Composition, Flavor, Chemical Foodsafety, and Consumer Preferences of Bottled Water

Authors


Direct inquiries to author Marcussen (E-mail: hma@life.ku.dk).

Abstract

 The worldwide consumption of bottled water has experienced an annual increase of 5.5% since 2004 and different authors have speculated that this is due to beliefs of superior flavor and health qualities of bottled water over tap water. The content of certain minerals varies by 2 to 6 orders of magnitude between bottled water brands. The major cations Ca, Mg, Na, and K can be found in concentrations up to 730, 450, 1900, and 270 mg/L but are usually below 100 mg/L; higher concentrations have only been found in some European and North American bottled water samples. The dominating anions are HCO3, Cl, and SO42−. Sensory evaluation and description of water strongly depend on personal preferences and sensitivities. The major cations can be detected by taste in concentrations down to 2 digits in the mg/L range; however, the intensity depends on which anions are present. Most cations add different degrees of salty, sour, sweet, and bitter tastes to water depending on their concentrations. Al, Ca, Cu, Fe, Mg, and Zn may introduce metallic, astringent, and irritative sensations and for Fe and Cu also retro-nasal odors may influence the metallic sensation. Drinking water with total dissolved solids in the range of 100 to 400 mg/L results in good sensory quality. Known off-flavor problems originate from by-products of ozonation and leaching of organics from bottling material which may also be enhanced by ozonation, and microbial by-products from the source water such as geosmin and 2-methylisoborneol, lubricants, or cleaning agents used in the bottling industry.

Introduction

Bottled water was 1st introduced as a commercial beverage category in France by Evian in 1829. Today, bottled water has become a truly global commodity found even in remote parts of third world countries and world bottled water consumption has been increasing every year (Mascha 2006; Rodwan 2010). Yet bottled water is 240 to 10000 times more expensive than tap water and the quality of tap water is generally getting better (Natural Resources Defense Council 1999; Doria 2006). Therefore, it is intriguing why bottled water has become such a commercial success.

Different authors speculate that the increase in bottled water consumption is due to a lack of trust in tap water quality, beliefs that bottled water has superior health promoting properties and better flavor qualities compared to tap water, and the fact that bottled water is marketed as trendy and a sign of youth (Levallois and others 1999; Anadu and Harding 2000; Castaño-Vinyals and others 2002; Lou and others 2007). Loss of trust in tap water quality may have originated from episodes in which pathogenic microorganisms had been discovered in water sources, storage facilities, or the distribution system (Parag and Roberts 2009). Media reports of chemical pollutants such as pesticides may also be considered to decrease the general trust in water purity or the health aspects of drinking tap water. Poor sensory quality of tap water is a 3rd reason for the lack of consumer trust (Levallois and others 1999; Doria and others 2009; Pintar and others 2009). Yet, while some studies indicate that consumers believe bottled water to have superior health and sensory qualities over tap water (Ward and others 2009; Doria and others 2009), there is little scientific evidence to demonstrate the superior quality of bottled water over tap water. Furthermore, the understanding of what determines good sensory quality in water and hence consumer preference for one type of water over another is not well understood (Moskowitz and Krieger 1995).

This paper presents the current state of knowledge about the organoleptic quality, positive and negative chemical health implications, and consumer perception of bottled drinking water. The influence of inorganic and organic constituents on the flavor and chemical health implication of drinking water is reviewed. Concentration ranges of inorganic and organic compounds in bottled waters are reported and their possible influence on flavor and health are discussed. Microbial food safety in bottled water is a major topic. It will not be addressed here as it must be addressed specifically.

Bottled Water Definition and Regulation

Bottled water is defined as all types of waters for human consumption marketed in bottles. Bottled water is normally divided into different categories depending on its origin and treatment. The regulation with respect to bottling and microbiological and chemical quality differs between the different categories. In the European Union (EU), the following 3 categories are used: (1) natural mineral water originating from a ground water reservoir; (2) spring water, namely, groundwater in a natural state but bottled at the source; and (3) other types of bottled drinking water, including bottled tap water (Table 1). In this review, Category 3 water will be denoted miscellaneous bottled water. Miscellaneous bottled water is regulated under the same directive as tap water, namely the EU Drinking Water Directive (EC 1998). It must therefore conform to all standards in this directive, which comprises 48 quality parameters, including limit values for anions and cations, conductivity, flavor, color, and microbial quality. These parameters must be regularly monitored and the water in permitted to be treated if it fails to comply with the standards. Natural mineral water is regulated under the EU Mineral Water Directive (EC 1980, 1996). This directive was adopted to allow for flavor differences and the high mineral content in some mineral waters that is not allowed under the Drinking Water Directive. Natural mineral water is defined as microbiologically safe water originating from an underground water reservoir and tapped from one or more springs or extraction wells. The regulation for mineral water is less strict compared to the regulations for tap water; it is assumed that the underground origin protects the water from being polluted. The composition, temperature, and other essential characteristics of natural mineral water sources must remain unaffected by extraction, which means the water characteristics must not be affected by variations in the pumping flow rate. The directive contains no limits for electrical conductivity and the content of major anions and cations, although maximum limits have been set for the content of potentially toxic ions and elements: antimony, arsenic, barium, boron, cadmium, chromium, copper, cyanide, fluoride, lead, manganese, mercury, nickel, nitrate, nitrite, and selenium (EC 2003). The source must be free of pathogenic microorganisms and parasites, which means the maximum total colony count after bottling must not exceed 100/mL at 20 to 22 °C for 72 h and 20/mL at 37 °C for 24 h on agar–agar or agar–gelatine medium. Bottled water sold as natural mineral water can be treated, but only to reduce the contents of Fe, Mg, S, and As. In addition, filtration, decantation, oxidation, and treatment with ozone-enriched air are allowed if they do not alter the elemental composition of the water. Elimination, introduction, or reintroduction of carbon dioxide to mineral water is also allowed. The bottled spring water must conform to the standards of the EU Drinking Water Directive (EC 1998). Further, it is subjected to many of the requirements of the Natural Mineral Water Directive such as restrictions on treatment and addition of minerals. An overview of the different requirements of the 3 categories of bottled water is given in Table 1.

Table 1. European bottled water categories, origin, treatment, and regulation
 Bottled water categories
 Natural mineral waterSpring waterMiscellaneous
OriginGround water tapped from natural spring of extraction wellsGround water spring bottled at source.All underground or surface waters.
TreatmentOnly filtration, decantation, oxidation, andSame treatment as for natural mineral water isTreatment and addition of minerals are
  ozonation to remove iron, manganese, sulfur, and arsenic is allowed. Addition of carbon dioxide is allowed. allowed. allowed.
RegulationMineral water directive (EC 1996). Directive of concentrations and labeling of mineral water (EC 2003).Mineral water directive (EC 1996). Directive of concentrations and labeling of mineral water (EC 2003). Drinking water directive (EC 1999).Drinking water directive (EC 1999).

Consumption of Bottled Water

Water is an essential food and the human body consists of about 70% water. The general recommendation is for an adult person to consume approximately 2 L of water per day, excluding beverages such as coffee, tea, and other drinks containing caffeine or alcohol, which seems to be more a myth than sound advice based on scientific evidence (Valtin 2002; Negoianu and Goldfarb 2008). However, it is clear that a substantial daily water intake is required and that the required amount depends on the climate and the activity level of the person in question (Valtin 2002). Sources of water would include all beverages as well as fruits (approximately 90% water) and solid foods (2% to 70% water). For the average human that would be about 2 L per day as evidenced by the daily loss of 2 L with excretion, mainly urine, as well as exhalation and perspiration.

Worldwide consumption of bottled water has been steadily increasing. According to the Beverage Marketing Corp., the worldwide annual growth rate in bottled water consumption for the period 2004 to 2009 was 5.5%, resulting in a consumption of 203 million metric tons in 2009 (Rodwan 2010). This is a significant increase considering that the corresponding world population growth rate was below 1.3% for the same period (U.S. Census Bureau 2010). The per capita annual consumption of bottled water ranged between 82 and 234 L in 2009 for the 20 countries with the highest consumption but the global average of 30 L per person per year was still relatively low, even though it had increased by 23% since 2004 (Figure 1) (Rodwan 2010). European countries have traditionally had the highest bottled water consumption and 12 of the top 20 countries with the highest per capita consumption are still European (Rodwan 2010). However, today Mexico is the country with the highest per capita consumption and the United Arab Emirates, Lebanon, the United States, Thailand, Saudi Arabia, Qatar, and Hong Kong also have high consumption rates (Figure 1) (Rodwan 2010). Although the consumption in most Asian countries is still fairly low, the Asian share of the global market has been increasing sharply. China is today the country with the 3rd highest total bottled water consumption and the annual growth rate is 12.3% (Rodwan 2010). Global bottled water consumption has increased even though the quality of municipal tap water is improving in many countries and bottled water does not generally seem to have a better quality than tap water (Hunter 1993; Lalumandier and Ayers 2000; Saleh and others 2001; Ward and others 2009). Furthermore, bottled water may in many cases just be tap water filled into bottles with or without further treatment. For instance, a consumer survey in 2006 showed that 44% of all bottled water in the United States was in fact bottled tap water (Gleick and Cooley 2009). Bottled water drinkers often cite flavor as the main reason for choosing bottled water over tap water, but there is little scientific evidence to support that bottled water has a superior sensory qualities over tap water (Napier and Kodner 2008). A consumer survey of residents in Canada showed that the main reason for not drinking tap water was dissatisfaction with flavor and that up to 37% had bottled water as their only drinking water source (Levallois and others 1999; Pintar and others 2009). Consumer surveys elucidating the consumption of bottled water and the motives for choosing bottled water over tap water have not been published from low-income and medium-income countries where tap water is often of poorer quality. However, it is known that the average per capita consumption of bottled water is low in such countries and that, among the low- and medium-income countries, anywhere from 2.8% to 100% of the population treat their tap water before drinking it (Rosa and Clasen 2010). The increased use of bottled water is likely a combined result of beliefs in sensory and health benefits, marketing campaigns and convenience, but lack of trust in the quality of tap water may also be a factor (Parag and Roberts 2009; Ward and others 2009).

