Short answer
Less than the confident phrasing of most accounts suggests. In Colourwise's dataset of 45 organism records, each drawn from peer-reviewed work on how a colour is made, 6 have a function that was experimentally tested in the cited studies, 9 rest on comparison or correlation, 11 are proposals, and 19 have no function established at all. Mechanism is usually measured; purpose is usually inferred.
Saying that a hummingbird's gorget is a stack of melanosome layers is a claim about anatomy and optics. It can be checked with a microscope and a spectrometer, and different laboratories get the same answer. Saying that the gorget is 'for' attracting females is a claim about evolutionary history and present-day behaviour: that females respond to it, that the response affects which males breed, and that this is why the structure persists. That claim needs a different kind of study, usually a manipulation in the field, and it is far harder to obtain. The two are routinely reported in the same sentence, in the same confident voice. Throughout this section they are kept apart and graded separately.
A function is graded experimentally tested when the cited work manipulated something and measured the outcome the function predicts — survival of prey models, mate choice, photosynthetic yield. It is comparative or correlational when it rests on association: across species, across individuals, across seasons, or between a measured colour and a model of a viewer's vision, without the outcome being manipulated. It is proposed when the authors or reviewers offer it without testing it, often in a single sentence at the end of an optics paper. And it is not established when the cited work declines to assign a function or argues there may be none. The grade describes the cited studies. It is not a verdict on whether the function is real.
The tally is in the table below: 6 tested, 9 comparative, 11 proposed, 19 not established, with 4 of the total marked as contested because published studies disagree. Multilayer reflectors and photonic crystals — the most intensively characterised structures — have the highest share of proposed or unassigned functions, because the papers that characterise them are physics papers. These numbers describe a set of records chosen to illustrate mechanisms. They are not a sample of nature or of the literature, and a different selection would give different proportions. What they do show is that it is entirely normal for a colour to be understood in exquisite physical detail while its function remains a suggestion.
Even the tested cases carry limits. Most camouflage and warning-colour experiments use artificial prey — paper triangles, pastry or clay models — because real animals move, hide and cannot be randomised; the result shows that a pattern can protect, on that background, against those predators. Laboratory choice tests show what an animal can perceive or learn, not what it encounters. A function can be tested and still be contested: bumblebees learn floral iridescence in an arena, and optical measurements suggest real petals do not present it in daylight. And a tested function need not be the visible one — the blue sheen of a shade begonia accompanies a measured gain in photosynthesis, with no evidence that the blue itself is seen by anything.
A colour may also have no function. Fluorescent porphyrins in mammal hair and the yellow of an autumn leaf are both explained adequately as by-products of other chemistry.
Colourwise writes 'associated with', 'has been proposed' or 'evidence suggests' unless an experiment tested the function, and names the experiment when one did. 'The peppered moth's dark form survives better on dark bark' is supported by releases of thousands of moths. 'The Morpho's blue is for signalling to rivals' is not supported by anything in the optics papers that describe the scale. Readers meeting an unqualified 'for' elsewhere can ask three questions: was anything manipulated, was the outcome survival or reproduction and not merely visibility, and was the viewer's own vision taken into account? The remaining pages of this section apply those questions to camouflage, warning colour and sexual signalling in turn.
