Ask why plants are green and the answer comes back in one line. Chlorophyll absorbs red and blue light and reflects green, so green is what reaches your eye. Every textbook says a version of it, and the sentence is doing something sneaky: it describes green as the color chlorophyll sends back, as though the pigment picked it.
Build the leaf numerically and the causation runs the other way. A healthy leaf absorbs 87.8 percent of the photosynthetically useful photons that land on it, and the green you see is made of the 7.4 percent that got away through the one window chlorophyll covers badly. Green is not a reflection the plant produces. It is the residue of an absorber that did not quite finish the job.
The test that settles it is simple. Keep adding chlorophyll to the model leaf until it absorbs 96 percent of the available photons, and watch what happens to the color. It does not get greener. It goes #333939, a neutral grey with a chroma of 2.80. A leaf that was genuinely good at its job would not look green at all.
How the leaf was built
Nothing here is a paint swatch or an eyedropper on a photograph. Every color below comes out of a physical model fed with published measurements, so each number can be checked against its source.
- The leaf. PROSPECT-D, the radiative transfer model that treats a leaf as a stack of scattering plates loaded with chlorophyll, carotenoid, anthocyanin, brown pigment, water and dry matter. Specific absorption coefficients come from Feret and colleagues, unchanged.
- Chlorophyll itself. Measured absorption spectra for chlorophyll a and chlorophyll b in diethyl ether, from the PhotochemCAD database, scaled to their published molar extinction coefficients. These are the pigment on its own, outside any leaf.
- The light. D65 daylight for everything colorimetric, and the ASTM G173 global tilt spectrum when the question is how many photons are actually available to a plant outdoors.
- The color maths. CIE 1931 2-degree observer, sRGB, and CIEDE2000 for every difference quoted. The just noticeable difference threshold of 2.3 units is the one measured in the JND work on this site.
- The reference leaf. Chlorophyll 40 micrograms per square centimetre, carotenoids 10, no anthocyanin, structure parameter 1.5. Ordinary values for a healthy broadleaf in summer.
The reference leaf renders #4A6133, a dark olive. That is the same figure the model produced for the autumn color piece, which is the check I care about most: the two articles are running the same physics and agreeing to the byte. It also matches what a leaf looks like held in your hand, which is much darker and duller than the green anyone reaches for in a crayon box.
One limit worth stating up front. This is a single leaf under even illumination. A canopy adds gloss, shadow, sky color and the light bouncing between leaves, and all of those shift the number. What the pigments control is what is modelled here.
What chlorophyll actually absorbs
Start with the pigment on its own, because the measured spectrum is more lopsided than the usual diagram suggests. Reading the peaks straight off the PhotochemCAD files:
- Chlorophyll a, blue peak (the Soret band): 427.8 nm, extinction 111,700 per molar per cm.
- Chlorophyll a, red peak (the Qy band): 660.1 nm, extinction 85,265.
- Chlorophyll b, blue peak: 453.0 nm, extinction 159,100.
- Chlorophyll b, red peak: 643.0 nm, extinction 57,050.
Now the part the diagrams flatten. Across 500 to 600 nm, the whole green window, chlorophyll a averages an extinction of 4,243. That is 26.2 times weaker than its own blue peak. The pigment does not decline to absorb green. It absorbs green about as well as a pale tea absorbs anything, and then the leaf compensates by stacking so much of it in the path that even that feeble absorption adds up.
Which is exactly what the finished leaf shows. Split the visible range into thirds and measure how much of each the reference leaf takes in:
- Blue, 400 to 500 nm: 95.53 percent absorbed, 4.20 percent reflected.
- Green, 500 to 600 nm: 79.00 percent absorbed, 10.87 percent reflected.
- Red, 600 to 700 nm: 90.75 percent absorbed, 5.48 percent reflected.
Four fifths of the green light that hits a leaf never comes back out. At 550 nm, dead centre of the green window and the single most reflective wavelength anywhere in the visible range for a leaf, reflectance is 14.88 percent. A surface that returns fifteen percent of the light at its brightest point is not a green object in any ordinary sense. It is a dark object with a slight preference.
Chlorophyll is not the color of a plant
Here is where the standard answer really comes apart. If green were chlorophyll's color, a pot of chlorophyll should look like a leaf. Push the measured a-and-b spectra through Beer-Lambert absorption at a 3 to 1 molar ratio, which is roughly what a higher plant carries, and the extract lands somewhere else entirely.
