Ten fair-skinned volunteers lay face down in the sun with rows of squares marked on their backs in opaque tape, each square 2.5 centimeters across. Some squares were bare skin. Others were under an experimental lotion containing 8 percent homosalate, a sunscreen chemical. A meter beside them counted the sunburning ultraviolet as it arrived, and at each step of the count someone covered one more square to end its dose. The air was about 35 degrees Celsius and 85 to 90 percent humid, and some of the volunteers were sweating. 1
The same ten had already been tested indoors under a 150-watt xenon lamp, in a room kept at 22 degrees. On bare skin the lamp and the sun agreed, and the dose that just reddened each person was about the same under both. Under the lotion they disagreed. The lamp gave the lotion a sun protection factor, or SPF, of 3.6, and the sun gave it 1.6, or 2.4 when the team compared patches of equal redness. 1
Three of the four researchers worked in the photobiology department of Schering-Plough, the maker of Coppertone, in Memphis, and the fourth at Temple University in Philadelphia. They warmed the backs of five volunteers to 33 to 35 degrees with an infrared heat lamp and tested them under the xenon lamp again. The lamp's SPF fell to 2.8, against 2.2 in the sun. "The need for standardized indoor controlled testing to make data and product comparisons possible is clear," they wrote in 1978. Fifteen years later, when the US Food and Drug Administration proposed to drop sunlight from sunscreen testing, it cited their paper. 1; 2
Of the 784 studies of sunlight and the body in our evidence map, 588 used a lamp: 405 a solar simulator, a lamp built to copy the ultraviolet in sunlight, and 183 another lamp or device. Ninety-eight put people, skin, or cells under the open sky, and 18 used daylight through a window. Another 72 measured no light at all and used a stand-in for it, such as latitude, the season, or a questionnaire about time outdoors.
Where researchers have put the same skin under a lamp and under the sun, the answers differed: a sunscreen protected less in the sun, and in Sydney the lamp reddened skin at a lower dose. The SPF on a sunscreen sold in the United States, including the one in your bathroom, is measured only under a lamp, the one light source the FDA has named for the test since it proposed dropping sunlight in 1993. 1; 3; 2; 4
The records behind this site have no field for who paid for a study. In the papers we read by hand, sunscreen, cosmetics, food, and drug companies paid for many of the lamp experiments on skin, and one of them supplied the lamp. 5
This page covers how light medicine moved between the sun and the lamp, how American sunscreen testing dropped sunlight, what a lamp leaves out, how a dose of light is counted, what the largest studies measure in place of light, who pays, and how we chose the studies this site is built on. Each study is named by its exposure: real sun, a lamp, cells, animals, or a stand-in such as latitude or a questionnaire.
Key facts
- Of the 784 studies in our evidence map, 588 used a lamp and 116 used real sunlight, direct or through a window; the rest used stand-ins or could not be classed.
- Under a xenon lamp an experimental sunscreen measured SPF 3.6 on ten volunteers, and in the sun on the same ten it measured 1.6 to 2.4. 1
- On 31 pale volunteers in Sydney a solar simulator reddened skin at 68.7 millijoules of burning ultraviolet, or UVB, per square centimeter, and the sun at 93.6. 3
- When seven contract laboratories tested the same sunscreens by the international standard method, results for some products labeled SPF 50+ ran from 62.4 down to 5.5. 6
- Of 83 dermatology trials published in 2019, 47 were paid for entirely by industry, and 46 of those had authors with financial conflicts of interest. 7
Early light cures used arc lamps, and sunlight was tested against them
Niels Finsen, a doctor in Copenhagen, began to study light because of where he lived. "I suffered from anaemia and tiredness, and since I lived in a house facing the north, I began to believe that I might be helped if I received more sun," he wrote, and he spent as much time in its rays as he could. The Finsen Institute of Phototherapy opened in Copenhagen in 1896, and in 1903 his treatment of lupus vulgaris, a tuberculosis of the skin, with concentrated light won him a Nobel Prize. The presentation speech described the method: "Sunlight, or more frequently the light from a powerful electric-arc lamp" was focused through lenses into a beam "from which the heat rays have been as far as possible eliminated." 8; 9
In the summer of 1922 Benjamin Kramer and Frank Boone of Johns Hopkins University in Baltimore put seven Black children with active rickets, aged 6 to 16 months, in direct sunlight, for fifteen minutes a day at first and then an hour. Their paper listed "radiation with mercury vapor quartz lamp, carbon arc lamp, cadmium open spark" among the measures already known to heal rickets and set out to learn whether direct sunlight could raise blood calcium and phosphorus "as occurs with the agents above named." Every child showed healing on X-ray during the third week, and the authors wrote that the changes in the blood were "identical with those which follow the administration of cod liver oil or radiation with the mercury vapor quartz lamp." Theirs is the oldest study in our evidence map. 10
In August 1932 an international congress on light, meeting in Copenhagen, split ultraviolet into three bands: UVA from 400 to 315 nanometers, UVB from 315 to 280, and UVC below that. A nanometer is a millionth of a millimeter. Brian Diffey, a medical physicist at Newcastle General Hospital, wrote in 2002 that the bands "are arbitrary and differ somewhat depending on the discipline involved." Dermatologists usually put the border between UVA and UVB at 320 nanometers, a choice Diffey called "perhaps more arbitrary." 11
A solar simulator copies the ultraviolet in sunlight and leaves out most of its visible light and heat
The words "solar simulation" were in use at the Jet Propulsion Laboratory in Pasadena by June 1962, where carbon arc lamps lit a vacuum chamber six feet across that tested the Ranger spacecraft before launch. 12 Solar Light Company of Pennsylvania, which now sells equipment for sunscreen testing, says it "invented the world's first Solar Simulator in 1967." 13 The lamp inside most simulators is a xenon arc, an electric arc burning steadily through xenon gas. Osram, by its own account, first sold its XBO xenon lamp in 1954 for cinema projectors, where it replaced the carbon arc. 14
In 1978 xenon solar simulators ran "from 150 to more than 6,000 watts," according to an advisory panel of the FDA, and a 150-watt unit produced "10 or 12 solar constants," ten to twelve times the energy of sunlight at the ground. "If the MED for a normal subject is 20 minutes of sunlight exposure, then the solar simulator would produce an MED of 2 minutes," the panel wrote. The MED, or minimal erythema dose, is the smallest dose that leaves a person's skin just visibly red a day later, erythema being the medical word for redness. In the same passage the panel listed the lamp's drawbacks: "The full solar spectrum output is low in the visible and infrared wavelengths." 15
Diffey gave the working reason for lamps in 2002: "For many experimental studies in photobiology it is simply not practicable to use natural sunlight." Photobiology is the study of what light does to living things. Ultraviolet makes up about 5 percent of the sunlight that reaches the ground, and a laboratory simulator is built to copy that 5 percent: a xenon arc behind a glass filter that cuts the shortest wavelengths, with more filters or mirrors to remove most of the visible and infrared light, lighting a patch usually smaller than 15 by 15 centimeters. "No such source will match exactly the spectral power distribution of sunlight," he wrote. 11
The 1999 American sunscreen rule allowed a simulator to put "not more than 5 percent of its total energy output" at wavelengths longer than 400 nanometers, where visible light begins. 16 In 2010 Sayre and a colleague in Memphis compared several simulators that met the proposed American standard with standard spectra of sunlight, including the sun directly overhead, and wrote that "none of the solar simulators we examined would be expected to produce the same SPF as sunlight." 17 The FDA's 2011 rule states that "solar simulators cannot perfectly replicate natural sunlight." 4
US sunscreen testing had a sunlight method in 1978 and has named only a lamp since 1993
On 25 August 1978 the FDA published its advisory panel's plan for testing sunscreens, with two light sources. The xenon simulator was "the preferred artificial light source," and any other lamp needed "corroborating natural sunlight testing for acceptance." The sunlight method had its own section: "The test situation more closely approximates the actual ways the sunscreen product will be used by the consumer. The test subject is exposed simultaneously to the full solar spectrum, the heat, and the humidity." 15
