ELSEVIER Food Chemistry 50 (1994) 389-392 © 1994 Elsevier Science Limited Printed in Great Britain. All fights reserved 0308-8146/94/$7.00 Differentiation of legumes through elemental chemical composition using factor analysis E. Barrado, R. Pardo, B. Camarero Departamento de Quimica Analitica, Facultad de Ciencias, Universidad de Valladolid, 47005 Valladolid, Spain A. Tesedo & H. Romero Departamento de Nutrici6n y Diet~tica, Hospital Clinico Universitario, Valladolid, Spain (Received 9 June 1993; revised version received and accepted 23 September 1993) The proximate composition (water, carbohydrates, proteins, fats and ash and also sodium, potassium and chloride ions) of some legumes in the diet of a clinical hospital, over a 10-year period, has been determined. From the average values, or their graphical representations, large differences between several kinds of legume can be observed, but these differences are clearer and better explained by chemometrics methods (factor analysis and varimax rotation) which produce groupings of the samples into three types (fresh, canned and frozen) according to two factors: nutrients and salt. INTRODUCTION The composition of food is determined by many factors: order, family, species, climate, age, soil compo- sition, etc. According to Forina and Lanteri (1984) these parameters constitute the 'cause space' and deter- mine the quality, price, taste, odour, etc, which consti- tute the 'space effects'. The chemical composition or 'chemical space' can be used for the interpretation of cause-effect relationships. This can be done by analysing all components of food, or by measuring a few components and using chemometric methods to obtain the maximum information from the data. There are many examples, such as those of Castino (1975) who uses linear statistical discriminant analysis (LSDA) to classify Italian wine, and of Forina and Tiscornia (1982) who apply some methods (SIMCA, etc.) to the geological classification of olive oils. Modern methods of analysis, such as gas chromato- graphy, HPLC and voltammetry, provide numerous parameters in only one determination. This has greatly contributed to knowledge of chemical composition, but requires the use of computerized multivariate methods (Martens & Harries, 1983). On the other hand, the nutritional value of food products depends upon their composition in essential nutrients. In general, fresh legumes are important in ali- mentation because they offer a rather balanced composi- tion of carbohydrates, proteins and fat (Mataix & Salido, 1985). Other products on the market (canned and frozen legumes) have been industrially treated and their compo- sition and nutritive value might have been modified. In this paper a chemometric elemental analysis study (determined over 10 years, 1981-1991) of a legume group included in the diet of a clinical hospital has been carried out. When these data are treated by 'prin- cipal component analysis', this shows the relationships and differences between three kinds of legume (fresh, canned and frozen) which are badly perceived or diffi- cult to see by simple observation of the data or their graphic representation. MATERIALS AND METHODS The elemental analysis includes: % water (wind furnace method), % total protein (Kjeldahl method, conversion factor 6.25), total ash (%), carbohydrates (%) and % fat (AOAC, 1975), Na + and K ÷ (ppm) by flame photo- metry and C1- (ISE). All calculations were done on a PC computer using a home-made BASIC program available upon request. 389 RESULTS AND DISCUSSION Table 1 shows the average results of 30 analyses carried out between 1981 and 1991, three each year, of differ- ent kinds of legumes: • natural: chick-pea (cp), lentil (1), white bean (wb) and coloured bean (cb); • frozen legumes: green bean (fgb), broad bean (fbb) and pea (fp); • canned legumes: pea (cp) and green bean (cgb).