Compared at matched lightness the two are 21.12 CIEDE2000 units apart, which is nine times the threshold at which two colors stop looking like the same color. The hue gap is 34 degrees, with the extract sitting firmly on the blue side of green and the leaf sitting on the yellow side. Chroma is 77 against 29, so the extract is nearly three times as colorful.
The difference is not the pigment. It is what the leaf does with it. Strip the scattering out of the model and let the same pigment load absorb in a single straight pass, and the color goes to #A5C500 at chroma 91.30, a fierce chartreuse that exists nowhere in a garden. Put the scattering back and the multiple internal bounces drag light through the pigment again and again, killing the chroma and pulling the hue 16 degrees toward yellow.
So when a search asks what color chlorophyll is, there are two honest answers and they are far apart. The molecule in a flask is an emerald near #008A43. The molecule doing its job inside a leaf contributes to #4A6133. Pointing at a leaf and saying that is the color of chlorophyll is like pointing at a brick wall and saying that is the color of iron oxide. Related, and not the same claim.
The greener it looks, the less it is catching
This is the measurement that reframes the question. Sweep the chlorophyll load from nothing to saturation and track two things together: how colorful the leaf looks, and how many useful photons it absorbs.
Chroma falls monotonically the whole way down the ladder, from 62.93 to 2.80, while absorptance climbs from 40.60 percent to 95.99 percent. There is no point on that curve where a leaf becomes both green and efficient. The two properties trade against each other, every step.
The bottom rung is the one worth sitting with. At 320 micrograms per square centimetre the leaf absorbs 96 percent of the photons available to it, and it renders #333939: chroma 2.80, which is inside the range most people would call grey. Its residual hue angle has even wandered round to 197 degrees, a cold blue, because at that density the last light escaping is coming from the scraps at both ends of the green window rather than its middle.
Run it backwards and the same logic holds. Take all the chlorophyll out and the leaf is #C89120, an autumn gold that comes from the carotenoid underneath, absorbing only 40.60 percent. Green is the colour of a chlorophyll layer caught partway between empty and saturated, and real leaves sit in that middle because pigment costs nitrogen and the returns flatten out fast.
That last point is the actual answer to the question. Plants are green because absorbing the final few percent is not worth what it costs, and the light they decline to chase happens to be the light our eyes are most sensitive to. We named the leftovers.
What the green window is actually worth
The obvious follow-up is how much a plant sacrifices by leaving that window ajar. Weight the ASTM G173 daylight spectrum by wavelength to get photon flux, since photosynthesis counts photons and not joules, and the accounting comes out smaller than the usual framing implies.
- Photons in 400 to 700 nm absorbed by the reference leaf: 87.81 percent.
- Share of all available photons sitting inside the 500 to 600 nm window: 34.96 percent.
- Photons escaping through the green window, as a share of everything available: 7.39 percent.
- Gain from closing the green window to the red band’s absorption rate: 4.66 percent more photons.
- Gain from becoming a perfect black absorber: 13.88 percent more photons.
Under five percent. That is the entire prize for solving what gets written up as a puzzle. And the model already shows the price: getting there means doubling the chlorophyll to 80 micrograms, which buys 92.80 percent absorptance for twice the nitrogen investment, then tripling it again for the next 2 percent. Diminishing returns arrive early and hit hard.
There is a popular counter-story that plants avoid green because sunlight peaks there, so the green window is a deliberate dodge of an overload. The premise does not survive contact with the spectrum. In photon terms, which is the currency photosynthesis actually spends, daylight at the ground peaks at 669 nm, deep in the red. The same point comes up measured from a different direction in the piece on what color the sun is, where the claim that the sun peaks in green turns out to be an artefact of which units you plot.
What is genuinely striking is where chlorophyll a’s red peak sits. Its Qy band is at 660.1 nm. The photon flux peak of ground-level daylight is at 669 nm. Those are nine nanometres apart. Whatever else evolution was doing, it parked the pigment’s strongest long-wavelength absorption almost exactly on the most abundant photon available.