| Primary mechanism | Records | Experimentally tested | Comparative or correlational | Proposed | Not established |
|---|---|---|---|---|---|
| Pigment (selective absorption) | 12 | 2 | 3 | 0 | 7 |
| Thin-film interference | 1 | 0 | 0 | 0 | 1 |
| Multilayer reflector | 11 | 2 | 0 | 6 | 3 |
| Diffraction grating | 3 | 1 | 1 | 0 | 1 |
| Photonic crystal (2D or 3D) | 4 | 0 | 2 | 1 | 1 |
| Coherent scattering, quasi-ordered | 3 | 0 | 0 | 0 | 3 |
| Incoherent scattering | 2 | 0 | 0 | 2 | 0 |
| Fluorescence | 3 | 0 | 1 | 0 | 2 |
| Bioluminescence | 4 | 1 | 0 | 2 | 1 |
| Chromatophore system | 2 | 0 | 2 | 0 | 0 |
| All organism records | 45 | 6 | 9 | 11 | 19 |
| Organism | Colour | Proposed function | Evidence grade | What the evidence is |
|---|---|---|---|---|
| Peppered moth, dark form (Biston betularia f. carbonaria) | Sooty black, against the typical form's speckled pale grey | Concealment from birds on dark, soot-covered bark. | Experimentally tested | A six-year release of 4,864 moths found stronger bird predation on the dark form on today's cleaner bark, with daily selection of about 0.1 against it. |
| Asian jewel beetle (Sternocera aequisignata) | Iridescent green shifting to blue and purple | Camouflage: changing colour and gloss make the outline hard to pick out. | Experimentally tested | In two field experiments, models carrying real iridescent wing cases survived bird predation best and were hardest for human searchers to find; the authors argue camouflage explains it better than warning colour. |
| Fireflies (Family Lampyridae) | Yellow-green flashes | Courtship signalling and mate choice. | Experimentally tested | Reviewed evidence from Photinus shows females choosing mates on male flash characteristics; predators exploit the same signals. |
| Shade-dwelling begonias (Begonia spp.) | Blue sheen on the leaf in dim light | More efficient photosynthesis in very low light. | Experimentally tested | Photosynthetic yield was measured directly and compared with ordinary chloroplasts. |
| Flower-of-an-hour (Hibiscus trionum) | A faint blue-to-ultraviolet sheen over dark pigment at the petal base | A cue for pollinating bees. | Experimentally tested; contested | Bumblebees learned to use iridescence as a cue on artificial targets in the laboratory; a later optical study found the signal too weak to matter under natural light. |
| Red-osier dogwood (Cornus stolonifera) | Red autumn leaves | Screening the leaf from excess light while nutrients are recovered. | Experimentally tested; contested | Red leaves recovered from a high-light treatment that damaged yellow ones. A later review treats the adaptive value of autumn colour as unresolved, with signalling to insects a rival explanation. |
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 analysisModerate evidence
Of 45 organism records in the Colourwise biological-colour dataset, 6 have a function graded experimentally tested, 9 comparative or correlational, 11 proposed and 19 not established; 4 are marked contested.
Based on: Each organism record in the Colourwise biological-colour dataset is counted once, under its primary mechanism and its function-evidence grade. Non-living analogies are excluded. The records were chosen to illustrate mechanisms, not sampled from nature, so the counts describe this dataset and the studies it cites — not how common each mechanism or each kind of evidence is among living things.
Method: Each record graded by Colourwise from the cited papers' own description of what was manipulated and measured.
Caveat: A count over an illustrative, hand-selected dataset. The grades reflect the studies Colourwise cites for each record and would change if further studies were added.
Source: The biology of color; Physics of structural colors; Gold bugs and beyond: a review of iridescence and structural colour mechanisms in beetles (Coleoptera)
FactStrong evidence
A review of iridescence mechanisms in beetles, which classifies the structures in detail, describes the functions of that iridescence as putative.
FactStrong evidence
The introduction to a themed journal issue on animal camouflage noted that much of the classical theory had, until recently, received little rigorous experimental testing.
Source: Animal camouflage: current issues and new perspectives; Disruptive coloration and background pattern matching
FactStrong evidence
Photosynthetic quantum yield measured in Begonia iridoplasts was 5–10% higher under low light than in ordinary chloroplasts, an experimentally tested function for a structure whose visible blue reflection has no demonstrated role.
Source: Photonic multilayer structure of Begonia chloroplasts enhances photosynthetic efficiency
Colourwise interpretationLimited evidence
The gap between how well biological colour mechanisms are characterised and how well their functions are tested is largely a consequence of who does the work: structures are measured by physicists and materials scientists, and functional claims in those papers are typically a closing suggestion.
Based on: Colourwise's reading of the dataset: records sourced from optics and materials journals carry most of the 'proposed' and 'not established' grades, while 'tested' grades come from behavioural-ecology and plant-physiology studies.
Caveat: An observation about the cited literature, not a measured property of the field as a whole.
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.