Figure 1.

Per capita consumption of bottled water by countries with highest bottle water consumption and world average (Rodwan 2010).

The increasing bottled water consumption has a considerable environmental impact since the production of bottled water requires about 1000 to 2000 times more energy compared to tap water (Gleick and Cooley 2009). Life-cycle assessments (LCA) where the environmental impact of a product is assessed in stages from cradle to grave have shown that the environmental impact of bottled water is 90 to more than 1000 times higher than that of tap water, depending mainly on how far the water is transported (Jungbluth 2005). Production of bottles and transportation is the most energy-consuming part; the handling of discarded bottles may also result in great energy demands but has not been included in energy calculations due to many unknowns (Gleick and Cooley 2009).

Composition of Bottled Water

Inorganic constituents

The chemical composition of bottled water is relatively well known, particularly with regard to the major elements and ions. Through a literature review we collected 40 peer-reviewed articles and 1 book showing data on the inorganic content of up to 67 constituents and other parameters of a total of 1935 bottled water samples, some of which may refer to the same brand of bottled water as the brands were not stated in all articles. In total, 35 of the 41 studies reported values based on the authors’ own analyses, whereas the remaining 6 reported values were from bottle labels or certificates supplied by the producer. The range of the inorganic composition of bottled waters based on this literature review is presented in Table 2 (major ions and elements) and 3 (minor elements).

Table 2. Chemical composition of bottled water. The number of studies (N) that included the parameter and the total number of bottled water brands (Nb) analyzed are shown. Units are in mg/L except for conductivity, which is in μS/cm
 RangeNNbReferences
  1. 1Ahmad and Bajahlan 2009;

  2. 2Alabdula'aly 1997a;

  3. 3Alabdula'aly 1997b;

  4. 4Allen and others 1989;

  5. 5Al-Saleh and Al-Doush 1998;

  6. 6Azoulay and others 2001;

  7. 7Baba and others 2008;

  8. 8Barroso and others 2009;

  9. 9Chiarenzelli and Pominville 2008;

  10. 10Chiba and others 2006;

  11. 11da Costa Grec and others 2008;

  12. 12Dabeka and others 1992;

  13. 13Dabeka and others 2002;

  14. 14Dobaradaran and others 2008;

  15. 15Feru 2004;

  16. 16Fiket and others 2007;

  17. 17Garzon and Eisenberg 1998;

  18. 18Güler and Alpaslan 2009;

  19. 19Güler 2007;

  20. 20Hardisson and others 2001;

  21. 21Ikem and others 2002;

  22. 22Karamanis and others 2007;

  23. 23Kim and Lee 1997;

  24. 24Krachler and Shotyk 2009;

  25. 25Lalumandier and Ayers 2000;

  26. 26Mahajan and others 2006;

  27. 27Mayne and Edwards 1990;

  28. 28Misund and others 1999;

  29. 29Momani 2006;

  30. 30Nkono and Asubiojo 1997;

  31. 31Pip 2000;

  32. 32Rizk 2009;

  33. 33Rosborg and others 2005;

  34. 34Saleh and others 2001;

  35. 35Saleh and others 2008;

  36. 36Shotyk and others 2006;

  37. 37Soupioni and others 2006;

  38. 38Saad and others 1998;

  39. 39Toumba and others 1994;

  40. 40Versari and others 2002;

  41. 41von Wiesenberger 1991.

  42. TC, total carbon; TOC, total organic carbon; TIC, total inorganic carbon; TH, total hardness; LOD, limit of detection.

Ca<LOD-7292514141,4,6,7,9,10,13,15,17,18,19,21,23,24,26,27,28,29,32,33,34,35,37,40,41
K<0.05 to 268187891,7,9,10,13,18,21,23,26,28,29,32,33,34,35,37,38,40
Mg<0.005 to 4472514031,4,5,6,7,9,10,13,15,17,18,19,21,23,24,26,28,29,32,33,34,35,38,40,41
Na<0.3 to 19072413821,4,6,7,9,10,13,15,17,18,19,21, 23,24,26,28,29,32,33,34,35,38,40,41
Si<LOD-46.662847,18,28,33,34,35
Sr<0.00017 to 80.1116925,7,10,13,16,18,24,28,32,33,35
B<0.001 to 21.4118774,7,13,16,18,19,28,32,33,35,40
SiO2−103113240
S−321189
Br<LOD-5.332181,13,34
Cl<LOD-532188461,4,7,9,15,18,19,21,26,28,29,33,31,32,34,35,38,40
ClO3<LOD-1.0119913
CN<LOD-0.6229419,35
Cr6+<LOD-0.00072119913
F0.006 to 8.4199861,2,4,11,12,14,18,19,20,25,26,29,32,33,34,35,38,39,40
HCO31 to 402674877,15,18,19,32,33,40
NH4+<0.009 to 3544118,19,23,33,40
NO2<0.03 to 10.1550118,19,34,35,40
NO3<0.04 to 129137211,3,18,19,26,28,29,31,32,33,34,35,38,40
PO43−<0.01 to 1053004,18,28,34,40
SO42−<0.1 to 1371168111,4,7,15,18,19,21,28,29,31,32,33,34,35,38,40
SO32−<LOD-0.117018
pH4.17 to 9.16136701,7,9,15,18,19,21,26,32,33,34,35,40
CO20.1 to 2100113240
O21 to 86113240
TDS5 to 340082931,9,15,26,29,31,32,40
Conductivity6.1 to 52700114794,7,18,21,26,29,32,33,34,35,40
Alkalinity2.7 to 28924821,29
TC0.67 to 7812421
TOC0.48 to 42.3412421
TIC0.2 to 7112421
TH as CaCO33.9 to 2644721, 26,29,32
Table 3. Concentration in μg/L of minor elements in bottled water. The number of studies (N) that included the parameter and the total number of bottled water brands (Nb) analyzed are shown
 RangeNNbReferences
  1. For references see footnotes in Table 2.

Ag0.0004 to 9296154,13,16,21,24,28,33,34,35
Al<0.06 to 25001512034,7,8,12,13, 16, 18,19,21, 24,28,33,34,35,40
As0.006 to 49128999,12,13, 16, 18,19,21,28,33,34,35,37
Ba0.02 to 2000138324,7,10,13,16,18,24,28,33,34,35,37,40
Be0.00004 to 3185884,5,13,24,28,33,34,35
Bi<0.001 to 252717,28
Br0.6 to 8890416318,28,33,37
Cd0.0006 to 242112154,5,7,9,10,12,13,16,18,19,21,24,26,28,29,30,31,33,34,35
Ce<0.002 to 1.3715628
Co0.0009 to 11137084,7,13,16,18,21,24,28,30,33,34,35,37
Cr0.001 to 102199964,5,7,9,10,13,16,18,19,21,22,24,28,29,30,33,34,35,37
Cs<0.002 to 3628928,33
Cu<0.01 to 7611810054,5,7,9,13,16,18,19,21,22,24,26,28,30,31,33,34,35
Dy<0.002 to 49212618,28
Er<0.002 to 48212618,28
Eu<0.002 to 0.1215628
Fe0.07 to 4500157434,5,7,9,16,19,21,22,24,28,29,32,33,34,40
Ga<0.002 to 0.03815628
Gd<0.002 to 0.2615628
Ge<0.001 to 119218824,28
Hf<0.002 to 0.215628
Hg<0.005 to 7594884,5,9,19,21,28,33,34,35
Ho<0.001 to 0.05315628
I<0.1 to 1400322128,33,40
In<0.001 to 0.00215628
La0.001 to 1.2326028,37
Li0.03 to 546086304,13,16,23,24,28,33,40
Lu<0.001 to 0.02415628
Mn<0.005 to 11501610164,5,7,13,16,19,21,22,24,28,29,30,33,34,35,40
Mo0.003 to 179117014,13,16,18,21,22,24,28,33,34,35
Nb<0.002 to 0.615628
Nd<0.002 to 1.1515628
Ni<0.01 to 420147924,5,7,9,13,16,18,19,22,28,30,33,34,35
P0.21 to 718421318,33,34,35
Pb<0.001 to 741710954,9,12,13,16,19,21,22,24,26,28,29,30,31,33,34,35
Pr<0.002 to 0.2915628
Rb0.005 to 840322124,28,33
Sb0.001 to 51188013,16,18,19,24,28,33,34,35,36,37
Sc0.0005 to 8.54216524,33
Se<0.01 to 293137465,10,13,16,18,19,22,28,30,33,34,35,37
Sm<0.002 to 0.1815628
Sn<0.002 to 30462624,7,22,28,33,35
Ta<0.002 to 0.4315628
Tb<0.001 to 0.03615628
Te0.0006 to 0.18218824,28
Th0.00002 to 2842309,21,24,28
Ti<0.005 to 13673087,16,18,28,33,33,35
Tl0.00004 to 0.9652113,16,24,28,34,35
Tm<0.001 to 0.02315628
U0.0002 to 9283067,9,16,21,24,28,33,37
V0.0006 to 93137364,5,9,10,13,16,18,22,24,28,33,34,35
W<0.002 to 1628928,33
Y0.002 to 2.315628
Yb<0.002 to 0.1215628
Zn<0.001 to 980199924,7,9,10,13,16,18,19,21,22,24,26,28,29,30,33,34,35,37
Zr0.003 to 731047,28,33