The panel also listed the weather, the moving clouds, the changing angle of the sun, and sweat, and warned that when exposures were timed by the clock, "the results of 1 day's testing probably cannot be duplicated on another day." It recommended a meter that counted sunburning ultraviolet instead and took a count of about 400 as one MED on "typical" white skin. Readings of the MED by different investigators "vary ±20 percent," the panel noted. "No instrument has proven so reliable and consistent as the human eye," it added. 15
Volunteers had to be skin types I, II, or III, the three palest of six, and their type was judged by how their skin reacted to the "first 30 to 45 minutes sun exposure after a winter season of no sun exposure." The FDA's 2011 rule kept the same test of skin type. 15; 4
In 1993 the FDA proposed to drop the sunlight method. One comment it answered reported testing the panel's reference lotion, 8 percent homosalate, on nearly 500 volunteers, with an SPF of "4.0 ± 0.5" under a xenon simulator and "2.5 to 3.0" outdoors. Another said lamps had given higher values than natural sunlight and argued that both indoor and outdoor testing should be required, "adding that the cost of such testing is not an excuse when the principal issue concerns the health of people." Another wrote that an SPF that "might range from 8X to 18X in the laboratory environment" narrowed "from 4X to 6X with some formulations" under natural sunlight. 2
A reply comment answered with the 1978 Schering-Plough paper and an argument: "solar simulator testing provides a means of standardization, a means of assuring reproducible results and ultimately, and most importantly, a means of insuring uniform product labeling." The agency wrote that it "believes solar simulator testing is a more accurate method of determining SPF values for product labeling than outdoor testing," and added that "there is not enough sunlight in a day to conduct outdoor testing of sunscreens with high SPF values." The 1999 final rule named only the lamp. The FDA stayed that rule in 2001 before it took effect, and the rule that replaced its test in 2011 also named only the lamp and cut the volunteers per test from 20 to 10. 2; 16; 4
In a study paid for by the Barcelona company ISDIN and published in 2020, seven contract laboratories tested eight sunscreens marketed as SPF 50 or 50+ under lamps by the international standard method, and for some products the values "differed from a maximum of 62.4 to a minimum of 5.5." 6 In a 2025 ring study, which sends the same coded samples to several laboratories, 32 products went to 12 laboratories for the standard lamp test. By the study's own figures, a product that measured SPF 30 in one laboratory could measure anywhere from 18.1 to 49.5 in another, and the two results would count as the same. 18 The sunscreen page tells how the SPF number began outdoors in the Alps.
In 2021, in Arequipa, Peru, 17 people spent 1 hour 59 minutes under a clear sky with patches of ten sunscreens labeled SPF 30 to 110 and six sun-protective fabrics on their skin. Some reddening or darkening showed at 97 percent of the sunscreen-protected scores, and the two products made only with sunscreen chemicals the FDA allows, labeled SPF 30 and 45+, reached "at a maximum, SPF of 5-7 in natural sunlight." The fabrics "provided excellent sun protection." The first author, Shaun Hughes, founded Sun Precautions, a Seattle company that sells sun-protective clothing. 19; 20
In 1992 the cancer agency judged sunlight a proven cause of cancer and each band of ultraviolet only a probable one
In 1992 a working group of the International Agency for Research on Cancer, the cancer agency of the World Health Organization, found "sufficient evidence in humans for the carcinogenicity of solar radiation" and put sunlight in Group 1, its list of proven human carcinogens. For sunlamps and sunbeds it called the human evidence "limited" and put them in Group 2A, probable. Each band of ultraviolet on its own went in Group 2A as well. "Ultraviolet A radiation is probably carcinogenic to humans," the evaluation reads, and it says the same of UVB and UVC. 21
For each band the evaluation lists "sufficient evidence" in experimental animals and no separate finding in people. The people in the human studies had been exposed to sunlight, which at the ground carries UVA and UVB together with visible light and heat, and almost no UVC. Those human studies estimated sunlight from where people lived and what they remembered, and the melanoma page follows them one by one.
From the 1980s the studies in our map moved under lamps
Before 1980 our evidence map holds 15 studies, and 9 of them used real sunlight. The 1980s added 66, 36 of them under lamps, and the 1990s 117, with 95 under lamps. In the 2000s 225 of 257 studies used a lamp, 88 percent, and 19 used the sun. Since the 1980s the number of real-sun studies has stayed between 13 and 31 a decade, while lamp studies rose from 36 to 225 and fell back to 143 in the 2010s. The 2020s so far hold 123 studies, 86 under lamps and 27 in the sun.
A program that tags the studies in our map from their abstracts and full texts marks 62 of them as randomized trials, the design in which chance decides who gets what. Forty-five of those 62 used a lamp, and 15 used sunlight, direct or through glass. Nobody has yet checked those tags by hand, and 412 of the 784 studies carry no usable design tag at all.
Outdoor testing, the Schering-Plough team wrote in 1978, "requires simultaneous scheduling of a large number of volunteers and support personnel who are willing to endure the summer heat for a relatively long period of time," and testing across the range of outdoor conditions "would be prohibitively expensive and time-consuming." 1
In 1983 Hersey's group at Sydney Hospital studied sunlight and the immune system by copying a course at a commercial solarium, twelve sessions of half an hour. Fifteen volunteers sunbathed in the spring in minimal clothing, for half an hour on each side at midday, on 12 days over two weeks, while 13 matched volunteers stayed out of the sun. When the weather was bad they were asked to make up the lost time on later days, and for 5 of the 15 the weather never allowed the full 12 hours. 22
On the same skin, a lamp reddened people at lower doses than the sun
In March 2001 a team at Royal Prince Alfred Hospital in Sydney compared burn thresholds under a lamp and under the sun on 31 pale volunteers, skin types I and II. Each person's lower back took a row of rising doses from a 1,000-watt xenon simulator and, "just after midday on the roof of the Department of Dermatology," another row from the sun, with both lights measured on the same spectroradiometer, an instrument that splits light by wavelength and measures each part. The lamp reddened skin at an average of 68.7 millijoules of UVB per square centimeter and the sun at 93.6, about a third more. Counted in standard erythema doses, a unit weighted by how strongly each wavelength reddens skin, the lamp took 3.4 and the sun 11.7. 3
The lamp's figure was lower on every test day. On the one windy day, 59.9 against 63.2, the difference was too small to be sure of, and the authors suggested that "infrared heat, wind and sweating may play a role." They judged simulators "useful for performing tests where ratios are involved, such as for sunscreen SPF testing," and cautioned that absolute doses such as the MED differ from those in sunlight. Two of the seven authors worked at the Shiseido Basic Research Center in Yokohama, and the paper names no funder. 3
In Seoul, in a study published in 2008, 16 volunteers lay with their buttocks bare to the summer sun from 11 in the morning to 2 in the afternoon. Some patches sat under a filter that blocked ultraviolet, the part of sunlight a simulator keeps, and others under black cloth that let through only heat. "Only the UV region recruited neutrophils into the dermis, while UV, IR/visible light and heat contributed to macrophage infiltration," the authors wrote. Neutrophils are the white blood cells that arrive first at an injury, and macrophages are larger immune cells that follow. Seoul National University Hospital, a Korean science foundation, and a research agreement with the Amore-Pacific Corporation paid for the study. 23
In Rotterdam in 2016, hair from nine volunteers lost a median 54 percent of its cortisol, a stress hormone that researchers measure in hair, after 40 hours in natural sunlight. Hair from the same people lost 75 percent after 137.55 hours at 3.5 centimeters from a 26-watt UVB spiral bulb made by a company called Reptile Technologies. 24 In Dresden the same year, two months of summer sun raised two other hormones in hair samples, dehydroepiandrosterone, or DHEA, and progesterone, and a short lamp exposure did not. 25 Both teams wrote that sunlight should be counted as a confounder in studies of hair hormones, meaning a third factor that can distort what such a study finds.
