532 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS Table VII Solar Simulator Vs. Westinghouse FS-20 Sunlamps Fluorescent FS-20 Sunscreen Sunlamp Tubes Solar Simulator (5/xl per cm 2) S.P.F. - S.E. S.P.F. - S.E. J 13.4 - 1.07 8.8 - 0.68 A 7.2 - 0.37 3.5 - 0.22 I 62.0 + 2.54 8.0 + 0•42 COMPARISON OF UV-LIGHT SOURCES In comparison to the solar simulator, the fluorescent sunlamps gave strikingly higher S.P.F. values for all the formulations (Table VII). The difference was most pronounced with I, where there was about an eightfold increase. DISCUSSION Recently, the Over-The-Counter (OTC) panel on topical analgesics has provided guidelines for the laboratory appraisal of sunscreens (4). These require the use of the solar simulator to determine the immediate S.P.F. of a specified UV-absorber at a dose of 2 mg]cm 2. This is intended to provide the consumer with an estimate of potency so that a choice can be made according to individual needs (skin type, geographic location, etc.) Valuable as this is, we regard the immediate S.P.F. as furnishing too limited in- formation. Millions of persons have occupations or life styles which require the regular use of sunscreens under very diverse circumstances and different environmental stresses. Our data show that proprietory formulations vary markedly in their ability to withstand sweating, water wash-off and abrasion. Moreover, since we recommend daily use for persons at risk, it is important to assess other features of safety such as extent of percutaneous absorption as well as buildup in the horny layer reservoir. Often a sunscreen performs well in one test and poorly in another. Each formulation has a "per- sonality" a complex of features which may make it either exceptional or ordinary for particular uses and users. A comprehensive evaluation cannot be made without assess- ing these various properties. To develop appropriate methodology, we had to consider a number of factors of varying importance. Since these have not been adequately dis- cussed in the literature, we shall briefly review our own thoughts and experiences. Early on, the question arose whether S.P.F. values were related in any way to skin type. Fair-skinned, blue-eyed celts are far more susceptible to sunburn than darkly pig- mented Mediterraneans. Would the S.P.F. for a given formulation be different in the two groups? We found that the mean S.P.F.'s were not different (unpublished observa- tions). The S.P.F. is a ratio and is not influenced by the susceptibility of skin to sunburn. It is advantageous, however, to use fair-complexioned subjects since their MED's are lower redness is more easily perceived and time is saved. Another question is how do laboratory S.P.F.'s compare with those obtained with sun- light? The solar simulator mimics sunlight mainly in the sunburning UV-B range but not in other regions of the spectrum. UV intensity, for example, falls sharply above 360 nm. Sayre et al. found that the S.P.F. obtained with the solar simulator was higher than with sunlight (3.6 vs. 2.4) (5). Outdoor testing was done, however, at a tempera-

EFFICACY OF SUNSCREENS 533 ture of 34øC and 85 to 90% relative humidity, conditions which favor sweating and dilution of the sunscreen. When laboratory testing was repeated after preheating the skin to 35øC, the S.P.F.'s were comparable. We have obtained excellent agreement for one well studied sunscreen (F). At a dose of 2/xl per cm 2, our laboratory S.P.F. was 4.1 ___ 0.57 in ten subjects. In sunlight with a larger panel the S.P.F.'s at the same dose were 3.6 --- 0.4 before and 4.1 + 1.96 after swimming (studies conducted by Paul Finkelstein, Ph.D., Johnson and Johnson Laboratories). This experience validates the solar simulator as a realistic instrument. Other pilot studies of our own show that the results secured with the solar simulator are applicable to the usage situation. We found a wide range of individual S.P.F.'s for any single preparation differences by as much as 50% being not uncommon. This variability is familiar to investigators who test sunscreens. Repeatability poses a real problem. Occasionally, tests conducted on the same person at different times will yield divergent results. It is well known that the MED may vary by as much as 50% in the same person from one day to another (6). The skin's surface does not resemble a glass slide. A variable amount of the test ma- terial is lost on the glass rod used to spread the test agents. This can have a large effect in view of the small amount applied. Even distribution of the test agents is difficult, especially with ointments and creams. We have long been aware that the variance is greater with ointments than with liquids. Delayed pigmentation may develop in dark complexioned persons, making it difficult to identify minimal erythema. This pig- mentation is due to UV-A (7), which is transmitted freely by UV-B absorbers such as PABA and its esters. As a check on undue variability it is advisable to include a standard formulation in every test. The OT.C panel recommends a 4% ethanolic solution of PABA or, more recently, an 8% solution of Homomenthyl salicylate. These are convenient for UV-B absorbers. At 2 /xl/cm 2, our S.P.F. for the 4% PABA standard is 4.6 + 0.27 (unpublished observations). We customarily include a standard UV-B proprietary sunscreen, generally I, and when appropriate a UV-A absorber, generally J. It was previously shown from this laboratory that hot quartz mercury lamps, which emit line spectra throughout the UV range, are woefully misleading for assessing sunscreens (2). The values were either falsely high or low depending on the absorption characteristics of the sunscreen. Our present findings show that this same handicap ap- plies to fluorescent suniamps. These emit a continuous spectrum that extends from about 280 to 360 nm, with about 55% of the energy below 320 nm. The inordinately elevated S.P.F. for I can be explained by the fact that the peak absorption of PABA parallels the maximal UV-B emission of the fluorescent sunlamp. We reemphasize that only the xenon solar-simulator provides realistic S.P.F.'s. Likewise, we would argue against the use of monochromatic radiation. Different wavelengths in the UV have dif- ferent biologic effects and these may be augmentative rather than additive (8). Sunlight is of course polychromatic. It is inappropriate, therefore, to use monochromatic radiation. It is clear that S.P.F.'s are significantly influenced by dose. The O.T.C. panel recom- mends a test dose of 2 mg/cm 2 (or 2/xl/cm • for liquids). This is entirely reasonable. We used 5/xl/cm • when we began our study and were constrained to continue with this dose for consistency. In several published studies, doses have ranged from 10 to 60 /•l/cm (2, 9-11). Such amounts are greatly in excess of normal usage and yield artificially elevated S.P.F.'s. Schlagel et al. estimated the quantity of ointment and