 | 59 | 0% | +0° |
| Daylight, CIE D50 (graphic arts viewing) | 61 | +2% | +2° |
| Incandescent tungsten, CIE A | 63 | +6% | +8° |
| Cool-white fluorescent, CIE FL2 | 52 | -13% | +11° |
| Narrow-band tri-phosphor fluorescent, CIE FL11 | 63 | +6% | -5° |
| Phosphor-converted white LED, CIE LED-B3 | 57 | -4% | +5° |
| Red-green-blue mixed LED, CIE LED-RGB1 | 74 | +25% | -5° |
Why: Low deep-red output: low R9 and negative Rcs,h1 despite acceptable Ra.
Fix: Choose a lamp that publishes a clearly positive R9, or TM-30 red chroma shift near zero or slightly positive.
Why: Showroom lamps boost red chroma; the camera and later screens do not reproduce it.
Fix: Photograph under a high-fidelity source and compare with the showroom by eye before approving.
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
CIE 13.3 uses eight test-colour samples for the general colour rendering index Ra and six further samples, including a strong red, for special indices.
Source: CIE 013.3-1995 Method of Measuring and Specifying Colour Rendering Properties of Light Sources
FactStrong evidence
Because CRI uses the non-uniform CIE U*V*W* space, R9 is on a different scale from other indices, sources that increase red chroma tend to lose CRI points, and CIE 13.3 values can be negative.
FactStrong evidence
TM-30's red-bin fidelity Rf,h1 shares R9's purpose, and of the 99 samples the one most correlated with CIE test sample 9 is sample 7.
Colourwise analysisModerate evidence
For one saturated red surface seen by an eye adapted to each light, apparent chroma is lower than under D65 for the halophosphate fluorescent FL2 and the phosphor LED LED-B3, and higher for the RGB LED LED-RGB1.
Based on: Colourwise calculation: a smooth reflectance fitted to sRGB (180, 30, 40), integrated with each CIE illuminant at 10 nm, Bradford-adapted to D65 and expressed in CIELAB; see the table.
Caveat: One synthetic surface, one adaptation model and 10 nm sampling; not a CIE R9 or TM-30 calculation.
Source: CIE datasets (colour-matching functions, illuminants)
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.