Small changes in a lamp moved results in cells threefold or more
Randall Gill and Thomas Coohill at Western Kentucky University compared their simulator with August noon sunlight in a paper of 1987 and found the two matched in intensity at only one wavelength, about 300 nanometers. Between 310 and 345 they differed by as much as a factor of seven. Six strains of cells, three of them from people with xeroderma pigmentosum, an inherited disease in which skin cannot repair ultraviolet damage, varied sixfold in sensitivity under a germicidal lamp, which gives off UVC at 254 nanometers, and twofold under the simulator. "At best, the results reported here can only be interpreted as representative of this artificial source," the authors wrote. 26
A Philadelphia team reported in 2000 that light from 295 to 320 nanometers makes up about 4 percent of solar ultraviolet at the ground and 47 percent of the ultraviolet from a common fluorescent sunlamp filtered with Kodacel plastic. They titled the paper "Common fluorescent sunlamps are an inappropriate substitute for sunlight." 27 Unfiltered, such lamps also give off UVC, which the ozone layer stops before it reaches the ground. A 1993 analysis by researchers at Schering-Plough calculated that UVC made up 3.2 percent of an unfiltered sunlamp's power and 11.1 to 16.7 percent of its power to redden human skin. 28
In 2001 and 2003 physicists in Milan fitted one simulator with two filters, one passing light longer than 284 nanometers and the other longer than 293, which left the UVA the same and changed only the shortest UVB. Their cells were a human hybrid line, made by fusing a cervical cancer cell with a skin cell. Light through the 284 filter killed them three times as effectively and caused five times as much chromosome damage, and at doses that killed equal numbers of cells it transformed at least seven times as many into cancer-like cells in the dish. Ozone absorbs nearly all sunlight below about 286 nanometers, Gill had noted, so most of the difference between the two filters lies in light that barely reaches the ground. 29; 30; 26
In Dublin in 2010 a pulsed simulator that delivered, by the team's estimate, about 1,600 watts of ultraviolet per square meter in each flash killed fewer skin cells than a steady simulator at about 55 watts and a UVB lamp at 6.4. The authors traced the difference to the spectrum, mainly in the UVB, and wrote that without careful measurement of each lamp it "may have resulted in erroneous conclusions." 31
In Brighton in the summer of 1994, a Medical Research Council team set human blood cells outdoors in droplets on ice between noon and 4 in the afternoon and compared their DNA damage with that from a sunlamp. The sun left a much higher share of Dewar isomers, a kind of damage that longer ultraviolet makes out of another kind, the (6-4) photoproduct. 32
The standard dose of sunburning light rests on a fitted curve, and the curve for skin cancer on mice
Studies in this field often report dose in minimal erythema doses, each volunteer's own burn threshold under a particular lamp. Diffey called that use of the MED "absurd because the MED is not a standard measure of anything but, on the contrary, encompasses the variable nature of individual sensitivity to UV radiation." The international standard for the unit lists what changes an MED: the lamp, the size of the patch, the person's pigment, where on the body it is measured, how long after exposure it is read, and the light in the room where it is read. 11; 33
The standard's alternative is the standard erythema dose, or SED, a fixed 100 joules per square meter of ultraviolet weighted by how strongly each wavelength reddens skin. A clear summer day in Europe delivers about 30 to 40 SED, and 4 SED will moderately redden white skin that has not seen the sun for a while. 11 The weighting is an action spectrum, a curve of how strongly each wavelength produces an effect, and the reference curve was published by McKinlay and Diffey in 1987. The World Health Organization's 1994 review of ultraviolet describes it as based "on a statistical analysis of the results of minimum erythema dose studies carried out over the past 20 years or so" and drawn as "three straight lines when plotted on a semi-logarithmic scale," adding that "individual action spectra would not have the two inflection points." 34
In 1993 de Gruijl and colleagues built an action spectrum for skin cancer from about 1,100 albino hairless mice kept under 14 different broadband ultraviolet lamps in Utrecht and Philadelphia. "Direct information on the lambda dependency can only be obtained from animal experiments," they wrote, lambda meaning wavelength, and until then "the wavelength dependencies of other detrimental UV effects, such as sunburn, have been used as substitutes." Their curve peaked at 293 nanometers and carried "a large margin of uncertainty above 340 nm." 35 The international standard curve for skin cancers other than melanoma, first issued in 2006, is "derived from experimental laboratory data and modified to estimate the non-melanoma tumor response in human skin." 36
Few studies have measured the sunlight that reaches a person
"While evaluating the weathering characteristics of the plastics polysulphone and polyphenylene oxide (PPO) we found that they both darkened when exposed to ultraviolet radiation," Davis, Deane, and Diffey wrote in Nature in May 1976, and they proposed polysulphone as a badge to measure the ultraviolet reaching skin. 37 The film, 40 to 50 micrometers thick, sits in a cardboard or plastic holder, and its absorbance at 330 nanometers rises with the dose. 11 A Copenhagen group later showed, on a beach and on summer holidays, that a spore film on the wrist records about half the dose that reaches the top of the head, and in 2005 put a sensor weighted like the erythema curve into a 14-gram wristwatch that stores 32,000 time-stamped readings. 38; 39
On a 2007 expedition to the north side of Everest, 15 climbers and 10 Sherpas wore spore dosimeters, devices that add up the dose they receive, on the crown of the head. Fourteen dosimeters were lost on the mountain, temperatures above 6,500 meters fell below minus 20 degrees Celsius, which may have made the spores undercount, and skin measurements stopped at base camp "because none of the expedition laptops worked at elevations above" it. The climbers' median dose over the season was 93.6 MED and the Sherpas' 102.5, which the team set beside the roughly 100 MED a northern European office worker gets in a year. The Ontario Centres of Excellence paid for the study. 40
In Karachi in 2012, researchers pinned polysulphone badges to the clothes of 54 adults for four days and had them read at the University of Southern Queensland in Australia. They were building a questionnaire because dosimeters "are prohibitively expensive, therefore cannot be used in large epidemiological studies." Time outdoors on the questionnaire correlated with the badges at 0.60, on a scale where 1 is perfect agreement and 0 is none, and the questionnaire scores correlated with blood vitamin D at 0.36 to 0.48. 41 In a Perth trial of vitamin D drops for babies, the researchers gave dosimeters "to the first 3–4 infants enrolled per month over the course of two years," 86 of 195, "due to funding constraints." 42; 43
In the Canary Islands holiday study, 25 people from Denmark wore wrist meters for a week while a second meter sat on the hotel roof. The daily dose on the wrists tracked the roof with an R-squared of 0.33, meaning the roof reading accounted for about a third of the day-to-day and person-to-person spread in what the wrists recorded. The European Union paid for the study, and the sunburn page tells what happened to the volunteers' skin. 44 Our automatic tags mark 3 of the 784 studies in our map as using a personal dosimeter, and at least five of the studies cited on this page used one.