As seen from Table 2 and 3, the constituent concentrations vary by 2 to 6 orders of magnitude. The chemical composition of bottled water depends to a certain extent on the category of bottled water to which it belongs. Natural mineral water and spring water are not allowed to be submitted to extensive treatments that alter the chemical composition of essential minerals, and the composition therefore depends very much on the geochemistry of the ground water reservoir. In contrast, miscellaneous bottled water (Category 3) may be extensively treated, as is the case for desalinated seawater with added minerals (EC 1980). Contamination during the bottling process, or from the bottles during storage, may influence the chemical composition of all types of bottled water (Misund and others 1999; Shotyk and others 2006; Krachler and Shotyk 2009; Reimann and others 2010). The major elements in bottled water are Ca, Mg, Na, and K, which can have concentrations up to 730, 450, 1900, and 270 mg/L, respectively. These elements thereby constitute a high percentage of the total dissolved solids (TDS), which can amount up to 3400 mg/L but in most cases will be much lower (Table 2). Elements other than Ca, Mg, Na, and K are found in concentrations below 100 mg/L. The elements Ca, Mg, Na, and K are ubiquitous in most rocks and sediments and hence their concentrations are usually high, depending on the geochemical composition of the aquifer. Calcium, Mg, and Na concentrations in individual bottled waters are presented in Figure 2. Bottled waters from Asia, Melanesia, and the Middle East had concentrations of these 3 elements below 100 mg/L. This was also the case for most bottled waters from Europe and North America; however, some bottled waters from these 2 regions had much higher concentrations of one or more of these elements. For all regions, except the Middle East, most waters were dominated by Ca. Slightly fewer waters were dominated by Na, whereas only very few waters had a high percentage of Mg. Bottled waters from the Middle East were dominated by Na.

Figure 2.

Calcium, Mg, and Na concentrations in bottled water in different regions (Kim and Lee 1997; Garzon and Eisenberg 1998; Saad and others 1998; Azoulay and others 2001; Dabeka and others 2002; Ikem and others 2002; Feru 2004; Rosborg and others 2005; Momani 2006; Saleh and others 2008; Chiarenzelli and Pominville 2008; Ahmad and Bajahlan 2009; Güler and Alpaslan 2009).

The dominant anions are HCO3, Cl, and SO42− and these have been found in concentrations up to 4030, 530, and 1370 mg/L, respectively (Table 2). Concentrations of these anions for individual bottled waters have only been reported for Europe and northern Asia (Figure 3). All samples from northern Asia had SO42− concentrations below 1 mg/L and Cl concentrations up to 20 mg/L, whereas HCO3 concentrations were up to 190 mg/L; most were, however, below 100 mg/L. Samples from Europe had much higher anion concentrations corresponding to their higher cation concentrations (Figure 2 and 3). In most cases, Cl and SO42− concentrations were below 100 mg/L. HCO3 concentrations were above 500 mg/L for most samples with some high extremes of 2000 to 3420 mg/L. That HCO3 is the dominant anion is clearly seen from Figure 3, which shows the distribution of Cl, SO42−, and HCO3 as mass percentages. An important anion with regard to chemical safety is NO3. High NO3 concentrations in drinking water are normally a result of contamination from agricultural production and cities. Nitrate concentrations up to 129 mg/L have been reported by Momani (2006), whereas other studies have found concentrations up to 50 mg/L, which is the limit value for NO3 in tap water in the European Union (EC 1998; Misund and others 1999; Saleh and others 2001; Versari and others 2002; Rosborg and others 2005; Saleh and others 2008). Other inorganic anions such as Br, ClO3, CN, CrO42−, F, NO2, PO43−, and SO32− are all present in concentrations below 10 mg/L (Table 2).

Figure 3.

Chloride, SO42−, and HCO3 in bottled water in Europe and Northasia (Feru 2004; Rosborg and others 2005; Güler and Alpaslan 2009).

A study of the chemical content of 128 bottled waters from Italy showed that the parameters TDS, conductivity, HCO3, Ca, Mg, Na, and K were highly correlated; whereas Cl only correlated with Na (Versari and others 2002). In another study Cl was also found to correlate with TDS, which shows that such correlations depend very much on the specific bottled waters analyzed (Pip 2000). Misund and others (1999) reported on the concentrations of 66 elements in 56 brands of mineral water purchased all over Europe. They found that for most minor elements, the concentrations correlated with the total element content. However, the concentrations of elements such as Cd, Cu, Zn, Fe, Pb, and As in mineral water may not correlate with the total element content due to pollution, their differences in geochemistry (for example sorption, solubility) with respect to pH and redox sensitivity, or regional differences in aquifer geology (Misund and others 1999; Fiket and others 2007). Misund and others (1999) further found a tendency for bottled mineral water marketed in Russia, the Baltic countries, and Germany to have relatively high total element contents. They did not interpret this as a result of the regional groundwater reservoir geology but rather that people from these countries favor or are accustomed to drinking water with high salt concentrations. However, Misund and others (1999) did discover regional trends in a few elements that were not linked to the total element content. There was a tendency toward higher As concentrations in waters from France and Germany and the highest Cu concentrations were seen in samples from Russia and the Baltic countries.

As mentioned, the mineral content of bottled water may be influenced by packaging and storage. Studies have indicated that Pb, U, Zr, Li, K, Na, and Th leach from glass bottles, since water packed in glass had much higher concentrations of these elements compared to water in plastic bottles (Misund and others 1999; Shotyk and Krachler 2007; Krachler and Shotyk 2009). Lead concentrations were 26 to 57 times higher in glass bottles compared to plastic bottles, and longer storage time further increased the concentrations (Shotyk and Krachler 2007). On the other hand, high Sb concentrations have been found in water stored in plastic bottles, since Sb2O3 is used as a catalyst in polyethylene terephthalate (PET) production (Shotyk and others 2006). Reimann and others (2010) studied the leaching of 57 elements from 126 different bottles including glass, and soft and hard PET bottles in different colors. They found that leaching concentrations for most elements were low compared to the concentrations found in natural mineral water. However, some elements, including Ce, Pb, Al, Zn, Cu, and the rare earth elements, leached from glass bottles and Sb from PET bottles to an extent that could significantly influence the concentration of these elements in the bottled water. The study also showed that colored bottles leach more elements compared to none colored bottles (Reimann and others 2010).

The concentration of selected major elements and ions are given on the label of many bottled waters. However, there is no EU regulation regarding the quality of the chemical analyses, labeling, and specification of the chemical composition of bottled waters. Consequently, the number and types of elements with reported concentrations of bottled water labels differ between brands. Misund and others (1999) generally found that concentrations reported on bottle labels agreed well with the actual content while other studies have found deviations of up to 500% from the determined concentrations (Allen and others 1989; Saad and others 1998; Pip 2000; da Costa Grec and others 2008). This discrepancy could be due to a change over time in the composition of the water source, changes in element concentrations during storage, or the analytical procedures.

Source water which contain bromide and has been subjected to ozone treatment may also contain bromate which is a potential carcinogen for humans (Nyman and others 1996; Liu and Mou 2004; Huang and others 2009; Aljundi 2011; Kim and Hyun 2012). Bromate in ozone-treated bottled water has been observed in concentrations up to 38 μg/L (Nyman and others 1996; Warner and others 1996).

