J. zyxwvuts Agric. Food Chem. zyxwvut 1984, 32, 651-655 65 1 Smith, A. K.; Watanabe, T.; Nash, A. M. zyxwvutsrq Food zyxwvutsrqp Technol. (Chicago) Staswick, P. E.; Hermodson, M. A.; Nielsen, N. C. J. Biol. Chem. Thanh, V. H.; Okubo, K.; Shibasaki,K. Plant Physiol. zyxwvut 1975,56, Thomson, J. A.; Schroeder, H. E. Aust. J. Plant Physiol. 1978, Tombs, M. P. Biochem. J. 1965,96,119. Tumer, N. E.; Richter, J. D.; Nielsen, N. C. J. Biol. Chem. 1982, Utaumi, S.; Nakamura, T.; Mori, T. Agric. Biol. Chem. 1982,46, 1960,14,332. 1981,256,8752. 19. 5, 281. 257, 4016. 1923. Utsumi, S.; Yokoyama, Z.; Mori, T. Agric. Biol. Chem. 1980,44, Wang, H. L.; Swain, E. W.; Kwolek, W. F. Cereal Chem. 1983, Wolf, W. J.; Tamura, T. Cereal Chem. 1969,46, 331. Yamagishi, T.; Yamauchi, F.; Shibasaki, K. Agric. Biol. Chem. 595. 60,245. 1980,44, 1575. Received for review October 28,1983. Accepted February 13,1984. This work was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science and Culture of Japan (No. 57760101). Pattern Recognition Analysis of Fatty Acids. Application to Beef Fat Tissue Classification Emile M. Gaydou,* Marie-Christine Iatrides, Romaine Ramananarivo, Jacques Artaud, and Gilbert Peiffer Thirty-six fat samples of hump, kidney, and depot fat of Malagasy zebu (Bos indicus), eight samples of kidney and tallow of Bos taurus, and two butter samples of Bos taurus were analyzed for 36 fatty acids by gas-liquid chromatography (GLC). Repeatability and GLC reproducibility analyses of the main fatty acids were checked. Intercorrelations among the fatty acids and their relations to fat tissue origins were investigated by analyzing the data with pattern recognition techniques. Nineteen fatty acids and four combinations of measurements were used for principal component analysis (PCA) and discriminant analysis (DA). By DA, 36 fat tissue samples of hump (15) and kidney (15) of B. indicus and tallow (6) of B. taurus were classed in three categories. Supplementary fat tissues were also successfully checked. Misclassifications were observed by DA, when only the main fatty acids were used. Nutritionists need detailed data on the amounts of fatty acids in foods, in order to evaluate current dietary habits and to execute nutritional research. Such data would require quantitative analysis of a vast number of foods. The use of conventional gas chromatography has allowed the identification of many fatty acids in beef tallows. The latest advances in glass capillary gas chromatography en- able even finer separation of fatty acids in complex mix- tures such as depot fats and fats from organs of mono- gastric species, polygastric species, and fishes (Flanzy et al., 1976; Ramananarivo et al., 1981). The use of long (60 and 100 m) and highly efficient glass capillary columns can solve the problems of the separation between peaks, the positional isomers, and the separation of the geometric isomers (cis and trans) for the determination of the amount of trans fatty acids (Slover and Lanza, 1979). However, the analysis is relatively long (1-3 h) with animal fats. The use of short glass capillary columns reduces considerably analysis time, and a comparative study on loo-, lo-, and 2-m glass capillary columns coated with SP 2340 was made by Lanza et al. (1980). The identification of every fatty Laboratoire de Phytochimie, Universit6 de Droit, zyxwvu 8- Economie et des Sciences, Centre de St. JBrCme, 13397 Marseille Cedex 13, France (E.M.G.), and Laboratoire des Organo-Phosphor& et Institut Universitaire de Technol- ogie, 13337 Marseille Cedex 14, France (M.X.I., R.R., J.A., and G.P.). acid can be time-consuming, and generally the workers limit their studies to fatty acids having concentrations higher than 0.1 zyxwv % . In this study, the fatty acid composition of various depot fats of French beef lipids (Bos taurus) and Malagasy zebu lipids (Bos indicus) was investigated. The repeatability and GLC reproducibility will be discussed. Multivariate statistical techniques, including discriminant and cluster analyses, which have been successfully applied in enological research (Kwan and Kowalski, 1980; Noble et al., 1980), were used to distinguish among various animal depot fats. Pattern recognition techniques were applied to 46 samples and proved successful in the distinction of various origin depot fats. The data presented in this investigation show the importance of minor fatty acids for the characterization of beef fat tissues. EXPERIMENTAL SECTION Fat Samples. Fifteen samples of hump tissue and fifteen samples of kidney fat, extracted from fifteen male zebu animals (B. indicus) were given by Sevima Co. (An- tananarivo, Mad%-). Three kidney tissue samples and three external depot fat samples of zebu were collected from a local market in Antananarivo. Six beef tallow samples, two kidney fat samples, and two samples of butter (B. taurus) were collected from a local market in Mars- eilles. Physicochemical Determinations. Determination of the iodine value (Wijs) and the saponification value were 0021-8561/84/1432-0651$01.50/0 0 1984 American Chemical Society