Why the test uses people
The sun protection factor is defined against a biological endpoint: the reddening of skin. That definition forces the method. A quantity measured on an instrument would be a different quantity, and the number on a pack would mean something else.
This is a source of persistent tension in the field. Human panel testing is slow, expensive and variable, and there has been sustained work on in vitro alternatives for decades. What has kept the in vivo method in place is that the thing being measured is defined in terms of skin, and any substitute has to be validated against the method it would replace.
The consequence for a reader is that the number on a pack came from people. Somewhere, a panel of volunteers was recruited, screened, marked out with a template and exposed to a graduated series of ultraviolet doses so that a manufacturer could print an integer.
The panel
Volunteers are recruited against criteria set by the standard: skin type within a defined range, no relevant skin condition, no recent ultraviolet exposure on the test area, no medication that alters photosensitivity, and informed consent. The test area is the back, because it is flat, large, easy to shield and rarely exposed.
The standard sets a minimum number of valid individual results before a factor can be reported, and testing continues, up to a maximum, until the statistical requirement is satisfied. The reason for that construction is that people differ. Two volunteers given the same product at the same quantity will not produce the same individual factor, and the reported result is a mean of a distribution rather than a single reading.
Establishing the baseline
Before any product is applied, each volunteer's unprotected response has to be established, because the factor is a ratio and the denominator is personal.
A series of small sub-areas is exposed to a graduated series of ultraviolet doses, and the sites are read after a defined interval, conventionally the following day. The lowest dose that produces a just perceptible reddening across the whole exposed sub-area is that individual's minimal erythemal dose. It is read visually, by a trained assessor, under standardised lighting.
Visual assessment is the part of the procedure that surprises people who assume the whole thing is instrumental. It is deliberate, and it is why assessor training and consistency form part of the method.
Application, which is the controlled variable
The product is weighed. Two milligrams per square centimetre of the marked test area is dispensed, distributed in small dots across the area and then spread with a gloved finger using a specified technique and a specified time, and the film is allowed to settle before exposure.
Every element of that sentence is a control. The quantity is weighed rather than estimated because the factor depends on it steeply. The spreading is timed because a film worked for longer behaves differently. The settling period exists because some formulations change as they dry.
This is the point at which the laboratory and the beach diverge completely, and it is why the application quantity the number assumes is the most consequential condition attached to the figure.
The solar simulator
The source is not the sun. It is a xenon arc solar simulator with filtering that gives it a specified spectral output within defined tolerances across the ultraviolet range, checked by spectroradiometry.
Standardising the source is what makes results comparable between laboratories and between years. Sunlight varies with elevation, season, latitude, altitude and atmospheric conditions, none of which can be held constant. A simulator can be held constant, and the standard specifies how constant.
The trade off is that the simulator's spectrum is a defined approximation of one condition of solar ultraviolet, and the factor is a property measured under that approximation. This is not a criticism. It is the reason the number is reproducible at all.
| Element | Controlled in the test by | In ordinary use |
|---|---|---|
| Quantity | Weighing to two milligrams per square centimetre | Not measured |
| Spreading | Specified technique over a specified time | By hand, variable |
| Source | Solar simulator with verified spectral output | Sunlight, varying continuously |
| Site | Flat prepared area on the back | Curved, mobile, partly covered |
| Interval | Defined settling period before exposure | Often none |
| Reading | Trained assessor, standardised lighting | Not read at all |
Framework of this publication. It describes how published rules and guidance fit together and is not a measurement, a survey or a study.
Exposure and reading
Protected sub-areas receive a graduated series of doses, scaled around the expected factor so that the reddening threshold falls within the series. The unprotected sub-areas receive their own series. Everything else on the back is shielded.
Sites are read after the defined interval by the trained assessor, and each volunteer's individual factor is the ratio of the protected threshold dose to the unprotected one. Individual results that fail the validity criteria in the standard are excluded rather than averaged in.
The reported factor is the arithmetic mean of the valid individual results, with a confidence interval calculated around it.
The statistics, and why the pack hides them
A panel mean has an uncertainty, and the standard requires the result to reach a stated precision before it can be reported. Where the precision is not met with the minimum panel, more volunteers are tested until it is, up to a maximum.
The printed figure discards all of this. The tested result is rounded down to the nearest permitted labelling figure, so a product tested somewhat above 30 is labelled 30. The confidence interval is not printed, the panel size is not printed, and the laboratory is not named.
Whether that is the right choice is a genuine question. Printing the interval would communicate the true precision of the measurement and would also be incomprehensible on a pack. The current convention chooses simplicity, and the cost is that a measured quantity with a known uncertainty is presented as a fact with none. That is examined further in measurement uncertainty and what a tested figure is.
The ethical dimension
A test that involves exposing volunteers to ultraviolet doses sufficient to produce visible reddening carries obvious ethical considerations, and they are managed through the same mechanisms as other human studies: screening, informed consent, dose limitation, ethical review and the exclusion of volunteers for whom the exposure would be inappropriate.
It is also the main practical driver behind the long standing effort to develop validated in vitro alternatives. Any replacement has to produce results equivalent to the in vivo method across a wide range of formulation types, which has proved harder than it sounds, because formulations differ in how they form a film on a substrate.
What the method costs, and why that shapes the market
An in vivo determination requires recruitment, screening, a controlled facility, a calibrated simulator, trained assessors and two visits per volunteer. It is not a quick test and it is not a cheap one, and both facts have consequences that reach the shelf.
Every reformulation that could affect the ultraviolet filter system needs retesting. A change of filter, of concentration, of emulsion type or in some cases of pack format is a change to the thing measured, and a figure obtained on the previous version supports a claim about the previous version. That is one reason product ranges are more stable than the pace of cosmetic marketing would suggest.
It also explains why the number is treated as an asset by the businesses that hold it. A determination is expensive enough that its result is defended, and the temptation at the margin is to describe it more generously than the report supports rather than to test again. That temptation is the origin of a large share of the claim problems this publication examines, and it is worth naming rather than pretending that claims go wrong at random.
What this means for reading a label
Three things follow. The figure is real, in the sense that it came from a measurement with a written method. The figure is narrow, in that it describes a property under one set of controlled conditions. And the figure is compressed, in that a distribution with an uncertainty has been reduced to an integer.
A reader who holds all three in mind reads a sun protection pack correctly. A reader who holds only the first treats a laboratory property as a promise about an afternoon.
What to take from this
Volunteers, a weighed film, a standardised source, a visual endpoint and a statistical requirement. The number on the pack is the compressed, rounded down output of that procedure, and everything the procedure controlled for is the list of things that will not be controlled in use.