Organic constituents

Compared to the content of inorganic constituents much less is known about the content of organic compounds in bottled water, and such information is not provided on bottle labels. A few studies have focused on organic compounds in bottled water and due to health concerns there is increasing attention on the concentrations of organic constituents (Leivadara and others 2008). Organic compounds in bottled water may comprise dissolved organic matter (DOM) originating from natural processes. Anthropogenic contaminants may result from industrial production, agricultural practices, households use, and so on, which contaminate the aquifer or may be introduced to the bottled water during bottling or by leaching from bottle material. Another important type of organic contamination adheres from bottling equipment or containers or are disinfection by-products from treatment water by ozonation, and to a less degree chlorination and chloroamination (Richardson 2003; Chen and Westerhoff 2010). The total organic carbon content of bottled water has been found to range between 0.48 and 42 mg/L and it therefore contributes little to the TDS, which may range from 5 to 3400 mg/L (Ikem and others 2002) (Table 2).

Disinfection by-products of ozone

Ozonation, reverse osmosis, distillation, and de-ionization are common treatment methods of bottled water. Ozonation can be used for disinfection, removal of manganese, organic micropollutants and can improve the color, taste, and odor of drinking water (Schneider and Rump 1983; Kruithof and Masschelein 1999; Kuo and Chan 2012). However, the oxidation of organic matter in drinking water by ozone may result in different types of disinfection by-products, some of which are known or suspected to be carcinogenic or harmful in other ways (Richardson 2003; Richardson and others 2007; Huang and others 2009; Kim and Hyun 2012). Ozone disinfection by-products have been studied both in laboratory and treatment plants and the following groups of compounds have been identified: aldehydes, ketones, ketoaldehydes, carboxylic acids, aldo-acids, keto-acids, hydroxy-acids, alcohols, and esters (Richardson and others 2000; Huang and others 2009; Qi and others 2009; Kim and Hyun 2012). Even though many ozone disinfection by-products have been identified, the potential health effects of these are poorly understood and it is expected that more than 50% of the disinfection by-products formed are still not accounted in chemical analysis (Richardson 2003). In cases where the source water contains Br ozonation may result in brominated organic disinfection by-products such as bromoform and bromoacetic acids which are more carcinogenic and mutagenic and considered more harmful to humans than their chlorinated analogues produced by chlorination (LaLonde and others 1997; Nobukawa and Sanukida 2001; Richardson and others 2007; Huang and others 2009). Concentrations of brominated by-products increase with increasing Br, O3, and DOM concentrations (Huang and others 2009).

Constituents from bottle material

Organic contaminants in bottled water may also originate from the bottle material itself. The most widely used materials for bottles, caps, or liners are PET, high-density polyethylene (HDPE) and polypropylene (PP) and ethylene vinyl acetate (EVA). Dabrowska and others (2003) found leaching of formaldehyde and acetaldehyde from PET bottles to water stored in the bottles which resulted in water concentrations up to 60 and 78 μg/L, respectively after a storage time of 170 days. Different authors have found that the leaching of formaldehyde and acetaldehydes increase with increasing storage time, temperature, and carbonation level (Nawrocki and others 2002; Dabrowska and others 2003; Ewender and others 2003). In the study by Dabrowska and others (2003) 3 additional carbonyl compounds were found in ozonated water which had been used for wash of PET bottles namely propanal, nonanal, and glyoxal indicating that ozone may increase the release of aldehydes from PET material. Song and others (2003) studied the release of volatile low molecular weight compounds from different plastic materials used for packing of water. They found no leaching under none ozonated conditions, while in presence of ozone they observed a release of small aldehydes with 4 to 11 carbon atoms and hexanone. Several studies have shown that ozone can oxidize and degrade PET, HDPE, and PP materials and modify their functionality (Peeling 1983; Oueslati and Catoire 1991; Hill and others 1995; Niew and others 1999; Ton-That and others 1999; Teare and others 2000). Bach and others (2012) reviewed the literature regarding leaching of different compounds from PET both without and in the presence of ozone. PET is the preferred material for bottles as it is produced with the fewest additives. They concluded that there is consistent evidence for leaching of formaldehyde and acetaldehyde but reports of other leaching substances such as phthalates, alkylphenols, antioxidants, UV stabilizers, lubricants, and other carbonyl compounds still need further proof and may be a result of contamination or bottled lids.

Plasticizers such as phthalates and adipates are not firmly bound to the bottle polymer material and may migrate into foods stored in plastic containers. Leivadara and others (2008) found the plasticizer diethylhexyl phthalate (DEHP) in 13 brands of Greek bottled water. The DEHP concentration increased with storage time and the maximum observed concentration was 6.8 μg/L. In Chinese bottled mineral water the plasticizers diethyl phthalate and di-n-butyl phthalate were detected in concentrations of 2.8 and 4.6 μg/L, respectively (Xu and others 2007). Yamini and others (2009) did not find the plasticizers DEPH and di-2-ethylhexyl adipate (DEHA) in bottled water kept in polyethylene before the expiration date, while 3 mo after the expiration date concentrations of DEHA and DEPH rose to 0.41 and 0.71 μg/L, respectively. Fayad and others (1997) studied the migration of vinyl chloride and plasticizers from polyvinyl chloride (PVC) drinking water bottles. They found that phthalate esters were the major organic compounds identified in the water samples and the migration of phthalate esters was highly dependent on storage time, temperature, and exposure to sunlight. The observed vinyl chloride concentrations were up to 0.6 μg/L (Fayad and others 1997). Styrene has also been found to leach into bottled water packed in polystyrene bottles in concentration up to 46.4 μg/L which is higher than the WHO guideline value of 20 μg/L (Ahmad and Bajahlan 2007; Al-Mudhaf and others 2009).

Page and others (1993) found that 23 of 182 bottled waters marketed in Canada contained cyclohexane, with a maximum concentration of 108 μg/L. Cyclohexane may originate from the production of polyethylene as it is used as a solvent in the polymerization process. Ethyl benzene, which is used in the production of polystyrene, has also been detected in bottled water in concentrations up to 3.2 μg/L (Ahmad and Bajahlan 2007).

Constituents from source water

As Category 3 bottled water may be bottled tap water, which has been subjected to chlorination another range of disinfection by-products may be found in some bottled water. Trihalomethanes (THMs), including chloroform, trichloromethane, bromodichloromethane, dibromochloromethane, tribromomethane, and haloacetic acid (HAA) and, in particular, dichloroacetic acid and trichloroacetic acid, may form as a result of the disinfection of drinking water with chlorine or chloroamine (Chen and Westerhoff 2010). These products can therefore be found in some bottled waters though the concentrations must be expected to be lower than concentrations found in tap water. Leivadara and others (2008) studied disinfection by-products in 13 brands of Greek bottled waters. In most bottles the concentrations were below the detection limit and the highest observed concentrations were 21.7 μg chloroform per liter and 1.8 μg bromochloroacetic acid per liter. The study showed that the concentrations of the different disinfection by-products depended on storage time and temperature. THMs formed during storage, whereas HAAs initially present in the sample tended to decrease during storage (Leivadara and others 2008). A study of 182 different bottled waters sold in Canada showed that 12 contained chloroform and 4 contained dichlormethane in concentrations up to 70 and 97 μg/L, respectively (Page and others 1993). Dichlormethane is considered a contaminant and not a disinfection by-product though it can be formed during disinfection with chlorine. Al-Mudhaf and others (2009) found THMs in 48% of 71 analyzed bottled water brands purchased in Kuwait. The main THM was bromoform followed by chloroform with maximum concentrations of 37.5 and 1.9 μg/L, respectively. Ikem (2010) found that bottled water from Central Missouri contained fewer THMs compared to tap water. The highest total THM concentration in bottled water was 18.1 μg/L, whereas tap water had concentrations up to 322 μg/L. Most bottled waters had THM concentrations below 1 μg/L.

Other types of contaminants found in bottled waters are benzene and toluene (Page and others 1993; Fayad and others 1997; Ahmad and Bajahlan 2007). Benzene was detected in 1 of the 182 Canadian bottled waters, and toluene was detected in 20 bottles, with maximum concentrations of 2 and 63 μg/L, respectively (Page and others 1993). Fayad and others (1997) also detected benzene in Greek bottled water. Al-Mudhaf and others (2009) found toluene, ethylbenzene, xylene, and naphthalene in 5.3% to 23.0% of 71 analyzed bottled water brands purchased in Kuwait, whereas maximum 2 of the bottles contained trimethylbenzene, 1,2-dichloropropane, trichloroethene, iso-propylbenzene, and 1,3-dichlorobenzene.

Polychlorinated biphenyls (PCBs) may be found in bottled waters in low concentrations. Salinas and others (2010) found that the average total PCB concentration in 6 brands of bottled water collected in Mexico City every month for 1 y was between 0.035 and 0.067 μg/L, which is below the Mexican threshold limit for PCBs in drinking water of 0.5 μg/L. However, considerable variations in concentrations of some PCBs were observed during the 1-y of monitoring, and a concentration of 0.58 μg/L was observed for one brand on one sampling occasion (Salinas and others 2010).

