162 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS cosmetics? Because they are neither essential nor lifesaving, cosmetic products must be "trustworthy." No cosmetic manufacturer or distributor would knowingly distribute a preparation that is not safe. To do so would classify him as not trust- worthy. His liability insurance rates would skyrocket, the public would refuse to buy his products, and he could not remain in business profitably for very long. Thus, it is in the best interest of all who produce and sell cosmetic products to establish the safety and trustworthiness of their products in advance of marketing. The assurance that cosmetics be 100% safe is financially impractical and also scientifically impossible. Nevertheless, certain predictive testing is performed to insure that the number of reactions is minimal and that any reactions that might result from the use or abuse of the product are minor. For purposes of predictive testing, safety is frequently defined mathe- matically and is usually reported in statistical terms. Generally, safety statistics are based on one or more tests on a limited number of animal and/ or human subjects (4). The data from these tests are then extrapolated in order to classify a compound or a cosmetic as either "safe" or "not safe." However, there is still a statistical probability that the product may or may not elicit adverse reactions when used by the population at large. In many instances the tests or criteria do not take into considera- tion the actual conditions of use, climatic variations, individual differences, and other factors which should be examined before the statistical analysis attesting to the safety of the product is declared valid. Regardless of the tests performed and the number of subjects used, there is always, without exception, someone who might and will react to one or more of the in- gredients of a cosmetic. Evaluation of the safety of a cosmetic should be based primarily on tests that include as many variables as possible and that simulate conditions of actual use. The tests must provide a margin of safety, take into account the number of exposures the consumer might have, and also allow for a cer- tain amount of misuse. To disregard any or all of these factors is not concordant with the purpose of safety-testing, for the few of the many users who might experience an adverse reaction do not consider themselves as statistical freaks but rather as unfortunate injured parties who demand satisfaction. It has been the experience of the authors that there is very little correla- tion, if any, between the results obtained from animal tests and the actual use of the product by the consumer. There is also only a very limited correlation between prophetic human patch testing and actual in-use experience with the product (4). Last, but not least, there is some- times little relationship between the "established" safety of a raw material and its performance in a finished product. The following discussion of

EXPERIENCES IN SAFETY TESTING OF COSMETICS 163 these problems is primarily concerned with some of the practical experiences of the authors during the testing of cosmetics. THE CORRELATION BETWEEN ANIMAL AND HUMAN TESTS It is commonly accepted that animal skin or eye tests are sound guides to the safety of products or of raw materials. It must be remembered, how- ever, that animal skin is entirely different from human skin and that there may be no correlation between the mildness of a raw material on a rabbit's back and its safety during use on a human face. For example, it is widely recognized that isopropyl myristate is relatively irritating in the Draize rab- bit skin test (5). This compound, nevertheless, seems to be a safe ingredient for use on the human integument. Hair loss due to squalene has been demonstrated by Flesch (6) in animals, but squalene is a normal constituent of human sebum (7) and has been widely and safely used in cosmetics (8). Butcher (9) elicited increased mitosis, parakeratosis and desquamation through application of olive oil to the skin of rats, although olive oil has been used to anoint the human body--apparently xvith impunity--since Biblical times. The authors have also established that some finished cosmetic products, which are classified as mildly irritating (by the Draize rabbit skin test procedure), cause no reactions of any type under closed patch tests per- formed on humans and are safe when used under normal conditions by the population at large. Certain materials appear to be more irritating in the Draize rabbit skin tests than in human patch tests. If reliance were placed exclusively on the results obtained in animal tests, many useful raw materials or even finished preparations could not and would not be marketed. What is even more distressing is that many products that have been used for years with- out apparent ill effect would also have to be abandoned. This does not mean or should in no way imply that the use of animal skin tests should be abandoned for these tests, when properly used and intelligently inter- preted, can be valuable tools. If properly judged, i.e., not on an all or none basis, animal skin tests have considerable merit and probably can predict whether a finished product or new raw material may elicit severe irritation or toxic symptoms on application to human skin. The Draize rabbit eye test (5) has been widely used as a test of safety for products used in, on, around, or above the eyes. The great confidence placed by many in the validity of this test does not seem fully justified. Information has been presented (10) indicating that different laboratories, presumably using the same technique, have obtained conflicting results with the same product. This might indicate that results of the Draize rabbit eye test depend either on the conduct of the test, method of inter- preting the results, or perhaps even on the particular groups of rabbits