The largest studies of sunlight measured a stand-in for it
In a US Army study published in 2013, researchers assigned each of 421,461 male and 90,141 female recruits a sunlight exposure from the average annual UV index at their "home of record." They expected recruits from low-UV regions, thought to have less vitamin D, to break more bones in basic training. The recruits from low-UV regions had slightly lower odds of a stress fracture, 0.92 for men and 0.89 for women, where 1 would mean no difference. The true values could lie anywhere from about 3 to 13 percent lower for men and 5 to 16 percent lower for women. 45
Environment Canada introduced the UV index in 1992, and the World Meteorological Organization and the World Health Organization adopted it as a standard in 1994. It is the strength of sunlight on a flat surface, in watts per square meter weighted by the erythema curve, multiplied by 40. 46; 47
The World Health Organization's guide to the index says "REPORTING BURN TIMES IS NOT RECOMMENDED," because "people tend to interpret burn times to mean that there is a safe level of unprotected sun exposure." 47 A 2025 review of 39 studies found that about 48 percent of the people surveyed had heard of the index, and in the studies that asked for the day's value, at most 9 percent answered correctly. 48
A Madrid study of 704 heart patients, published in 2017, counted blood drawn from April to September as high sun exposure and blood drawn from October to March as low. 49 A Milan hospital sorted 415 unipolar and 187 bipolar depressed inpatients by whether their rooms faced east or west. On a bright May morning at 9 o'clock an east room measured 15,500 lux, a unit of how bright light is to the eye, and a west room 1,400, and "no exact recording of time spent in rooms is available." Bipolar patients in east rooms left a mean 3.67 days sooner, in what the authors called "a naturalistic retrospective observation." 50
More than 300,000 people in the UK Biobank reported their time outdoors on questionnaires, and the authors of a 2024 preprint, not yet peer reviewed, named that self-report, which "might involve recall bias," as the first limitation of their analysis. 51 A Spanish case-control study published in 2018 measured sun exposure two ways. It compared 2,140 people with colorectal cancer with 3,950 without and found lower risk in those who recalled frequent sun the previous summer, and no association with a score built from every place they had lived and the solar radiation at each. 52 The authors listed as a limitation that "there is no direct measurement of sun exposure." Case-control studies of this kind, which compare the recollections of people with a disease against people without it, are also the base of the sunburn and melanoma findings on the melanoma page.
Companies that sell sunscreen, cosmetics, and food paid for many of the lamp studies we read
Our records of these studies have no field for who paid for them. They do include a field meant for study design, copied from the source databases, and the most common entry in it, on 254 of the 784 sunlight studies, is "Research Support, Non-U.S. Gov't," a label PubMed uses for a kind of funding. The field holds one label per study, so it neither counts funders nor describes designs.
Procter & Gamble funded a 1994 simulator test of an iron-binding skin cream "entirely" and paid for a 2002 Boston study of sunscreen and DNA damage under a simulator. 53; 54 Mars supplied the cocoa for a 2006 sunburn trial under a simulator, Nestec the green tea extract for a 2015 one, and Croda the omega-3 capsules for a 2013 immune study under a simulator. 55; 56; 57 L'Oréal's own researchers ran a 2008 study titled "The benefit of daily photoprotection," in which the daily light was one minimal erythema dose from a xenon simulator, five days a week for six weeks, on "the untanned upper buttock region." 58 In a 2016 New York University study of skin cells, Estée Lauder "provided the solar simulated UV radiation instrument." 5
In the 2010 London lamp study of SPF at different latitudes, three of the five authors worked for BASF Schweiz, CIBA formulated both test sunscreens and supported the project, and the paper reads "The authors state no conflict of interest." 59 In a 2006 Sydney study, one author worked for Novogen, which synthesized the compound in the skin cream being tested, and the paper states "no conflict of interest." That study's "solar-simulated" light came from six fluorescent black-light tubes and one UVB tube. 60 The lead author of the Nambour trial, the one randomized trial of daily sunscreen in real sun with a melanoma count, disclosed research funding from L'Oréal Recherche. 61
Johnson & Johnson fully funded a 2018 trial in the sun of Vail, Colorado, that put its SPF 100+ and SPF 50+ products on the two halves of 199 faces, and four of the authors were its employees. 62 Sayre's 1978 comparison and his group's 1990 measurement of 26 simulators against the noon sun of Albuquerque both came from Schering-Plough's laboratory in Memphis, as did the 1993 analysis of UVC in sunlamps. 1; 63; 28 The Sydney rooftop study had two Shiseido scientists among its authors, and the seven-laboratory SPF study was paid for by ISDIN, which employed two of its authors. 3; 6
The UK Ministry of Defence paid for a trial that gave army recruits simulated sunlight from fluorescent tubes in whole-body cabinets, and the European Union paid for the Canary Islands holiday study. 64; 44 A 2014 Finnish trial of daylight photodynamic therapy, a treatment in which daylight activates a cream on sun-damaged skin, put "nonsponsored" in its title and was supported by the Novo Nordisk Foundation. An Australian trial of the same treatment that year was "funded by Galderma R&D," two of whose employees were authors. 65; 66 At least eight papers in our library carry a notice that because the authors paid page charges, the article "must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734." 32
We found no study that measures what share of sunlight or sunscreen research companies pay for. What has been measured is whether industry money goes with the result. A 2017 Cochrane review, one of a series that weighs medical evidence by fixed rules, pooled 75 papers on drug and device studies. Those paid for by the maker reported favorable results 1.27 times as often and favorable conclusions 1.34 times as often, with no difference in most of the usual checks for bias. Its authors wrote of "an industry bias that cannot be explained by standard 'Risk of bias' assessments." 67
In dermatology, a 2005 review of 179 trials in four leading journals found that the 43 percent with at least one author with a conflict of interest were more likely to report a positive result, and were also of higher methodological quality and larger. 68 Of 83 dermatology trials published in 2019, 47 were paid for entirely by industry, and 46 of those had authors with financial conflicts of interest. 7 Of 691 articles on indoor tanning, 39 of the 50 with financial links to the tanning industry favored tanning, against 27 of 620 without such links. 69
In the United States the Open Payments database lists company payments to doctors, but its rules cover makers of prescription drugs and of devices the FDA must clear, and a company that sells only nonprescription products such as sunscreen falls outside it unless it shares an owner with one that is covered. 70
Our map kept 784 of 157,842 records on the agreement of two AI models
The 784 studies in our evidence map came from a search we ran in August 2026. Eleven sources, among them Europe PMC, Semantic Scholar, OpenAlex, Crossref, PubMed, and ClinicalTrials.gov, returned 212,219 records, and removing duplicates left 157,842. Screening set aside 34,028 records outside biology and medicine, 13,437 about other species with no bearing on people, 6,801 that reported no original research, 2,617 with no real exposure to light, 716 about sun knowledge or habits alone, and 397 with no bodily outcome. Another 18,436 were reviews, kept only for the studies in their reference lists.