Several researchers have analyzed for selected pesticides (Mathur and others 2003; Turgut and Gokbulut 2008; Díaz and others 2009). Díaz and others (2009) reported total concentrations of 0.29 to 0.46 μg/L of selected organochlorine pesticides in 6 brands of Mexican bottled waters. The highest concentration of single compounds was detected for different isomers of hexachlorocyclohexane (HCH), which ranged between 0.012 and 0.152 μg/L. The aromatic dichlorodiphenyldichloroethylene (DDE), dichlorodiphenyldichloroethane (DDD), and dichlorodiphenyltrichloroethane (DDT) were detected in concentrations up to 0.06 μg/L and similar concentrations of cyclodienes were found. Indian bottled water has been shown to have a high content of pesticides. Eight organophosphorus and 12 organochlorine pesticides were analyzed in 17 brands of bottled water sold in India, which included the 5 most popular brands (Mathur and others 2003). Total concentrations up to 0.011 and 0.051 mg/L were determined for organochlorines and organophosphorus pesticides, respectively, while only one brand did not contain any of the analyzed pesticides. The organophosphorus pesticides observed in most samples and in the highest concentrations were chlorpyrifos and malathion, whereas for organochlorines they were γ-HCH and DDT. Turgut and Gokbulut (2008) found different organochlorine insecticides in 12 of 14 analyzed brands of Turkish bottled water. The highest pesticide concentration, 0.014 μg/L, was found for endrin aldehyde.

Health Risks and Benefits of Bottled Water

The safety of bottled water is determined by its characteristics and content of chemical and microbiological substances. A high content of certain essential elements may be beneficial for health, whereas potentially toxic elements above a certain concentration may constitute a chemical safety problem. High contents of organic carbon are normally only found in surface waters or upper ground water reservoirs. Such waters are also more likely to be influenced by urban, agricultural and industrial activities. High organic carbon contents may therefore warrant further analysis of the water. However, other types of organic contaminants such as plasticizers, pesticides, and THMs can constitute a food safety problem at lower concentrations (Page and others 1993; Ahmad and Bajahlan 2007; WHO 2008; Al-Mudhaf and others 2009; Díaz and others 2009). Studies investigating the chemical quality of tap and bottled water have shown that some bottled waters had better quality than tap water, whereas for other bottled waters the opposite was true (Lalumandier and Ayers 2000; Saleh and others 2001).

Our review of 40 articles reporting on the inorganic content of 1935 bottled waters showed that some contained potentially toxic element or ion concentrations that exceeded the EU or WHO limit values for drinking water (Table 4). An example was As, which was determined in concentrations above the limit value (10 μg/L) in 6 of 12 studies and had a maximum concentration of 49 μg/L. Other elements and ions, which were observed in concentrations above the limit value for drinking water were Ba, Cd, Cr, Hg, Mn, Ni, Pb, Se, CN, F, NO2, and NO3 (Table 4). Despite the fact that concentrations of inorganic solutes were observed above the limit values for drinking water, most of the studied bottled water brands contained concentrations below the limit values and in this respect were safe for drinking purposes.

Table 4. EU drinking water (EU-D), EU mineral drinking water (EU-M), and WHO threshold limits for potentially toxic elements and ions (μg/L) in bottled water (EC 1998, 2003; WHO 2008). Observed concentration range for bottled water and number of studies where the limit values are exceeded out of total number of studies
  Limit value 
   Studies above limit
 RangeEU-DEU-MWHOvalues/total studies
As0.006 to 491010106/12
Ba0.02 to 200010007001/13
Cd0.0006 to 245334/21
Cr0.001 to 1025050502/19
Cu<0.01 to 761200010000/18
Hg<0.005 to 751.01.063/9
Mn<0.005 to 11505004002/16
Ni<0.01 to 4202020704/14
Pb<0.001 to 741010105/17
Sb0.001 to 55.05.0200/11
Se<0.01 to 2931010104/13
CN<LOD-6000.070.050.071/2
F6 to 84001.55.01.58/19
NO2<30 to 101000.500.10.21/5
NO3<40 to 1290005050502/13

There are indications that high drinking water concentrations of Ca and Mg may be beneficial to health. About 30 epidemiological national or regional studies performed around the world since 1957 have found a significant inverse relationship between rates of mortality from cardiovascular disease (CVD) and the hardness and calcium and/or magnesium concentrations of the local drinking water (Shaper and others 1980; Lacey and Shaper 1984; Marier 1986; Derry and others 1990; Rylander and others 1991; Sakamoto and others 1997; Yang 1998; Rubenowitz and others 1999; Sauvant and Pepin 2002). The studies are too numerous to be just a result of coincidence or confounding factors that are not chemically or physically related to the water hardness or content of Ca or Mg, even though they do not prove cause-and-effect or offer information regarding the mechanisms responsible for the relationship. Two possible hypotheses have been proposed. The 1st is that Ca, and especially Mg, may play a protective role in the cardiovascular system and that water is a good source of these minerals compared to other dietary intakes (Couzy and others 1995; Costi and others 1999; Rylander and Arnaud 2004). The 2nd hypothesis is that soft water with low calcium and magnesium content is more corrosive to the plumbing system and, hence, that soft water may dissolve Cd, Pb, and other potentially toxic elements from the tap water plumbing system which may further result in damage to the cardiovascular system through long-term exposure (Michaels and others 1991; Møller and Kristensen 1992; Schwartz 1995; Nishijo and others 2009). If the 1st hypothesis regarding the protective role of Ca and Mg is true, a similar relation must be expected for bottled water, especially since some bottled waters have very high concentrations of Ca and Mg. The concentrations of Ca, Mg, and/or the hardness have also been found to correlate negatively with death caused by other diseases such as type 1 diabetes mellitus; decreased bone density in elderly women; and esophageal, stomach, colon, prostate, rectal and breast cancers (Couzy and others 1995; Sakamoto and others 1997; Yang and others 1998; Yang and Hung 1998; Aptel and others 1999; Costi and others 1999; Yang and others 1999a, 1999b, 1999c; Yang and others 2000a, 2000b; Zhao and others 2001). The review of inorganic constituents in bottled water has shown that Ca and Mg concentrations may range from below the detection limit to up to 730 and 450 mg/L, respectively (section “Inorganic constituents”). Consequently, the potential health benefits of Ca and Mg in bottled water will depend very much on the brand and consumption of bottled water (Table 2).

Fluoride in drinking water may be both beneficial and damaging for dental health; fluoride concentrations of about 1 mg/L protect the teeth from caries whereas concentrations above 4 mg/L may lead to the development of severe fluorosis in children, resulting in cracking and pitting of teeth (Driscoll and others 1983; Grobler and others 1986; Burt 2002). The most important sources for human fluoride intake are drinking water, toothpaste, and other fluoride-containing dental products such as mouthwash, tablets, and gels (O'Mullane and Holland 1983). Fluoride concentrations in bottled water range from 0.006 to 8.4 mg/L (Table 2). Bottled water fluoride content may therefore be beneficial or harmful to the teeth.

Natural mineral water with high Fe content may be a good substitute for traditional Fe supplements. McKenna and others (2003) and Halksworth and others (2003) found that Spartone, a natural mineral water with a high Fe content (200 mg/L), helped prevent Fe deficiency and that this brand of mineral water did not have the side effects of normal Fe supplements that cause many patients to stop the treatment. Normally, Fe content is much lower in bottled water, between 0.07 and 4500 μg/L (Table 3).

As bromate is a potential carcinogen to humans and DBP of ozone treatment of bromide containing water ozonation of bottled water can be a potential health hazard (Nyman and others 1996; Liu and Mou 2004.) Bromate have been detected in concentrations above the WHO provisional guideline value of 10 μg/L in 9 of 18 analyzed commercial bottled waters by Nyman and others (1996) and Warner and others (1996) found similar results for bottled spring water.

