Sorting colour is harder than naming it
Almost everyone can look at a square of colour and say "that is green". Far fewer people can take twelve greens that differ by a few degrees of hue and put them in the right order. The two tasks use different parts of the same visual system. Naming relies on broad categories your language gave you; ordering relies on the raw discrimination limit of the three cone types in your retina and the opponent channels that compare them. This hue test targets the second one, because that is where real differences between people show up.
How the four rows are built
Each row is a continuous arc of hue at a fixed saturation and lightness, cut into twelve steps. The first row walks from red through orange into yellow, the second from yellow through green into teal, the third from teal through cyan into deep blue, and the fourth from violet through magenta back toward red. Holding saturation and lightness constant matters: if the caps also varied in brightness you could sort them by lightness alone and never judge hue at all. Every step inside a row is the same size, so no row contains an easy giveaway jump.
The two end caps are locked in place and marked with a small dot. They act as anchors, exactly as the reference caps do in a physical arrangement tray. Without them you could produce a perfectly smooth gradient running backwards and there would be no principled way to score it. The ten caps between the anchors are shuffled with a bounded shuffle every time you start, so a second attempt never gives you the same starting arrangement.
How your accuracy score is calculated
When you submit, the test compares where each movable cap ended up with where it belongs. A cap placed one position off contributes one point of error, two positions off contributes two, and so on. Those displacements are summed across all four rows and divided by the worst arrangement that is mathematically possible for rows of this length, then flipped so that a higher number is better. The result is a straightforward 0 to 100 accuracy figure. A single adjacent swap inside one row costs about one point; a row you gave up on and left untouched costs about sixteen.
| Score | Reading | Roughly |
|---|---|---|
| 100 | Perfect arrangement | About 1 in 8 attempts |
| 95–99 | Excellent hue discrimination | Top 20% |
| 85–94 | Normal — a few neighbour swaps | Middle 50% |
| 70–84 | Below average, or a rushed attempt | Bottom 25% |
| Under 70 | Retest on a brighter screen before reading anything into it | Bottom 10% |
Why one row always feels impossible
Hue discrimination is not evenly spread around the colour wheel. The eye is unusually sharp near the yellow region and near the blue-green boundary, and unusually blunt in the stretch between cyan and deep blue, where a large change in hue angle produces only a small perceived change. That is a property of how the short wavelength cones are wired, not a personal failing. Most people who score in the nineties lose all of their points in a single row, and it is usually the blue one. If your errors are instead spread evenly across all four rows, the likely cause is haste or a dim display rather than colour vision.
Getting a result worth trusting
Screen conditions dominate this test more than any other on the site. Turn brightness up, switch off night mode, warm-light filters and any accessibility colour correction, and avoid direct sunlight or a lamp reflecting off the panel. Sit square to the screen — cheap laptop panels shift hue noticeably when viewed at an angle, which can make a correctly sorted row look wrong and tempt you into breaking it. Work outward from the anchored end caps rather than trying to judge the whole row at once, and compare each cap only with the one beside it. Finally, do not rush: there is no timer, and the difference between a score of 84 and a score of 97 is usually two extra minutes of checking.
How this differs from a colour blindness plate
A plate test asks whether you can see a number hidden in a field of dots. It is a pass or fail screen for the common red-green deficiencies and it says nothing about how finely a person with normal vision can discriminate. An arrangement test like this one measures the resolution of your colour sense on a continuous scale, and it can pick up mild acquired changes that a plate test passes without comment. It is not a substitute for either, though. If you consistently score badly across every row on a good screen, or if your score has dropped noticeably from previous attempts, that is a reason to see an optometrist rather than a reason to trust a web page.
Frequently asked questions
What does a hue test actually measure?
It measures how finely you can tell neighbouring shades apart. Sorting caps into a smooth gradient forces you to judge very small hue steps, which is a different skill from naming a colour correctly. A person can name every colour on this page and still misplace half the caps.
Is a perfect score of 100 normal?
It is uncommon but not rare. Roughly one person in eight places every cap correctly on a good screen. Most people land between 82 and 95, because one or two swaps inside the blue-green row are almost expected.
Does this replace a clinical colour arrangement test?
No. A clinical arrangement test uses physical caps under standard daylight illumination. Your monitor, its brightness and the room light all shift the result here, so treat this as a screening exercise rather than a diagnosis.
Why is the blue and green row the hardest?
Human hue discrimination is not uniform. Sensitivity peaks around yellow and around blue-green boundaries, and the region between cyan and deep blue produces the smallest perceived difference per degree of hue, so caps there look nearly identical.
How can I improve my score?
Turn your screen brightness up, disable any night or warm colour filter, sit square to the display and compare each cap only with its immediate neighbours rather than with the whole row. Working outward from the two fixed end caps also helps.
Is this hue test free?
Yes, the hue test is completely free online, needs no account and no payment, and you can retake it as many times as you like. The cap order is reshuffled on every attempt.