The Instrument Has Existed Since 1991. Nobody Has Pointed It at the Shelf.

There are two halves to the subject of skin color and cosmetics. One is rigorous, standardized, international, and thirty-five years old. The other has essentially no evidence base at all.

Guess which half the conversation happens in.

The instrument

Individual Typology Angle was defined by Chardon, Cretois and Hourseau in the International Journal of Cosmetic Science in 1991. Not in medicine. In cosmetic science, which then exported it to dermatology and from there into standards.

It works from CIE Lab coordinates. ITA in degrees equals the arctangent of the lightness value minus 50, divided by the yellow-blue value, converted to degrees.

The categories, as published: very light above 55 degrees, light from 41 to 55, intermediate 28 to 41, tan 10 to 28, brown from minus 30 to 10, dark at or below minus 30.

Two things nobody mentions about those bins. They are not equal intervals — brown spans forty degrees, light spans fourteen — so the scale compresses variation at the dark end, which is precisely the end where the compression does damage.

And the formula uses lightness and the yellow-blue axis. It does not use the red-green axis at all.

Hold that second point. It comes back.

Why it is not Fitzpatrick, and why the conversion tables are wrong

Both systems have six categories. That coincidence has been mistaken for equivalence for years, and conversion tables between the two circulate freely in the cosmetics and laser device industries.

Osto and colleagues tested it directly. Their conclusion, verbatim: these cannot be utilized interchangeably. Poor correlation between the six objective ITA categories and the six subjective Fitzpatrick categories.

They measure different things. ITA measures what color skin is. Fitzpatrick asks what skin does in the sun, and was derived from self-reported burning in white patients.

A 2025 prospective comparison of Fitzpatrick, the Monk Skin Tone scale, the Pantone guide and an objective colorimeter put it as plainly as it can be put: Fitzpatrick is not a proxy for skin tone. That study also found that color values extracted from dermoscopic images correlated poorly with colorimetry, and that the Monk scale captured differences in AI melanoma classification scores better than Fitzpatrick did.

And a 2026 paper in the British Journal of Dermatology comparing five instruments across multiple anatomical sites states its objective outright: to demonstrate that pigment cannot be assumed from National Institutes of Health race categories.

Which is the sentence this whole section has been circling.

ITA is not a curiosity — it is regulatory infrastructure

This is the part that surprised me.

Determining the minimal erythema dose in unprotected skin is the most critical step in producing an SPF number, and it disproportionately influences the final value.

Cole compiled ITA and minimal erythema dose data on 9,931 individuals, received from twelve laboratories on five continents, and fitted a polynomial regression that predicts one from the other.

That regression has been incorporated into ISO 24444, the in vivo sunscreen testing method that must be used to determine a subject’s minimal erythema dose.

So every SPF number on every sunscreen tested to that standard now depends on an ITA-based subject-selection regression built from about ten thousand people across five continents.

The measurement exists, it is international, it is embedded in the standard that governs the largest product category in this entire subject — and the cosmetics industry’s own shade vocabulary uses none of it.

The study that has never been done

I ran the searches — foundation and shade range and skin tone, set against availability, market and diversity, and then tighter versions of the same query. Nothing that came back counts shades against a population skin tone distribution.

There is no peer-reviewed study that counted the foundation shades available in a market against the measured skin tone distribution of that market’s population.

The entire forty-shades discourse from 2017 onward is trade press and brand marketing. Not measurement.

The irony is complete: the instrument to do it properly has existed since 1991, was built in cosmetic science, and sits in an ISO standard. Nobody has pointed it at a shelf.

What does exist is instructive by contrast. A 2024 study mapped lip color across 410 women — French and American Caucasian, African American, Hispanic American — by full-face hyperspectral imaging. A follow-up extended it to 514 women across the US, France and China, explicitly to build a lip color chart to guide product development.

Its key finding: important overlaps were found between ethnic groups.

That is the template for what a foundation shade study should look like. It has been done for lips. And its result undermines ethnicity-keyed shade segmentation as a product logic — the groups overlap, so sorting product ranges by ethnic category is sorting on the wrong variable.

The one place the shade complaint has been measured at all is tinted sunscreen, where it is a clinical issue rather than a cosmetic one. A content analysis of over-the-counter tinted sunscreens and their reviews found that most products provided only one color shade, and that tone incompatibility was the most commonly cited negative feature — with the vast majority of those comments coming from consumers with dark skin tones.

Review-text content analysis, not a product census. It is also the best published evidence for shade availability in any cosmetic category, which tells you how thin the shelf is.

One genuinely measured finding across shade ranges, and it runs counter to the usual complaint: a study of 32 foundation shades across four brands, applied at controlled thickness and tracked by spectrophotometer, found that darker shades showed less lightness reduction over time than lighter ones within each brand. Deep shades oxidized less, not more.

Undertone has no measurement basis

Warm, cool, neutral, olive. Every foundation counter in the world runs on this vocabulary.

There is no peer-reviewed classification, validation or reliability study of it. Nothing that defines undertone operationally, tests whether two people agree on it, or validates it against colorimetry.

And here is where the earlier detail returns. The measurable axes are the yellow-blue one and the red-green one. ITA uses lightness and yellow-blue — and discards red-green entirely.

Which means the field’s best objective instrument is, by construction, silent on the axis closest to what the industry means by warm against cool. No published work maps those coordinates onto the commercial vocabulary.

The closest thing to a study is recent and more interesting than a simple debunk. A 2025 paper had 170 participants rate faces under standardized lighting with randomized, tone-adjusted images, and found that perceived facial yellowness depended on the yellow-blue value, lightness, and diffuse reflection — surface scatter, a non-color property.

So undertone is not meaningless. It is under-determined: a perceptual attribute that depends on at least three parameters, one of which is not a color at all, being described with a single-axis word.

Two questions I cannot answer

Why shade ranges were historically narrow has an obvious economic story — deep shades sell slower per unit, so inventory carrying cost and markdown risk are higher, so assortment planning truncates the tails of the distribution. That argument is coherent and it is entirely unsourced. No operations, marketing science or economics paper on foundation shade assortment exists that I could find. It is a hypothesis, or it is industry testimony somebody would have to get on the record. It is not a finding.

And whether the post-2017 shade expansion persisted, or whether deep shades were quietly discontinued, has no published answer either. No peer-reviewed source, no regulatory dataset, no retailer disclosure.

That second one is worth dwelling on, because it is entirely tractable. Archived retailer catalogs exist. Somebody with them and a weekend could settle it. Nobody has.

Ten thousand people for a number, nobody for a shelf

The measurement science is excellent. ITA is objective, reproducible, standardized, and load-bearing inside the regulation of sunscreen. The instrument comparisons are honest about their own limits — one 2024 letter even warns that ITA is now being misused in the pulse oximetry bias debate, which is the right kind of caution to have about your own best tool.

The commercial side has a vocabulary with no validation, a shade-range controversy with no census, and an economic explanation with no paper behind it.

Those two halves have been within reach of each other for thirty-five years. And the reason the gap persists is not technical difficulty. Ten thousand people across five continents were measured to make an SPF number reliable. Nobody has measured a shelf.

This article is educational and is not a substitute for care from someone who can look at your skin. Anything new, changing, painful or spreading should be assessed rather than researched.