It is difficult to evaluate the effect of organic contaminants on bottled water chemical safety since only very few studies of organic contaminants of bottled water have been carried out. In general, the few studies undertaken show that the content of disinfection by-products, plasticizers, and other leached compounds in bottled waters are below the WHO and EU limit values for drinking water (EC 1998; WHO 2008; Leivadara and others 2008; Al-Mudhaf and others 2009; Ikem 2010). Some bottled waters do contain concentrations of a few compounds above the limit values for drinking water. Page and others (1993) found dichloromethane in bottled water with a maximum concentration of 97 μg/L, thus exceeding the WHO limit value of 20 μg/L (WHO 2008). Vinyl chloride and styrene, which may leach from bottle material, were also found in bottled water in concentrations that may be of health concern (Fayad and others 1997; Ahmad and Bajahlan 2007). The maximum vinyl chloride concentration was 0.6 μg/L, which is above the EU limit value of 0.5 μg/L but below the WHO limit value of 3 μg/L. Styrene has been found in concentrations up to 46.4 μg/L in water bottled in polystyrene, which is above the WHO limit value of 20 μg/L (Ahmad and Bajahlan 2007; WHO 2008; Al-Mudhaf and others 2009). Of the investigated volatile organic compounds, only benzene exceeded the limit values. A maximum bottled water benzene concentration of 2 μg/L was detected by Page and others (1993), which is between the EU limit value for benzene in drinking water of 1 μg/L and the WHO limit value of 10 μg/L (EC 1998; WHO 2008). The EU limit value for different pesticides in drinking water is 0.1 μg/L and the total pesticide concentration should not exceed 0.5 μg/L (EC 1998). Díaz and others (2009) and Turgut and Gokbulut (2008) found pesticide concentrations above these limit values. However, this does not necessarily constitute a food safety risk, as the EU limit values for pesticides in drinking water are not based on toxicity and health risk considerations. Even though the concentrations of individual compounds with endocrine-disrupting effects such as plasticizers in bottled water are generally too low to have an effect on their own, there are indications that the mixture of different endocrine disruptors in bottled water may have a hormonal effect and that endocrine disruptors are leached from PTE bottles (Wagner and Oehlmann 2009).

Flavor of Bottled Water

Flavor is the overall sensory impression of ingested food and it is determined by taste, odor, and trigeminal sensations, also called mouth-feel, which is caused by chemicals but also by touch, pressure, and temperature (Lundström and others 2011). In most cases, an untrained person cannot distinguish between taste, odor, and mouth-feel stimuli. An example is the metallic odor of Fe which is part of the flavor sensation of Fe in water, to identify this as a retro-nasally odor tasting with clamping of the nose or a stream of air though the nose cavity has to be performed, since these conditions remove the metallic sensation of Fe it can be identified as a retro-nasally odor (Epke and Lawless, 2007; Epke and others 2009). The overall descriptor flavor is used when a sensation has not been identified more specifically as a taste, odor or mouth-feel (Lundström and others 2011). In many studies of water flavor a distinction between taste, odor, and mouth-feel have often not been made (Fong and others 2001; Lou and others 2007). Consumer complaints often refer to bad taste qualities; however, one should be aware that such problems are often due to odors or a combination of taste, odor, and mouth-feel sensations (Hettinger and others 1990; Young and others 1996; Suffet and others 1999). Here, we will use the term flavor as a general term for all sensory impressions perceived when water is ingested and when no distinction have be made between the different types of stimuli. However, during the 1980s the American Water Works Assn. Research Foundation funded a research project aiming at identifying odor categories for known odors in drinking water (Suffet and others 1995). This work was further developed into the taste and odor wheel (Mallevialle and Suffet 1987). The taste and odor wheel is a schematic presentation of all known flavor problems for drinking water, which serves to give a common language for sensory panels and water practitioners, and to provide the water industry with current knowledge of common reasons to flavor problems. The taste and odor wheel, which has become a widely accepted and applied tool in the water sector contains 8 odor categories, 4 basic taste categories, and a mouth-feel and nose-feel category (Figure 4 inner circle). The common flavors of each category, which have been defined by a sensory panel are given in the outer circle of the taste and odor wheel. Reference standards used by sensory panels are presented outside the 2 circles together with chemicals which have been identified to cause taste and odor problems (Figure 4). A flavor profile analysis (FPA) method, which uses standards to characterize tastes and odors of drinking water has been developed. This method named “Method 2170 flavor profile analysis” proposes conditions for sensory analysis with a trained panel (Franson and others 1995). By applying FPA and the standards and knowledge from the taste and odor wheel a trained sensory panel will be able to identify tastes, odors, mouth- and nose-feel stimuli and their intensity in drinking water (Mallevialle and Suffet 1987).

Figure 4.

The drinking water taste and odor wheel. Inner circle: Taste, mouth-feel and nose-feel categories. Outer circle: Common tastes and odors detected in drinking water by sensory panels. Outside the circles: Chemicals that have been confirmed as the causes of taste and odor problems in drinking water (reprinted with permission from Suffet and others (1999)).

Taste, odor, and mouth-feel are distinguished by different oral receptors and neural processing. Taste is caused by receptors on the tongue in response to variations in saliva composition. Similar receptors have been found at the epithelium of the palate, oropharynx, larynx, and the upper oesophagus, although their function has so far not been identified (Doty and others 1978; Lundström and others 2011). The basic tastes of sour, sweet, salty, and bitter are relevant for drinking water as described in the taste and odor wheel (Suffet and others 1999; Figure 4). Mouth-feel is caused by tactile stimulation when water comes into contact with the mouth. Known mouth-feel sensations for drinking water are given in the taste and odor wheel (Figure 4). Mouth-feel sensations can be caused by both organic and inorganic components in the water. The sensory impression of water is influenced not only by taste, odor, and trigeminal sensations; other factors such as thirst will also influence sensory perception. For example, the pleasant sensations of hydration decrease as an individual has consumed more water (Rolls and others 1983).

In the following sensations of drinking will be treated in general, as the principles are the same for tap and bottled water. However, focus will be on taste and odor causing elements and compounds, which are relevant for bottled water.

Traditionally, water of good sensory quality has been considered as water free from off-flavors and research within this area has mainly been concerned with understanding and solving off-flavor problems (Khiari 2004; Whelton and others 2007). Little focus has been given to the preferred flavor of water. The relationship between the chemical composition, sensory properties, and preferences that may lead to loyalty to a specific bottled water brand is therefore not well established (Whelton and others 2007). Nonscientific sensory guides establish mouth-feel determined by the carbon dioxide content as the most important factor when selecting a bottled water, and taste determined by the mineral content as the 2nd most important (Chapelle 2005; Mascha 2006).

Flavor of single salts

A few studies have investigated the flavor of drinking water in relation to concentrations of inorganic compounds, and most of the present knowledge is based on studies of flavor and flavor thresholds for single salts dissolved in distilled water (Zoeteman 1980; Tordoff 1996; Lawless and others 2003; Koseki and others 2005). Such studies have been carried out for the major alkaline earth cations Ca and Mg, the alkali metals Li, Na, K, Ru, and Cs, and trace metals such as Fe, Zn, and Cu (Cohen and others 1960; Zoeteman 1980; Murphy and others 1981; Tordoff 1996; Lawless and others 2003; Koseki and others 2005). Through these studies it has become widely accepted that cations are primarily responsible for the flavor properties of salts, whereas anions may modify the intensity of the flavor; for example, the intensity of flavor attributed to Zn will decrease as the anion is changed from sulfate to nitrate (Cohen and others 1960; Bruvold and Gaffey 1969b).

Lawless and others (2003) used a sensory panel of 28 persons to investigate the flavor of CaCl2, MgCl2, and MgSO4 salts in distilled water at concentrations of 400 to 4000 mg/L and 200 to 2000 mg/L for Ca and Mg, respectively. They found that both cations are primarily characterized by bitter and salty tastes and to a lesser extent by sour, metallic, astringent, and irritative sensations. Table 5 shows that the salty, sour, and bitter tastes are common for Ca, Mg, and Na. The Ca flavor detection threshold has been found to range between 40 and 800 mg/L depending on the anion of the added salt (Table 5). However, some persons are more sensitive to the flavor of Ca and can detect it at concentrations much below the average of a taste panel. For instance, Tordoff (1996) found an average flavor detection threshold of 800 mg/L for the PO43− salt, whereas the most sensitive panel members could detect Ca at 7 mg/L (Tordoff 1996). Using an untrained sensory panel Koseki and others (2005) found that the sensory quality of bottled mineral water decreased when more Ca was added. At lower Ca concentrations of 10 to 20 mg/L the water was perceived as sweet and tasted significantly better compared to water with a Ca concentration of 100 mg/L, which resulted in a bitter taste. Zoeteman (1980) found that NaHCO3 and CaSO4 in distilled water had an optimal sensory quality at Na and Ca concentrations of about 35 and 80 mg/L, respectively, whereas the flavor of dissolved NaCl and CaCl2 did not differ significantly from distilled water in concentrations up to 115 and 64 mg/L Na or Ca, respectively, but the sensory quality was impaired at higher concentrations (Zoeteman 1980). The study also showed that Cl salts of Mg, Ca, and Na were perceived as offensive at cation concentrations of 12, 105, and 184 mg/L, respectively. Hydrogen carbonate salts were offensive at similar concentrations, whereas much higher concentrations of SO4 salts were needed to produce an offensive flavor. Similar results were found by Bruvold and Gaffey (1969a) who studied the sensory quality of 1000 and 2000 mg/L solutions of different salts. They found that the sensory quality decreased in the order: NaHCO3, CaSO4, Na2SO4, MgSO4, NaCl, CaCl2, MgCl2, and Na2CO3. Bruvold and Gaffey (1969b) found the flavor intensity of Na to be lowest with HCO3 and SO4 salts, higher with Cl salts, and highest with CO3 salts. The Ca, Mg, and Na concentrations in bottled water analyzed were between the detection limits and 729, 447, and 1910 mg/L, respectively (Table 2). However, most bottled waters have concentrations of Ca, Mg, and Na below 100 mg/L (Figure 2). As the flavor detection thresholds range between 40 and 800 mg/L, 28 and 290 mg/L, and 11 and 216 mg/L for Ca, Mg, and Na, respectively (Table 5), these cations are likely to influence the flavor of bottled water and will decrease the sensory quality at higher concentrations (Zoeteman 1980).

