424 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS intestinal micro-organisms are sufficiently strong acids not to undergo appreciable absorption from the intestine. Thus the triarylmethane colourings are excreted almost quantitatively in the faeces (13). Never- theless, some absorption does take place. Hess and Fitzhugh (13) found that less than 5% of the dose of colouring admi'•istered by mouth appeared in rat, rabbit or dog bile, probably reflecting the excretory pathway of absorbed colouring. Although it is stated that none of the colours were found in the urine, one wonders whether re-examination of the question by up-to-date techniques might not reveal the presence of metabolites in urine. In the liver Lipid-soluble azo compounds do not undergo cleavage in the intestine but are absorbed intact and acted upon by liver azo reductase to form the corresponding primary amines (14). Other changes undergone by such azo compounds involve protein-binding, hydroxylation and other effects such as N- and O- dealkylation brought about by microsomal processing (drug metabolizing) enzymes. The subsequent fate of amines formed in the liver, or absorbed from the intestine, depends to a large extent on whether they are sulphonated and thus water-soluble, or whether they are alkyl-substituted anilines subject to extensive transformation within the body (15,16). In the former case they tend to be excreted intact in the urine or via the bile or they may undergo minor conjugation such as acetylation before urinary excretion. The lipophilic amines are subjected to hydroxylation, or oxidation of alkyl substituents which probably proceeds through hydroxy- lation. In the blood such acetylated amines or hydroxyamines exercise their well-known effect in inducing methaemoglobinaemia. This outcome ß has been recorded in man on numerous occasions following exposure to aniline or to "aniline dyes," as in children consuming coloured crayons. Methaemoglobinaemia is only the initial and, within limits, readily re- versible change. Further transformation of haemoglobin leads to the development of erythrocytic inclusions termed Heinz bodies, with increased destruction of the affected red cells, the development of anaemia, reper- cussions on the bone-marrow and deposition of haemosiderin in the spleen and liver. It is certainly true that in many instances these effects are only observed with high doses of colouring. Before one seeks to reach a

THE TOXICOLOGY OF ARTIFICIAL COLOURING MATERIALS 425 decision on what constitutes a maximum no-effect level, however, it is necessary to consider the following points: (a) The sensitivity of the method used to measure methaemoglobin (bearing in mind the fact that the rat normally has about 1 g methaemo- globin/100 ml blood). (b) The species differences in "resistance" to methaemoglobinaemia, presumably a reflection of the activity of methaemoglobin reductase (17). In this respect the cat is among the most sensitive species, man is almost as sensitive and the rat is a highly resistant species. Human genetically- determined deficiency of red-cell glucose 6-phosphate dehydrogenase creates a particular susceptibility to haemolytic amines, nitro compounds and the like (18). (3) Similar variations in degree of susceptibility to the formation of Heinz bodies. In our studies on food colourings, care has been taken to use a sensitive method for measuring methaemoglobin. In short-term feeding experi- ments carried out on rats, decreasing dietary levels of some colourings have given rise to Heinz bodies after increasingly long latent periods. At still lower dietary levels, although no effect was apparent, the application of a provocative test developed in our laboratories (19) revealed that a latent tendency to Heinz-body formation was still present. This example illustrates the errors that might have arisen from a mechanical, run-of- the-mill approach to safety evaluation, having as its objective the establish- ment of lack of toxicity rather than the discovery of the full facts. A comparison of the rat's susceptibility to the three isomeric dimethyl- anilines and to the trimethylanilines mesidine and pseudocumidine (derived from Ponceau 3R) has revealed the greater potency of the trimethylanilines in producing haemolytic anaemia, methaemoglobinaemia and Heinz bodies (20). General systemic toxicity Gastro-intestinal effects, and specifically catharsis, was an early concern and led to the prohibition of some colours for food use. On a chronic basis in animals, impaired weight gain and liver and kidney damage have often been the most striking effects brought about by high doses. Liver damage may be associated in long-term experiments with the production of liver rumours, which will be referred to below. Observations that present problems in interpretation are, firstly, the finding of an increase in relative liver weight in the absence of histopatho- logical evidence of liver damage, even at the highest dietary level tested