The final rule was written as instructions for two AI models. A study was eligible if it was original research that "actually administers or measures natural sunlight/daylight/solar exposure," or studied a lamp or a stand-in explicitly as a model of sunlight, and reported a bodily outcome in people, in human tissue or cells, or in an animal or cell model "explicitly inferring human effects." A lamp study of a drug or a cosmetic did not qualify "merely because that radiation can occur in sunlight." GPT-5.6 Luna, from OpenAI, and Claude Opus 5, from Anthropic, read the titles, abstracts, and index terms of 5,331 candidates separately, and a study went in only when both called it eligible and both named the same kind of light.
Each of the 784 carries the exposure label the two models agreed on, and nobody has checked all of them by hand. The 1978 Schering-Plough comparison, which used both a lamp and the sun, is labeled unclear. The Rotterdam hair study is labeled a direct-sunlight study of humans, and what sat in the sun was cut hair. 1; 24
The rule also kept out most research on what sunscreen chemicals do in the body. When we searched for that literature separately, 6 of the 1,539 studies we found were already in our library, and we added 1,335 of them as a second collection with no exposure label.
The two models split on 4,544 of the 5,331. Those records, with others the screens could not settle, make up 80,626 "unresolved candidates," more than half of all the records found. In searching, the share of relevant studies a search manages to find is called its recall, and the map's note gives the reason they were kept: "retained unresolved to protect recall."
Misconceptions
What people believe
- A solar simulator gives the same light as the sun, so what it does to skin is what the sun does.
- The SPF on a sunscreen was measured in sunlight.
- A minimal erythema dose is a fixed amount of sunlight.
- Large studies of sunlight measured how much sun people got.
- A study paid for by a company must have been badly designed.
What the studies show
- Simulators copy the ultraviolet of sunlight and remove most of its visible and infrared light. On the same Sydney volunteers a simulator reddened skin at 68.7 millijoules of UVB per square centimeter and the sun at 93.6. 15; 3
- The FDA's advisers wrote a sunlight test for sunscreens in 1978, and the agency dropped it in 1993. 15; 2
- An MED is one person's burn threshold under one lamp, read by one observer. Diffey called it "not a standard measure of anything." 11
- The largest studies used the UV index at a home address, the month of a blood test, or a questionnaire. 45; 49; 51
- In a Cochrane review, drug and device studies paid for by the maker did no worse on most checks for bias and still reached favorable conclusions more often. 67
Questions people ask
Is a UV lamp the same as sunlight?
No lamp in these studies matched sunlight. Solar simulators copy the UVB closely, give less of the longer UVA, and filter out most visible and infrared light, and on the same volunteers in Sydney a simulator reddened skin at a lower UVB dose than the sun. 15; 3
What does the UV index measure?
The UV index is the strength of sunlight on a flat surface at a place and time, weighted by how strongly each wavelength reddens skin and multiplied by 40. It does not measure what reaches a person. On a Canary Islands holiday, the doses on people's wrists tracked the meter on the hotel roof with an R-squared of 0.33. 47; 44
Is SPF tested in real sunlight?
Not for the number on an American label. The FDA's advisers included a sunlight test in 1978, and the agency dropped it in 1993, writing that "there is not enough sunlight in a day to conduct outdoor testing of sunscreens with high SPF values." 15; 2
What is a solar simulator?
A solar simulator is most often a xenon arc lamp with filters that remove the short ultraviolet the ozone layer stops and most of the visible and infrared light. In the FDA panel's 1978 example, a 150-watt unit gave 10 to 12 times the energy of sunlight and reddened skin in 2 minutes where the sun would take 20. 15; 11
How do scientists measure how much sun a person gets?
The largest studies on this page did not measure it. They used latitude, the UV index at a home address, the season, or questionnaires. A few studies put polysulphone badges, spore films, or electronic wrist meters on people, and the Karachi team called dosimeters "prohibitively expensive" for large studies. 45; 41
Who pays for sunscreen and sunlight research?
We found no count of it. In the papers we read, companies including Procter & Gamble, L'Oréal, Estée Lauder, and Johnson & Johnson paid for or supplied many lamp and sunscreen studies, and governments and foundations paid for others. 54; 58; 5; 62
How did The Sunlight Report choose its studies?
A search returned 157,842 records after duplicates were removed. Two AI models read the candidates separately, and 784 went into our evidence map because both judged them eligible and named the same kind of light. The 80,626 records that could not be settled were kept apart as unresolved.
What we do not know
We found no study that followed the same long-term outcome, such as skin cancer or aging, under a lamp and under the sun in the same people. The comparisons on this page measured redness, cell death, DNA damage, or hormones in hair over periods from minutes to two months.
Most simulators remove the visible and infrared parts of sunlight, so lamp studies of skin say little about them. The Seoul study found heat and infrared active in skin after one afternoon. 23
Who pays for most sunlight research is unknown. Our records have no funding field, and the funding lines on this page were read by hand from a small share of the papers.
What the 80,626 unresolved records hold is unknown. Some are likely eligible studies that the two models read differently.
Of the 98 studies in our map done under direct sunlight, 19 are missing from our library, against 31 of the 405 done under solar simulators.