Arp and colleagues argued in Science in 2020 that this is the real story: light-harvesting pigments are positioned not to maximise how much they absorb but to minimise how much their output jitters as light conditions swing. Absorbing at two points that flank a smooth region of the solar spectrum gives a steadier signal than absorbing everything would. On that reading the green window is not a gap to be explained away. It is the spacing between two well-placed absorbers, and the color is a consequence of where they sit.
Why is grass green, and why is every plant a different green
Grass, moss, pine needles, a rubber plant in an office and a lettuce all look green, and they all look like slightly different greens. The pigment is the same molecule in every one of them, so the variation has to come from load and structure. Running the model across plausible parameter sets for different plant types:
Every healthy plant in that table lands inside a hue band 16.7 degrees wide, from 118.3 to 135.0 degrees. Meanwhile lightness spreads across 12.9 units and chroma across 17.2. The greens of the plant world are one hue with a lot of different lightnesses, which is the same structure I found running the measurements for teal and for gray: vocabulary that sounds like it is naming hues is usually naming lightness.
Notice which dimension is doing the separating. A grass blade and a conifer needle come out 3.4 degrees apart in hue and 2.3 apart in lightness, and the hue difference is the one you could not report if the two were not side by side. That is a practical problem rather than a trivial one: hue differences of a few degrees are near the floor of what people can name reliably, so most of the describing we do about plant greens is really describing how dark they are.
And a stressed leaf breaks the band completely. Drop chlorophyll to 12 micrograms while leaving the carotenoid alone, which is what a plant short of nitrogen or water looks like, and it renders #867E2B at hue 99.5 degrees, nearly 20 degrees off the healthy cluster and heading for yellow. That is why a sick plant is legible at a glance from across a garden. The hue leaves the band that every healthy plant shares.
Most of what a leaf reflects, you cannot see
One last measurement, and it is the one that most changed how I think about the question. Follow the leaf’s reflectance past the red end of vision:
A leaf reflects 2.87 times more strongly in the near infrared than it does at its visible green peak, and the transition happens across about 40 nanometres. Remote sensing people call this the red edge and use it to count vegetation from orbit, because nothing else in a landscape does it.
Add up the radiant power a leaf actually reflects between 400 and 2400 nm and only 13.64 percent of it falls in the range human eyes respond to. Roughly six sevenths of a leaf’s reflection is invisible to us. If our vision ran a hundred nanometres longer, foliage would read as one of the brightest surfaces in an outdoor scene rather than one of the darker ones, and nobody would think to describe plants by their feeble green.
Which is a useful reminder that the question contains a hidden clause. Plants are green to us. The color is a fact about a particular three-cone visual system sampling a narrow slice of what the leaf is doing. Some birds and insects, with photoreceptors we do not have, are reading a different leaf. The same argument applies to what dogs can see, where the missing receptor changes which distinctions exist at all.
What I would retire
The sentence I would drop is that chlorophyll reflects green light. It absorbs 79 percent of it. Nothing is being reflected in the sense that word implies, any more than a nearly opaque curtain reflects the daylight that leaks around its edge.
The better sentence is longer and worth the extra words. Chlorophyll is an almost total absorber with one weak region in the middle of the visible range, plants stack enough of it to catch most of even that weak region, and the small amount that still escapes is the color we named the whole kingdom after. The green is a measure of how much the plant gave up on, and the number is 7.4 percent of the photons it could have had.
It also explains why the green of a real leaf keeps disappointing people who try to paint it. The web colour named green sits at a chroma of 71.85, and the one named lawn green at 108.73. A leaf is 29.37, so the nearest common name is more than twice as colorful as the thing it names. The mental image is built from what the color means rather than from what it measures, which is the same failure mode behind memory colors generally: people remember grass as more saturated than grass has ever been.
Test it on yourself
Judging a green in isolation turns out to be genuinely difficult, and the site is built around demonstrating exactly that. The color memory game shows a color and asks you to find it again from memory, which is where the gap between remembered green and measured green shows up most clearly. If you want to test the narrow-hue-band claim directly, the hue sort variant makes you order colors that differ by only a few degrees, and spot the difference pushes the same limit the other way.
For the numbers side, the hex variant makes you commit to coordinates rather than adjectives, and the catalogue of plant greens in shades of green shows how much of the vocabulary is doing lightness work. The companion pieces on why leaves change color and why the sky is blue run the same modelling approach on the two other questions people ask most about color in nature.