The Farnsworth-Munsell 100 Hue Test: What It Measures

How the classic arrangement test works, how it is scored, and what changes when it moves onto a screen.

Short answer: it is an arrangement test. You put 85 colored caps into a smooth hue sequence across four trays, and the score comes from how far each cap sits from its correct neighbors. It measures hue discrimination, and a bipolar pattern of errors can point to an axis of color confusion. It does not measure color taste, color memory or artistic ability, its normal range depends heavily on age, and an online version is a different instrument from the physical one.

What Dean Farnsworth built, and why

The test was published in 1943 by Dean Farnsworth, working for the United States Navy, alongside the shorter dichotomous test that carries his name. The problem he was solving was practical: selecting personnel for tasks where fine color judgement mattered, using something more informative than a pass-or-fail screening plate. His answer was to make the observer produce an ordering, which reveals not just whether you can discriminate but where your discrimination breaks down.

The caps are Munsell surfaces of roughly constant value and chroma, varying in hue around the circle. That constancy is the whole design: if the caps differed in lightness you could sort them by lightness and never touch hue at all. Anyone who has played an arrangement game on a screen has felt the temptation to do exactly that.

How it is scored

The scoring is arithmetic and unforgiving. Every cap carries a hidden number. For each cap you take the absolute difference between its number and the number of the cap on its left, add the absolute difference to the cap on its right, and subtract 2, because a cap in its correct place between correct neighbors already accumulates a difference of 1 on each side. Sum this across all caps and you have the total error score. A perfect arrangement scores zero.

Plotted on a polar chart, the per-cap error scores make the shape people actually look at. Scattered small errors all round the circle read as ordinary imprecision. A pair of peaks on opposite sides of the circle reads as an axis, and the position of that axis is what suggests a red-green or a blue-yellow confusion.

The score is less solid than it looks. The traditional method has measurable test-to-test variability, and an alternative scoring based on maximum-likelihood estimation was proposed specifically to reduce that variability and improve discrimination between levels of overall color discrimination, while keeping the resulting error scores comparable with the traditional ones. That is a useful reminder: the number is the output of a scoring rule, and the rule is a choice.

Age is not a footnote

The most common misuse of this test is quoting a score without an age. Normative work covering each year of age from 5 to 22 and each decade from 30 to 70 shows a clear pattern: children score poorly, performance improves through adolescence, is at its best in early adulthood, and worsens again with age. The same total error score can be entirely typical for one person and a clear outlier for another.

What it can and cannot tell you about color vision differences

An axis in the error plot is suggestive, not diagnostic. Studies that compared the 100 Hue test with threshold-based methods, such as measuring discrimination ellipses directly, have found that separating protan from deutan observers on the arrangement test alone is not reliable enough to stand by itself. The 100 Hue is good at showing that hue discrimination is impaired and roughly where; it is weaker at saying precisely which type of difference is responsible. That work is a useful corrective to online tests that hand out a confident diagnosis after five minutes.

If you want the background on the underlying differences themselves, our page on color vision differences covers what causes them and how common they are, and inclusive color teaching covers what to do about it in a classroom.

What changes when the test moves to a screen

Computer-based versions of the 100 Hue test have been built and evaluated in the ophthalmic literature, so a screen version is not automatically invalid. But a validated instrument on calibrated hardware and a web page you found are not the same thing, and it is worth being precise about what changes.

One consequence deserves emphasis. On a screen the caps are rendered through a nonlinear transfer function, so hues that were matched in lightness as pigments may not be matched in lightness as pixels. That gives you a lightness cue the physical test deliberately removed, and it can make an on-screen arrangement task easier for the wrong reason.

Where Huesort fits, stated plainly

Huesort uses the same core idea: order a set of shades into a smooth sequence. It is deliberately short, it runs daily, and it is scored so you can watch a trend. What it is not is a clinical test, a diagnosis, or a substitute for an eye examination, and we would rather say that here than bury it in a disclaimer.

What it is genuinely good for is finding out which region of the hue circle you personally find hardest, which is useful information for practice, and for building the habit of comparing neighbors rather than judging caps in isolation. That habit, comparing across an edge instead of against a memory, is the single most transferable thing the task teaches, and it connects directly to the just-noticeable difference work.

If your arrangement is consistently much worse in one region across many sessions, or if you have a family history of color vision deficiency, or if something has changed recently in how you see color, book a proper examination. An online score is a reason to go, never a reason not to.

Keep reading

• Color vision differences: what the axis in an error plot is pointing at.
• The just-noticeable color difference: the sensitivity that arrangement tasks are indirectly probing.
• Mach bands and banding: why a sequence can look wrong even when it is right.

Sources: Farnsworth, D. (1943), The Farnsworth-Munsell 100-Hue and dichotomous tests for color vision, Journal of the Optical Society of America 33(10), 568; Kinnear, P. R. and Sahraie, A. (2002), New Farnsworth-Munsell 100 hue test norms of normal observers for each year of age 5-22 and for age decades 30-70, British Journal of Ophthalmology 86(12), 1408; Craven, B. J. (1997), A method for increasing the scoring efficiency of the Farnsworth-Munsell 100-Hue test, Ophthalmic and Physiological Optics 17(2), 153; Bento-Torres, N. V. O., Rodrigues, A. R., Cortes, M. I. T., Bonci, D. M. O., Ventura, D. F. and Silveira, L. C. L. (2016), Psychophysical evaluation of congenital colour vision deficiency: discrimination between protans and deutans using Mollon-Reffin's ellipses and the Farnsworth-Munsell 100-Hue test, PLOS ONE 11(4), e0152214; Ghose, S., Parmar, T., Dada, T., Vanathi, M. and Sharma, S. (2013), A new computer-based Farnsworth Munsell 100-hue test for evaluation of color vision, International Ophthalmology 34(4), 747; Munsell Color, Farnsworth-Munsell 100 Hue Test.