Key studies
- Kramer and Boone, 1922, Baltimore. Real sun. Seven children with rickets healed, with blood changes that matched those after cod liver oil or a quartz lamp. 10
- Davis, Deane, and Diffey, 1976. A plastic that darkens in ultraviolet, proposed as a badge to measure the dose reaching skin. 37
- Sayre and colleagues, 1978, Memphis. Lamp and real sun on the same ten people. A sunscreen measured SPF 3.6 under the lamp and 1.6 to 2.4 in the sun. 1
- FDA advisory panel, 1978. A sunlight method for testing sunscreens beside the lamp method. 15
- Hersey and colleagues, 1983, Sydney. Real sun, timed to copy a course at a solarium. 22
- Gill and Coohill, 1987, Kentucky. Lamp and cells. Simulator and sunlight matched in intensity at one wavelength. 26
- Sayre and colleagues, 1990, Memphis. Twenty-six simulators measured against summer noon sun in Albuquerque. 63
- International Agency for Research on Cancer, 1992. Sunlight in Group 1, each band of ultraviolet in Group 2A. 21
- FDA, 1993. The sunlight method dropped from the proposed sunscreen rule. 2
- De Gruijl and colleagues, 1993, Utrecht and Philadelphia. Lamps and mice. The action spectrum for skin cancer from about 1,100 hairless mice. 35
- Learn and colleagues, 1993, Memphis. UVC from unfiltered sunlamps did 11 to 17 percent of their reddening. 28
- Clingen and colleagues, 1995, Brighton. Real sun and a sunlamp on blood cells. Different mixes of DNA damage. 32
- Brown and colleagues, 2000, Philadelphia. Short UVB made up 47 percent of a filtered sunlamp's ultraviolet and 4 percent of the sun's. 27
- Benedetti and colleagues, 2001, Milan. Window direction as a stand-in for morning sun. 50
- Bettega and colleagues, 2001 and 2003, Milan. Lamp and cells. A shift of 9 nanometers in a filter moved results threefold to sevenfold. 29; 30
- Diffey, 2002, Newcastle. How ultraviolet is produced, simulated, and measured. 11
- Poon and colleagues, 2003, Sydney. Lamp and real sun on the same 31 people. The sun needed more UVB to redden skin. 3
- Cho and colleagues, 2008, Seoul. Real sun with parts of the spectrum blocked. Heat and infrared drew immune cells into skin. 23
- Maguire and colleagues, 2010, Dublin. Lamps and cells. The spectrum and strength of the light, not its pulsing, decided cell death. 31
- Young and colleagues, 2010, London. Lamp. Regulatory test lamps for SPF carry more UVB for their UVA than the equatorial noon sun. 59
- Sayre and Dowdy, 2010, Memphis. No simulator examined would give the same SPF as sunlight. 17
- Humayun and colleagues, 2012, Karachi. Real sun. Badges against a questionnaire. 41
- Cheng and colleagues, 2013, Everest. Real sun. Spore dosimeters on climbers and Sherpas. 40
- Montain and colleagues, 2013, United States. The UV index at a home address as a stand-in for 511,602 recruits. 45
- Petersen and colleagues, 2013, Canary Islands. Real sun. Wrist meters against a meter on the roof. 44
- Wester and colleagues, 2016, Rotterdam, and Grass and colleagues, 2016, Dresden. Real sun and lamps on cut hair. 24; 25
- Lundh and colleagues, 2017. Industry-sponsored drug and device studies reached favorable conclusions more often. 67
- Rueter and colleagues, 2019 and 2021, Perth. Real sun. Dosimeters on 86 of 195 babies. 43; 42
- Trullàs and colleagues, 2020. Lamp. Seven laboratories, SPF values from 62.4 to 5.5. 6
- Hughes and colleagues, 2021, Arequipa. Real sun. Sunscreens labeled SPF 30 and 45+ reached SPF 5 to 7. 19
- Steele and colleagues, 2021. Conflicts of interest in 69 percent of 83 dermatology trials. 7
- Cole and colleagues, 2025. Thirty-two products in 12 laboratories. By the study's precision figures, a product measuring SPF 30 in one laboratory could measure 18.1 to 49.5 in another. 18
Definitions
- Action spectrum. A curve of how strongly each wavelength of light produces an effect, such as reddening or cancer.
- Case-control study. A study that compares people who have a disease with people who do not, and looks for differences in what the two groups recall or have on record.
- Cochrane review. A systematic review, a study that gathers and weighs all the studies on one question, done to the rules of the Cochrane network.
- Confounder. A third factor that can create or hide a link between the thing a study measures and the outcome.
- Correlation. A number from minus 1 to 1 for how closely two measures rise and fall together. Zero means no relation.
- Daylight photodynamic therapy. A treatment for sun-damaged skin in which a cream is spread on the skin and daylight activates it.
- Dewar isomer. A form of DNA damage that longer ultraviolet makes out of another form, the (6-4) photoproduct.
- Dosimeter. A badge, film, or meter worn or placed to add up the dose of light it receives.
- Erythema. The medical word for redness of the skin.
- Evidence map. Our list of the 784 studies of sunlight and the body that met the eligibility rule, each with an exposure label.
- Exposure label. The kind of light a study used, as our map records it: direct sunlight, filtered daylight, solar simulator, another lamp or device, or a stand-in.
- Germicidal lamp. A lamp that gives off ultraviolet at 254 nanometers, used to kill germs. Sunlight at the ground holds none of that wavelength.
- Lux. A unit of how bright light is to the human eye.
- Minimal erythema dose (MED). The smallest dose of ultraviolet that leaves one person's skin just visibly red a day later. It changes with the lamp, the skin, the body site, and the person reading it.
- Nanometer. A millionth of a millimeter, the unit for the wavelength of light.
- Odds ratio. How the odds of an outcome in one group compare with another. 1 means no difference.
- Photobiology. The study of what light does to living things.
- Polysulphone. A plastic that darkens in ultraviolet, used as a film badge to measure dose.
- Preprint. A paper posted publicly before peer review.
- Randomized trial. A study in which chance decides who gets which treatment, so that the groups differ only by chance in everything else.
- Recall. In searching, the share of relevant studies a search finds.
- R-squared. The share of the spread in one measure that another measure accounts for, from 0 to 1.
- Ring study. A test in which the same coded samples go to several laboratories to see how far their results differ.
- Solar constant. In the FDA panel's usage, the total energy of sunlight arriving on a square meter at the Earth's surface.
- Solar simulator. A lamp, most often a filtered xenon arc, built to copy the ultraviolet part of sunlight.
- Spectroradiometer. An instrument that splits light by wavelength and measures each part.
- SPF. Sun protection factor. The dose that reddens skin under a sunscreen divided by the dose that reddens bare skin, measured under a lamp.
- Standard erythema dose (SED). A fixed 100 joules per square meter of ultraviolet, weighted by how strongly each wavelength reddens skin.
- UV index. The strength of sunlight on a flat surface, weighted by how strongly each wavelength reddens skin and multiplied by 40.
- UVA, UVB, and UVC. The three bands of ultraviolet: UVA from 400 to 315 nanometers, UVB from 315 to 280, and UVC below 280, which the ozone layer stops. Dermatologists often put the border between UVA and UVB at 320.
- Xenon arc lamp. A lamp that makes light from an electric arc through xenon gas. Its light, filtered, is the basis of most solar simulators.
- Xeroderma pigmentosum. An inherited disease in which skin cannot repair ultraviolet damage to DNA.
