The two pigments under the word "undertone"

Skin is not one colour-bearing substance. It is mostly two. Melanin sits in the epidermis and absorbs broadly, more strongly at short wavelengths. Haemoglobin sits below it in the blood of the dermis and absorbs in sharper bands that shift with how oxygenated it is. Almost everything anyone means by "undertone" is some ratio of those two, seen through tissue that scatters the light on its way back out.

That ratio survives a camera, which is the part most people are surprised by. Tsumura and colleagues showed in 1999 that the spatial variation of melanin and haemoglobin can be pulled apart from an ordinary colour photograph, and refined the method for rendering skin in 2003. It is worth being exact about what that is: it separates two independent components under a model — assuming there are two pigments, that their quantities vary independently across the skin, and that the arithmetic is linear in optical density. It estimates. It does not assay you.

Where dermatology does put a number on skin colour, the number is not warm or cool. It is the Individual Typology Angle, built from two CIELAB coordinates — lightness and yellowness — as ITA = arctan((L* − 50) / b*), in degrees. Chardon, Cretois and Hourseau defined it in 1991, on a Caucasian population, and named five bands; the dark band below −30° was added later, by Del Bino and Bernerd in 2013. The bands are conventions laid across a continuous number, and where to cut them is still argued about.

So there is a standard measure of how light and how yellow skin is, and there is no standard measure of warm against cool. The nearest anyone has come to testing the idea directly is a 2024 paper by Branigan and colleagues, which set instrument readings beside human raters. A single rater on their own was unreliable — intraclass correlations below .30 — and their judgements of skin redness turned out to be essentially uncorrelated with the measured a* coordinate, around −.09 and −.19, which is fractionally the wrong way round. Yellowness they saw considerably better, at .64 and .54. None of that proves undertone is nothing. It does mean the word has not yet named a quantity two people measure the same way twice, and that the red half of it — the half the entire warm-cool axis turns on — is the half human eyes are worst at.

Which brings it to the wrist. The vein test has no literature behind it: not a contested result but an absent one. The biomedical indexes return nothing for it, and every source teaching it is a blog or a course selling the reading. The optics argue against it too. Kienle and colleagues asked in 1996 why veins look blue at all, and answered with vessel depth, vessel diameter, how oxygenated the blood is, and how the skin above scatters light — not one of which is a property of pigment.

The mechanism is not even the one usually told. A vein is not sending more blue light to your eye than red. Red light penetrates deep enough to reach the vessel and is heavily absorbed there, so the skin over a vein is short of red compared with the skin beside it, and vision, which judges colour by comparison, reads that shortfall as blue. Bring the same vessel nearer the surface and it reads red instead. What the inside of your wrist tells you is how deep your veins are.

The useful residue: the two-pigment picture is real, and it is photographable. The warm-cool label sitting on top of it is a convention, and the profession's own literature says so — Perrett and Sprengelmeyer noted in 2021 that stylists' categories are not defined scientifically and that the colours they advise are inconsistent between them. The same paper found the underlying intuition is not baseless: across 171 people there was strong agreement that cooler hues suit fair skin and warmer ones suit tanned. That is a consensus of taste, which is a real and useful thing to know, and a different thing from a measurement. Percol's own reading is a judgement about colours next to a face. It is worth knowing which of the two you are being handed.