426 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS and, secondly, the presence of pigment granules in reticuloendothelial cells, especially of the liver, and/or in epithelial cells of the proximal convoluted tubules of the kidney. The significance of liver enlargement under these circumstances has been discussed (21). Storage of pigment granules that are neither haemosiderin, lipofuscin nor melanin, but probably represent the colouring in its original or a modified form, may reasonably be inter- preted as an effect of exposure to high doses. In such cases there is no reason to believe that under the conditions of intended use any storage of this sort is likely. Evidence of the uptake of certain dyes (not food colourings) in lysosomes lends particular point to the occasional occurrence of increased amounts of lipofuscin, particularly in the liver cells, after long-term feeding experi- ments. In the past such accumulations of what was called "wear and tear pigment" were ignored. Nevertheless the possibility exists that they represent the remains of previously-existing autophagic vacuoles which had been engaged in the task of restoring the integrity of damaged parenchymal cell cytoplasm. Be that as it may, it must be realized that present-day methods of assessing toxicity to the liver, by means of tests of liver function and observation of histopathological changes, leave something to be desired - especially when the problem is one of detecting low-grade chronic irrever- sible damage. Moreover, the laboratory animal exposed to such agents as colours is not simultaneously exposed to the multitude of environmental toxic hazards by which man is surrounded in everyday life, and which may in certain circumstances exercise a synergistic or even a potentiating effect on hepatotoxic action (22). Tests involving topical application The availability of excellent reviews (23,24), makes it unnecessary to deal here with questions of methodology. Mention should, however, be made of special problems, such as hair dyes, application about the eye, inclusion of colours in dental products, and tests on abraded skin. The effects of colouring matehals on skin and mucous membranes of experimental animals are at most a preliminary guide to further tests of the complete formulation in animals and man. Consequently it is not proposed to deal with the subject further here. CARCINOGENICITY Tumour production by azo compounds has been the subject of many reviews. Clayson (25) distinguishes dimethylaminoazobenzenes. other

THE TOXICOLOGY OF ARTIFICIAL COLOURING MATERIALS 427 aminoazo compounds, derivatives of phenylazo-2-naphthol and other azo compounds containing neither amino nor hydroxyl groups. Of primary interest here is the group of colourings, which includes such compounds as Carmoisine and Sunset Yellow FCF, which are derivatives of phenyl- or naphthylazo-2-naphthol. Despite a few suggestive earlier observations, the weight of evidence today leads to the conclusion that food colourings of this class are not carcinogens. Liver carcinogenesis by fat-soluble colours, of which p-dimethyl- aminoazobenzene is the prototype, and by water-soluble colours such as Ponceau 3R (26,27) creates considerable problems. Closely-related materials, for instance Ponceau 2R, are suspect even if liver tumours are reported to have been produced by analogues of this sort, one cannot be certain that adequate steps were taken to exclude traces of carcinogen (i.e. Ponceau 3R) that might have been present as impurities in the product tested. More fundamental, and more controversial, is the question - are low levels of such hepatocarcinogens acceptable for external topical applica- tion? Local carcinogenesis elicited by application to skin or mucous membranes is, of course, of special interest. Fortunately the colouring materials with which we are concerned do not produce such effects nor has carcinogenesis, or promoting action being attributed to any of the colourings employed in food or cosmetics. Bladder carcinogenesis has been studied (25) but has merely confirmed other observations, such as the carcinogenicity of Ponceau 3R and the non-carcinogenicity of Ponceau 2R, Blue VRS and Patent Blue V (28). Weighing up the evidence, it seems reasonable to accept the use of colourings that are hepatotoxic or hepatocarcinogenic in animals, provided that use is severely restricted to preparations for external application where ingestion will not occur. Tests by subcutaneous injection For some unaccountable reason, synthetic colouring matehals have been extensively tested for carcinogenic potential by repeated injection under the skin of rats. Of those colourings that have been investigated in this way, a substantial proportion have produced malignant tumours, usually fibrosarcomas, at the site of injection. It is likely that many, if not most, of the colourings not yet tested in this way are capable of inducing subcutaneous sarcomas. The extensive background to this problem has been reviewed in con- siderable detail by Grasso and Golberg (29). When the host of complicating