Statistics
The 1983 Sydney team timed its sunbathing to match a course of 12 half-hour whole-body sessions in a commercial solarium. 22
A 1987 cell study's solar simulator delivered about 60 percent of the output of noon sunlight on a cloudless August day in Kentucky. 26
The same 1987 authors note that their solar simulator gave out less visible and infrared light than sunlight, so UV below 360 nm made up a larger share of its output than of sunlight's. 26
A 1991 cell study's solar simulator gave 0.4 percent UVB, 6.2 percent UVA, 57.9 percent visible, and 35.5 percent infrared, against 0.3, 5.1, 62.7, and 31.9 percent for sea-level sunlight. 71
A London team's solar simulator put 93 percent of its sunburn-weighted energy in the UVB band, against 81 percent for noon summer sunlight in the UK and 85 percent at 30°N. 72
In a 1998 immune-suppression study, the fluorescent solar-simulating lamps matched the sun's spectrum from 290 to 360 nm but gave out relatively less of the longer-wave UVA than sunlight. 73
In a Lausanne photodynamic-therapy study, mid-April afternoon sunlight measured 50 mW/cm2 in the 380 to 700 nm band, half the irradiance of the team's solar simulator in that band. 74
L'Oréal researchers ran the same sunscreen immune test indoors under a solar simulator in 75 recruits and outdoors in real sun in 32, of whom all 32 finished. 75
A Swedish study exposed six volunteers to both lamp UV and natural sunlight, with and without protection. 76
In 16 pale-skinned Sydney volunteers, the smallest dose that produced a visible tan a week later averaged 85.6 mJ/cm2 under the solar simulator and 118.3 mJ/cm2 under natural sunlight. 3
In 31 volunteers with pale skin (types I and II) in Sydney, the smallest dose that reddened the skin averaged 68.7 mJ/cm2 (3.4 standard erythema doses) under a xenon-arc solar simulator and 93.6 mJ/cm2 (11.7 standard erythema doses) under natural sunlight. 3
In a 2007 crude-oil toxicity study, sunlight on a clear June day measured about 8,000 foot-candles against 700 for the solar simulator used, about tenfold stronger. 77
In a London lamp study, UVA made up about three times more of the sunburn-weighted energy in a lamp built to mimic temperate daylight than in the standard European source used to test SPF. 59
UVB was 3.7 percent of the daylight-mimicking lamp's UV but 65.7 percent of its sunburn-weighted energy; in the standard SPF-test source, UVB was 8.8 percent of the UV and 87.5 percent of the sunburn-weighted energy. 59
A UVB-only sunscreen labeled SPF 6 tested at SPF 8.6 in people under the standard source, but its modeled SPF under the daylight-like spectrum was 3.9; the broad-spectrum sunscreen went from 7.0 to 6.7. 59
A Dublin cell study found a pulsed solar simulator delivering over 1600 W/m2 of UV per pulse was less toxic to cells than a second solar simulator at 55 W/m2 and a UVB fluorescent lamp at 6.4 W/m2, which the authors put down to differences in spectrum, mainly in the UVB. 31
Weighting their solar simulator's spectrum for sunburn, a team calculated that UVB caused 87 percent of the skin enzyme MMP-1 induction under the lamp, against 80 percent for noon summer sun in London. 78
A Chapel Hill fibroblast study's solar simulator emitted 94.3 percent UVA and 5.7 percent UVB. 79
In hair from nine volunteers, 40 hours of natural sunlight cut cortisol by a median 54 percent, against 75 percent for a high dose of lamp UV. 24
In a Chinese hemoporfin trial, 20 minutes of noon sunlight measured at 85 mW/cm2 delivered 102 J/cm2 to the skin. 80
In six adults, 20 to 50 minutes of sunlight at 7600 µW/cm2 raised the skin redness index by 40.22 to 47.12 units, while a 365 nm UV lamp raised it by at most 12.09. 81
A 1999 study found immune suppression in volunteers from daily lamp doses of UVB and UVA each rated as about 6 minutes of summer sun. 82
A Graz sunscreen study estimated that the lamp dose which halved volunteers' immune response, 0.63 to 0.79 of a minimal erythema dose, would come from 10 to 20 minutes of cloudless summer noon sun in central Europe. 83
A 2007 bacteria-and-skin-cell study rated its 30-minute xenon-mercury lamp dose of 18.6 J/cm2 as equal to 30 minutes of outdoor sunlight. 84
In Sydney nicotinamide studies, the lowest solar-simulator dose was rated as under 4 minutes of noon spring sunlight in Sydney and the highest as about 15 minutes of sunlight. 85
In an omega-3 trial, the solar simulator doses were rated as equal to about 8 minutes (lowest) and about 30 minutes (highest) of midday summer sun. 57
A Danish lamp study irradiated about 80 percent of participants' skin for nine weeks and took 220 blood samples. 86
The Danish lamp trial ran at 56°N from 1 October to December, a period its authors describe as having insignificant ambient UVB. 86
A visible-light sunscreen test gave four daily xenon-lamp doses, each rated equal to about one hour of midday sun. 87
A lamp study of vitamin D across skin types gave per-session doses of 2 standard erythema doses from Arimed B tubes and 1.5 from TL01 tubes, cumulative doses of 10 and 7.5. 88
The same skin-type study ran its two lamp arms in February to March and May to June, and relied on very poor weather in both periods to limit ambient UV, alongside unexposed controls. 88
In a 2026 British army trial, each week of simulated sunlight in the first phase was rated equal to about 15 minutes of unshaded midday summer sun six times a week in northern England at 53.5°N, about 12 minutes in Philadelphia and 18 minutes in Oslo. 89
In a Finnish heliotherapy course for eczema, the median personal UV dose over two weeks was 60 standard erythema doses for the January group and 109 for the March group. 90
Across four days of dosimeter wear in Karachi, the high-exposure group spent a mean 329 minutes in the sun, against 83.5 minutes for the moderate group and 69.5 minutes for the low group. 41
Of the dosimeters handed out on the Everest expedition, 72 percent were analysed; 14 were lost on the mountain. 40
In a New Zealand study, 201 volunteers wore electronic UV dosimeters during daylight for 10 weeks, then each got one full-body lamp dose of about 2 standard erythema doses from one of four lamp types. 91
In the Danish sun-holiday study, the daily dose on wrist dosimeters tracked the rooftop ambient dose with an R-squared of 0.33. 44
In the Australian daylight photodynamic therapy trial, patients judged the weather on a picture scale: 65 percent rated it sunny, 34 percent cloudy or partly cloudy, and 1 percent rainy. 66
In the Finnish daylight photodynamic therapy trial, patients sat in daylight for an average of 120 minutes (range 115 to 125) at an average 23 °C. 65
A Dutch daylight-and-sleep field study measured light with a wrist sensor that read up to 32,000 lux, sampling every 15 seconds for at most five days in a row. 92
In the Perth infant trial, dosimeters recorded a median 1018 J/m2 of UV over the first three months for winter-born babies and 1203 J/m2 for summer-born babies, a difference the authors found not significant. 43
In the Perth infants, dosimeter-measured UV did not correlate with blood 25-hydroxyvitamin D at 3 or 6 months of age. 43
In a Perth infant vitamin D trial, funding limits meant personal UV dosimeters went only to the first three or four infants enrolled each month over two years. 42
In the Perth infant trial, 3 infants did not wear their dosimeters, leaving 83 with usable readings and a median of 950 J/m2 of UV over the first three months of life. 42
The 2019 report from the same Perth trial counted 4 non-wearers among the 86 infants given dosimeters, leaving 82 with usable data. 43
In a Karachi validation study, time outdoors reported on a short-term questionnaire correlated with four days of UV dosimeter readings at 0.60. 41
In the Karachi study, the short-term sun questionnaire score correlated with blood vitamin D at 0.36, and the long-term score at 0.43 for summer and 0.48 for winter. 41
The Danish sun-holiday volunteers were asked to write diary entries every 30 minutes from 07:00 to 18:30. 44
