744 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS Another feature of the patch test which may give misleading results is the pressure used in applying the patch (2, 22). Fisher (2) pointed out that the very application of the patch to the skin is sufficient to cause an im- pregnated test substance to be forced to the periphery of the patch. In this way, uneven distribution of the test substance would be formed, giving rise to stronger reactions in the skin at the periphery of the patch than at the centre. For a mild irritant this could result in a score being given which would be consistent with that of a moderate or even a severe irritant. Evidence that certain substances appear less irritant when applied to the skin under occlusive patches (,24) and that other irritants such as solvents are unlikely to be detected using open patch tests (32) suggests that in a realistic assessment of primary irritants both open and closed patch tests are required (5, 18, 22, 39). Open tests involve an application of standard amounts of the test compounds to the skin followed by an evalua- tion of the cutaneous responses after a given period, e.g. 24 or 48 h. Such tests are carried out routinely in the evaluation of epidermal carcinogens. In this instance, applications are often repeated at frequent intervals for 30-80 weeks--or until tumours appear. In studies of early changes produced by carcinogens, an evaluation of the cutaneous response is occasionally carried out within 24 or 48 h (40-42). Repeated administration for testing irritancy has also been recommended for medicaments (17). However, no actual recommendations for repeated application appear to have been made concerning the testing of cosmetic chemicals for primary irritation. The Draize patch test in animals, although widely used for the detection of primary irritants, can duly be regarded as an approximate guide to the safety of a particular substance (1, 3, 5, 11, 43-45). Whereas tests in animals are usually satisfactory for detecting severe irritants, they are of considerably less value in tracing mild irritants due to their lower sensitivity (1, 25, 46). To increase the sensitivity of animal skins to optically applied substances, Finkelstein and his co-workers (47) pre-treated the skin with 20•o formalde- hyde solution. In their view, the sensitivity of the animal skins was increased such that the responses given were of a similar order to those given by human skin. Brown (48) reported recently that there was noticeably good agreement between the results obtained with occluded patches on formalin treated rat skin and those on untreated rabbit skin with a range of ionic and non-ionic surface active agents. It is considered, however, that formalin is sufficiently irritant to obscure the responses of an animal's skin to a mild irritant. Thus, the technique is of questionable relevance as a possible modification of the standard patch test.

APPRAISAL OF METHODS FOR DETECTING PRIMARY SKIN IRRITANTS 745 Since invariably animals are used in initial investigations for irritants, it would appear profitable for some basic studies to be carried out to ascer- tain whether or not any simple correlations do exist between animal and human predictive tests (1, 49). The existence of such a correlation was strongly doubted by Rieger and Battista (5) who also drew attention to the poor correlation of the results of human patch tests and 'in-use' investiga- tions. It has often been stressed that the use of animals for detecting primary irritants is likely to yield information only of a relative kind, i.e. compound A is more irritant than compound B and less than compound C. Such information could be of considerable value if substances of known degrees of irritancy to human skin were included in the animal testing (12-14, 25). The adoption of this principle, together with attention to the other factors described above, would greatly increase the value of animal skin testing. The relative merits of various animal species in tests for primary irritants have been reviewed previously (50, 51). The rabbit has been shown to be the most suitable species for detecting substances likely to irritate human skin, but it was considered advisable to use additional species. On the evidence available, rats, mice and guinea-pigs are satisfactory as secondary species. The difficulty in obtaining consistent results using the rabbit is well illustrated in a large scale collaborative study reported by Weil and Scala (52). If difficulty is encountered when using only one species, the use of additional species may further complicate the evaluation unless care is taken to ensure that the information sought is particularly relevant. Tests in man The patch test, as described by Draize (10, 53), for use in human studies is similar in detail to that used in rabbits. Occlusive patches are applied to suitable areas of skin on the forearm or back of subjects and the reactions are scored in a similar manner. Although an exposure period of 24 h was recommended for use in animal tests (10), Rostenberg (4) suggested that a test period of 48 h would be more suitable to detect obvious irritants in human tests. A modification of the Draize test was suggested by Lanman (24). It involves not just a single application of a test compound followed by a scoring of the skin reaction, but repeated application at daily intervals until such time as the skin shows some change in appearance. In this test, the