1. Introduction Stijn W.H. Van Hulle 1,2 and M. Cristina Ciocci 1 1 Research Group EnBiChem, Department of Industrial Engineering and Technology, University College West-Flanders, Kortrijk, Belgium 2 BIOMATH, Department of Applied Mathematics, Biometrics and Process Control, Ghent University, Ghent, Belgium For more information: Tel. ++32 56 241 251 Fax ++32 56 241 224 E-mail: Stijn.Van.Hulle@howest.be Objectives Statistical evaluation of the chemical quality of different water types There are two common types of drinking water: bottled water and tap water. Bottled water consumption has grown steadily over the past few decades. In 2003, bottled water became the second largest commercial beverage category by volume in the United States. Bottled water is considered to be a safe and readily available product and in several countries consumers prefer bottled water over tap water as drinking water. However, bottled drinking water consumption is associated with a higher economic and ecological cost. Furthermore, it can be demonstrated that the quality of tap water is not significantly different from the quality of bottled water in some developed countries. In view of this debate on the use of tap water and bottled water a statistical analysis was performed on the chemical composition of these waters in order to determine the differences in major ion content. As such this statistical analysis aims at classifying the different types of water as their chemical composition may differ. Methods The Piper diagram reveales that the main discriminating factor for tap water is the slightly increased chloride content. The PCA analysis revealed that the first principal component explains 48,1% of the total variance and contains most of the information on the monovalent ions. The first principal component (PC1) can therefore be considered as a measure for the water saltiness. The second principal component (PC2) explains 25,6% of the total variance and contains most of the information on the bivalent ions. As such, this principal component can be considered as a measure of the water hardness. The third principal component (PC3) explains 11% of the total variance and mostly contains information on the pH. Below the score plots are presented. The main difference between still and sparkling water is the pH (PC3) The results of the PCA were confirmed with LDA. Insight in classification of different types of water Conclusions The physico-chemical parameters of 84 different water types were collected. The chemical parameters used in this study were calcium (Ca 2+ ), magnesium (Mg 2+ ), potassium (K + ), sodium (Na + ), sulfate (SO 4 2- ), bicarbonate (HCO 3 - ), chloride (Cl - ), fluoride (F - ) and nitrate (NO 3 - ). As an independent check on the quality of the chemical analyses in the database charge balance error was calculated: Only water types with a calculated charge balance error of less than ± 10% were accepted in this study The Piper diagram of the 84 different water types is presented below: For statistical analysis, the software program SPSS version 17 was used (www.spss.com). ANOVA as well as PCA and LDA were used to select the most discriminating parameters and to investigate the overall variation of the data. Results and discussion There exists a difference between the different water types and that there also is exists a difference between different water brands. Principal component analysis revealed that the three main differences between different water types are saltiness, hardness and pH. Based on pH it was possible to discriminate between sparkling water and non-sparkling water Bottled still waters (♦), Bottled sparkling water (+ ) Tap water (o ).