A UK Biobank analysis of outdoor light and blood pressure enrolled over 300,000 people and took their time outdoors from questionnaires, not meters. 51
A hospital study took sunlight exposure from which way the room windows faced, east or west, for 415 unipolar and 187 bipolar depressed inpatients. 50
Bipolar inpatients in east-facing rooms, which got direct morning sun, stayed a mean 3.67 days less than those in west-facing rooms; the authors call it a naturalistic retrospective observation. 50
In the Health Professionals Follow-Up Study, researchers related six determinants to blood vitamin D in 1,095 men and used the model to assign a predicted level to each of 47,800 men in the cohort. 93
An Australian ecological study of lymphoid cancers diagnosed from 2002 to 2006 used three latitude bands, below 29°S, 29 to 36°S, and 37°S and beyond, as the stand-in for ambient UV exposure. 94
A US Army study of 421,461 male and 90,141 female recruits assigned each recruit's UV exposure from the average annual UV index at their home of record. 45
In the army cohort, recruits from low-UV home regions had slightly lower odds of stress fracture than those from high-UV regions, 0.92 for men and 0.89 for women, against the authors' hypothesis. 45
A Madrid study of 704 heart patients classed sun exposure only by the month their blood was drawn, April to September counting as high and October to March as low, and followed them for a mean 2.15 years. 49
A Spanish case-control study of 2,140 colorectal cancer patients and 3,950 controls took sun exposure from interviews and residential history. 52
A 1983 Sydney study put 15 volunteers in natural sunlight for 1 hour a day on 12 days over 2 weeks, against 13 matched controls. 22
In a 1991 lamp trial of sunscreen against cold sores, 38 patients each had four of their own minimal erythema doses of UV on the lip twice, once over sunscreen and once over placebo. 95
The 1994 cold-sore model was built on 40 subjects, 6 given 4 minimal erythema doses to the lip from a solar simulator and 34 given 3. 96
The Lausanne team tested skin photosensitivity with a solar simulator on the backs of 23 patients at times from 5 hours to 57 days after the drug. 74
The 2002 sunburn cream trial treated simulated-sunlight sunburn in 24 volunteers with skin type III. 97
A Seoul study gave the buttocks of 16 volunteers 1.1 to 3 minimal erythema doses of natural sunlight, covering some sites with a UV filter or black cloth to split the effects of UV, infrared and visible light, and heat. 23
The Karachi questionnaire study enrolled 54 adults picked by their reported sun exposure: 17 high, 18 moderate, and 19 low. 41
The Australian daylight photodynamic therapy trial enrolled 100 people for 24 weeks. 66
A Finnish split-face trial of daylight photodynamic therapy randomized lesions, not people: 177 actinic keratoses on 13 patients. 65
In a trial of melatonin cream in real sun, 23 volunteers had their backs exposed from 1:22 to 2:02 PM under a UV index of 9. 98
A Turkish nursing-home trial of morning sun for sleep had 61 residents aged 65 and over, 30 in the sun group and 31 controls. 99
The Dutch daylight-and-sleep field study used 14 adults, 9 women and 5 men, aged 65 to 80. 92
A 2020 psoriasis gel study enrolled 90 patients: 44 had clinic narrowband UVB and 46 used the gel at home before ordinary sunbathing. 100
In the Perth infant trial, 86 infants wore a UV dosimeter for their first three months and 109 did not. 42
A 2026 United Arab Emirates study piped sunlight indoors for 16 adults, four sessions a week for eight weeks, each 10 to 20 minutes. 101
A 1977 report treated nine people with solar urticaria by repeated exposure to natural sunlight or lamp light. 102
The Finnish authors named the limits of their daylight trial as 13 patients and a 3-month follow-up. 65
The Helsinki trial compared laser-assisted and plain daylight photodynamic therapy on the two sides of the head in 60 patients over a two-year follow-up. 103
A 1994 study that used a solar simulator to trigger cold sores was supported by Burroughs Wellcome Company. 96
A 1994 solar-simulator test of a topical iron chelator was funded entirely by Procter & Gamble. 53
A 1998 study of skin vitamin E loss under a solar simulator was partly supported by a grant from Unilever Research U.S.A. 104
In the 2001 skin-adaptation study of tanning-lamp courses, one author, De Winter, was supported by the lamp maker Philips Lighting. 105
The 2002 study of sunscreen and thymine dimers under a solar simulator was supported by a grant from Procter & Gamble. 54
A 2002 trial of steroid creams for lamp-induced sunburn was funded by Schering AG. 97
A 2003 study of nerve growth factor in lamp-irradiated human skin used tissue left over from a project funded by Procter & Gamble. 106
A 2003 Graz solar-simulator study of sunscreen immune protection was funded by a European Community grant, with the sunscreens provided by Beiersdorf AG. 83
Two authors of the Sydney sun-against-simulator study, Haruhi Iwaki and Masato Hatao, were listed at the Shiseido Basic Research Center in Yokohama. 3
A 2004 study of burn thresholds after repeated lamp exposures was supported by Unilever. 107
A 2006 cocoa trial that tested sunburn under a solar simulator used two cocoa preparations, both provided by Mars Inc. 55
A 2008 solar-simulator study of a vitamin C, E, and ferulic acid serum was supported by a grant from SkinCeuticals-L'Oréal, and one author, Dr Pinnell, consulted for the company. 108
The authors of a 2008 solar-simulator study of a rosemary extract were all employees of L'Oréal. 109
A 2009 study of tanning after UVA lamp exposure was supported by Unilever and a Danish hospital foundation. 110
Every author of a 2010 reconstructed-skin study under simulated sunlight worked for L'Oréal, which took part in study design, data collection and analysis, or writing. 111
A 2011 solar-simulator study of antioxidants added to a broad-spectrum sunscreen was funded by Clinique Laboratories. 112
A 2014 study of UVA1 lamp exposure in human skin received funding from La Roche-Posay Pharmaceuticals. 113
A 2014 Australian daylight photodynamic therapy trial was funded by Galderma R&D. 66
The Finnish daylight photodynamic therapy trial was supported by the Novo Nordisk Foundation and declared no conflicts of interest. 65
A green tea trial under a solar simulator randomized 50 people, 25 to catechins and 25 to placebo. 56
In a 2016 keratinocyte study, Estée Lauder partly funded the work and provided the solar simulator itself. 5
A 2017 study of keratinocytes under a solar simulator was partly supported by Estée Lauder, and one author was a full-time company employee. 114
A Dutch study of seasonal vitamin D in elite athletes was funded by DSM Food Specialties and the Dutch Olympic Committee. 115
The SPF 100+ against SPF 50+ trial in real sun was fully funded by Johnson & Johnson Consumer Inc., and four of its authors were company employees. 62
A 2018 Detroit lamp study of UVA1 and visible light declared no funding source, while five of its authors listed investigator, consultant, or advisory roles with companies including Estée Lauder, Johnson & Johnson, Allergan, Bayer, and Pfizer. 116
One author of the hemoporfin trial, which tested skin in outdoor sunlight, was an employee of Shanghai Fudan-Zhangjiang Bio-Pharmaceutical. 80
The hemoporfin trial that tested skin in outdoor sunlight was sponsored by Shanghai Fudan-Zhangjiang Bio-Pharmaceutical. 80
A 2020 solar-simulator study of skin fibres across ethnic groups was funded by a programme grant from Walgreens Boots Alliance. 117
A 2020 study of the skin's gene response to solar-simulated UV had its microarray processing paid for by La Roche-Posay. 118
A 2021 study of antioxidants on lab-grown skin under simulated sunlight was supported by a Pierre Fabre grant. 119
A 2026 Helsinki trial of daylight photodynamic therapy stated that no medical companies funded it, while two authors listed congress support or speaker fees from companies including Galderma. 103