Table 5. Taste descriptors of cations and taste detection thresholds (TDT) in units of mg/L
CationTaste qualityTDTReferences
AlAstringent12 (SO42−)Whipple 1907;
   Schiffman and others 1992.
CaBitter800 (PO43−)Lawless and others 2003;
 Salty177 (SO42−)Tordoff 1996;
 Sour90 (Cl)Whipple 1907;
 Metallic440 (OH)Lochart and others 1955.
 Astringent300 (Cl) 
 Irritative125 (Cl) 
 Sweet40 (CO32−) 
CsBitter Murphy and others 1981.
 Sour  
 Salty  
 Sweet  
CuBitter2.4 (SO42−)Béguin-Bruhin and others 1983;
 Astringent10 (SO42−)Zacarías and others 2001;
 Sour29.8 (SO42−)Lawless and others 2004, 2005;
 Salty0.4 to 0.8 (SO42−)Lim and Lawless 2005;
 Metallic0.5 (SO42−)Yang and Lawless 2005;
  0.52 (Cl)Cuppet and others 2006;
  6.6 (Cl)Epke and Lawless 2007;
  25.2 (Cl)Schiffman 2000;
   Whipple 1907.
FeMetallic3.6 (Cl)Yang and Lawless 2006;
 Bitter49 (Cl)Cohen and others 1960;
 Astringent1.7 (SO42−)Hettinger and others 1990;
 Sour1.8 (SO42−)Lim and Lawless 2006;
 Sweet7.9 (SO42−)Schiffman 2000;
 Salty5.5 (SO42−)Viñas and others 1998;
  3.4 (SO42−)Epke and Lawless 2007;
  8.8 (OH)Whipple 1907.
KBitter340 (Cl)Murphy and others 1981;
 Sour Lochart and others 1955.
 Salty  
 Sweet  
LiBitter Murphy and others 1981.
 Sour  
 Salty  
 Sweet  
MgBitter290 (HCO3)Lawless and others 2003;
 Salty121 (SO42−)Lochart and others 1955;
 Sour100 (SO42−)Whipple 1907.
 Metallic141 (NO3) 
 Astringent128 (Cl) 
 Irritative135 (Cl) 
  28 (CO32−) 
  34 (CO32−) 
  75 (PO43−) 
NaBitter11 (CO32−)Murphy and others 1981;
 Sour194 (Cl)Whipple 1907;
 Salty216 (NO3)Wiesenthal and others 2007.
 Sweet129 (SO42−) 
  252 (Cl) 
RbBitter Murphy and others 1981.
 Sour  
 Salty  
 Sweet  
ZnBitter18 (SO42−Keast 2003;
 Sour25 (Cl)Cohen and others 1960.
 Salty22 (NO3) 
 Stinging  
 Astringent  

The tastes of the alkali metals Li, K, Rb, and Cs are characterized as salty, bitter, sour, and sweet when present as I, Cl, or Br salts in the concentration range 26 to 750 mg/L, 156 to 4222 mg/L, 340 to 9230 mg/L, and 530 to 14350 mg/L, respectively (Murphy and others 1981; Table 5). However, half of the panel members were not able to detect the alkali metals at the lowest concentrations. The bitterness increased with increasing atomic weight of the metal. Only K can be expected to affect the flavor of bottled water since the maximum concentrations of Li, Rb, and Cs found in bottled water (Table 2 and 3) were below the lower concentrations used in this study.

Trace elements may have a strong effect on the flavor of water when they are present above the flavor detection threshold. Iron has been the focus of many sensory studies since it has high flavor intensity and foods are often fortified with Fe due to the high occurrence of Fe deficiency (Mehansho 2006; Khoshgoftarmanesh and others 2010). Sensory studies of the flavor of Fe are complicated by precipitation of Fe(III) when Fe(II) becomes oxidized to Fe(III). Different methods have been used to overcome this problem such as removing oxygen from the solution, lowering the pH (reducing the rate of oxidation), and preparing fresh solutions before tasting sessions (Hettinger and others 1990; Viñas and others 1998; Yang and Lawless 2006; Lim and Lawless 2006; Epke and Lawless 2007). The flavors of different ferrous salts have been found to be predominantly metallic, bitter, and astringent and to a lesser degree sour, sweet, and salty (Cohen and others 1960; Yang and Lawless 2006) (Table 5). Sensory qualities of ferrous salts are complex and consist of several basic tastes, odors, and mouth-feel (Hettinger and others 1990; Lawless and others 2004; Lim and Lawless 2006; Stevens and others 2006). Lim and Lawless (2006) found that the basic tastes bitter, sweet, and sour of ferrous salts increased with increasing Fe concentrations whereas the metallic flavor decreased. The metallic flavor of Fe is decreased by nasal occlusion consistent with the hypothesis that solutions of Fe salts generate volatile lipid oxidation related aldehydes and protein-carbonyls in the mouth that are perceived retro-nasally as a metallic odor (Epke and Lawless 2007; Epke and others 2009; Ömür-Özbek and others 2012). Hettinger and others (1990) found that ferrous sulfate could be detected as a metallic flavor at a concentration of 55 mg/L (1 mM) with an open nose, but with a clamped nose it was flavorless, indicating that detection by retronasal odoris more sensitive than the basic tastes. A similar tendency but with lower detection thresholds was found by Epke and Lawless (2007) who found that ferrous sulfate could be detected at Fe concentrations of 1.7 and 9.0 mg/L with an open nose and clamped nose, respectively, which is the same tendency as seen for ferrous chloride though the detection thresholds were slightly higher. The reported mean flavor detection threshold for iron ranges from 1.7 to 49 mg/L for different ferrous salts (Table 5). This wide range of determined flavor detection thresholds may be due to different sensory or calculation methods, and the application of different Fe salts such a sulfate and chloride. However, the flavor detection threshold may vary by several orders of magnitude between people with different sensitivities to iron flavors and it is therefore likely that the variation in the flavor detection threshold is at least partly a result of different panel member compositions. Lim and Lawless (2006) found that the flavor detection threshold for Fe salts differed 611 times between the most sensitive and least sensitive person in their panel, whereas Cohen and others (1960) found this difference to be as high as 6400. Results by Young and others (1960) further indicated that there can be large variations in the flavor sensitivity of the same person between different days. Cohen and others (1960) found that the flavor detection threshold for Fe was lower in tap water compared to distilled water. An Fe concentration of up to 4.5 mg/L has been found in bottled water; however in most bottled waters the Fe content will be much lower (Table 3). This shows that the Fe content of bottled water will only influence the sensory quality for very sensitive individuals and Fe taste and odor problems for bottled water is much less pronounced compared to tap water where Fe release from pipes can cause such taste and odor problems (Mallevialle and Suffet 1987; Suffet and others 1995).

The flavor of copper (Cu) has been described as both bitter, astringent, sour, salty, and metallic (Table 5). As for Fe retronasal odor seems to be a part of the Cu flavor and lipid oxidation and formation of aldehydes and carbonyls also play a role in the Cu taste though it is less significant compared to the taste of Fe (Ömür-Özbek and others 2012). Epke and Lawless (2007) found taste detection thresholds Cu from CuSO4 in distilled water to be 0.49 and 1.56 mg/L with open and clamped nose, respectively. However, in other studies a difference between detection thresholds for Cu with open and clamped nose has not been observed (Zacarías and others 2001; Epke and others 2009). Epke and Lawless (2007) indicate that this may be due to the lower and slower oxidation of lipids in the mouth cavity by Cu compared to Fe(II). Flavor detection thresholds for Cu in distilled water were found to be in the range of 0.4 to 29.8 mg/L (Table 5). Studies have shown that free and complexed Cu ions can be readily sensed while particulate Cu added only slightly to the Cu flavor as only part of it is solubilized in the saliva (Cuppett and others 2006; Hong and others 2010a, 2010b; Hong and Kim 2011). A couple of studies have shown lower flavor detection thresholds for Cu in distilled water compared to mineral water (Cohen and others 1960; Zacarías and others 2001; Cuppett and others 2006). Béguin-Bruhin and others (1983), on the other hand, found lower flavor detection thresholds in mineral water compared to distilled water of 0.8 to 1.0 mg/L and 2.4 to 3.2 mg/L, respectively. However, the lower flavor threshold detection limit in mineral water was attributed to the stabilization of Cu2+ in the sample by carbon dioxide (affecting Cu speciation). The Cu content of bottled water can be up to 0.8 mg/L, a concentration just above or below the flavor detection threshold (Table 3) (Béuing-Bruhin and others 1983; Cuppett and others 2006). Copper is therefore not likely to have an influence on the flavor of most bottled waters and only for the most sensitive persons, but flavor problems can occur with municipal water where the water may receive Cu from the piping system (Mallevialle and Suffet 1987; Suffet and others 1995).

Zinc (Zn) salts at concentrations of 327 to 3270 mg/L (5 to 50 mM) result in bitter, sour, salty, tingling/stinging, and astringent sensations, with astringent being most dominant (Keast 2003) (Table 5). Further, Zn seems to inhibit the sweet taste of other substances. Cohen and others (1960) found Zn flavor detection thresholds in distilled water between 18 and 25 mg/L (Table 5). The detection threshold increased in the following order with respect to the salt anion: sulfate < nitrate < chloride and it was slightly higher in tap water compared to distilled water. Zinc concentrations in bottled water are below 1 mg/L and, hence, Zn should have no influence on the flavor of bottled water (Table 3).

Flavor of salts in complex mixtures

In more complex solutions such as spring, mineral, and tap water, the flavor will not necessarily be the sum of the flavors of single salts. A few studies have focused on the flavor of drinking water as a result of the total inorganic content or complex mixtures of salts. These studies used tap water from consumers’ taps or artificial waters produced from distilled water plus a range of salts. Through these studies the relation between the sensory quality and content of the major inorganic components Ca, Mg, Na, HCO3, Cl, NO3, SO4, and TDS was investigated (Bruvold and Ongerth 1969; Bruvold and Gaffey 1969a, 1969b; Bruvold and Daniels 1990; Fong and others 2001; Lou and others 2007; Teillet and others 2010a, 2010b). Early research indicated that the sensory quality of water gradually improves with increasing salt concentrations when TDS is below 50 mg/L; at higher salt concentrations the quality will decrease with increasing TDS, but water with TDS up to 450 mg/L is considered good-tasting water (Bruvold 1975; Bruvold and Daniels 1990). The decreased sensory quality of water with low mineral content has been shown in other studies where waters with a medium mineral content was preferred over waters with very low mineral content such as distilled water (Falahee and MacRae 1995; Vann 2004). Fong and others (2001) found that distilled water had a metallic and sweet flavor. Bruvold and Ongerth (1969) found similar results, as reported earlier. However, they divided the quality of drinking water into 5 sensory quality categories with the following ratings: excellent for water with TDS up to 313 mg/L, good-tasting with TDS between 314 and 638 mg/L, fair-tasting with TDS between 639 to 896 mg/L, poor-tasting with TDS between 897 to 1129 mg/L, and unacceptable with TDS above 1130 mg/L. Bruvold (1977) reported results from a similar study, but in this study the TDS had a stronger negative influence on the sensory quality and the water was rated unacceptable at TDS above 634 mg/L. On the other hand, Lou and others (2007) found that changes in the TDS from 214 to 384 mg/L of drinking water were insufficient to be detected by the untrained taster. Teillet and others (2010a, 2010b) carried out a sensory evaluation of 6 bottled waters and 6 tap waters with TDS ranging from 19 to 2660 mg/L. They found a bitter and metallic flavor of water with low TDS (19 to 65 mg/L), a neutral and fresh flavor of water with medium TDS (274 to 329 mg/L), and a salty flavor for high TDS. In conclusion, these studies all showed that TDS of about 200 to 400 mg/L will result in good-tasting water. The sensory quality will decrease at higher and lower TDS, but the acceptability of such water will depend on the characteristics of the salts, the sensory panel, and the conditions under which the tasting is carried out. Even though TDS has a significant influence on the flavor of water, small differences in salt concentrations will often not be detected by consumers. In bottled water, the TDS has been found to range between 5 and 3400 mg/L (Table 2). The flavor of bottled water should therefore range from unpleasant metallic, bitter, and/or sweet for waters with low mineral content, to pleasant and neutral sensory quality at medium salt concentrations (about 100 to 400 mg/L), and a salty unpleasant to unacceptable flavor at high salt concentrations (Bruvold and Ongerth 1969; Bruvold 1975; Bruvold and Daniels 1990; Fong and others 2001; Teillet and others 2010a, 2010b). However, why are bottled waters produced with a flavor that is perceived as unpleasant or even unacceptable and why do people buy such water? Sensory science has shown that during a life span, and especially during nursing years, humans can develop and adopt new taste and odor preferences (Beauchamp and Mennella 2011). Some people may therefore acquire a preferred flavor for drinking water with low or high mineral content. However, new research shows that some consumers who drink bottled water with high TDS content prefer water with medium TDS under blind tasting (Teillet and others 2010b).

The influence of salt content on consumer perception of drinking water is changed, diminished, or disguised when other flavor determining factors are varied (Lou and others 2007). Other factors such as carbonation and temperature may have a significant influence on the sensory experience of water (Yau and McDaniel 1991; Fong and others 2001; Whelton and Dietrich 2004). Zellner and others (1988) found that cold (0 to 5 °C) water is preferred over water served at room or hot temperatures. However, the preference for cold water was partly explained by the panel expecting water to be served at this temperature. When panel members were told they would be served hot water they liked it more compared to tastings where they were not informed about the serving temperature (Zellner and others 1988). Carbonation significantly alters the sensory perception of water; this is not only a result of the physical sensation made by bubbles but is also caused by the chemosensory response of receptor cells on the tongue and is experienced as a sour, burning, and tingling sensation (Fong and others 2001; Yarmolinsky and others 2009; Chandrashekar and others 2009). A study by Fong and others (2001) indicates that low salt content is preferred for still water, whereas for carbonated water a higher salt content up to a certain limit results in the best sensory quality. This study showed that still water with TDS of about 490 mg/L was rated as being of good sensory quality, whereas water with lower TDS resulted in the rating of very good. However, carbonated water was rated as very good at a TDS of about 620 mg/L, but above 1100 mg/L it was only rated fair.

Off-flavors

Tastes and odors of drinking water have received attention for more than 20 y, especially from the American Water Works Assn. Focus has been on understanding off-flavors and how to avoid them since off-flavors will result in consumer complaints (Khiari 2004). In tap water off-flavors may typically be introduced due to chlorination and microorganisms, which produce odorants, resulting in a musty/earthy odor but for bottled water release of organics from bottling material, ozone disinfection by-products, and lubricants or organic cleaning agents used in the maintenance of bottling equipment are known off-flavor problems (White and others 1991; Villberg and others 1997; Suffet and others 1999; van Aardt and others 2001; Gouviea and others 2007; Richardson and others 2007; Cizková and others 2009). As mentioned the taste and odor wheel was developed in the mid-1980s in an attempt to better understand and handle taste and odor problems and has since then been improved as new knowledge has been obtained (Suffet and others 1999). Though the taste and odor wheel were developed based on knowledge of flavor problems in tap water, for example, the taste category chlorinous, all basic taste categories, mouth- and nose-feel and some of the odor categories are relevant for bottled water also (Figure 4). The chlorinous, bleach or swimming pool odor, which is a known flavor problem for tap water is a result of disinfection with free chlorine or chloramines. Flavor detection thresholds of chlorine have been found to be 0.14 mg/L Cl2 (Puget and others 2010), 0.16 mg/L Cl2 (Bryan and others 1973), 0.24 mg/L Cl2 (Krasner and Barret 1984), 0.10 mg/L Cl2 (McDonal and others 2009) and 0.05 mg/L Cl2 (Piriou and others 2004) whereas hypochlorous acid, hypoclorite, and monochloramine have flavor detection thresholds of 0.24, 0.30, and 0.09 mg/L Cl2, respectively (Krasner and Barret 1984; Piriou and others 2004). Chlorine flavor detection thresholds determined by untrained panels may be much higher around 1 mg/L Cl2 especially for panels accustomed to chlorinated water (Mackey and others 2004; Piriou and others 2004). Such high concentrations of chlorious compounds are not likely to occur in bottled water.

Traditional bottled water treatment methods such as filtration, reverse osmosis, and ozonation will be able to remove some flavor problems but will not always be sufficient and will in some cases even increase the problem or create new odor problems (Lalezary-Craig and others 1988; Glaze and others 1990; Suffet and others 1995). Ozone disinfection is known to result in low molecular weight nonhalogenated aldehydes as described earlier (Richardson and others 2007; Huang and others 2009; Qi and others 2009; Kim and Hyun 2012). Such compounds are volatile and have low odor thresholds in air and water (Devos and others 1990; Suffet and others 1995). Suffet and others (1995) compiled odor description and thresholds for low molecular weight aldehydes and found that they are characterized as fruity, orange, or citrus like at low concentrations and have odor threshold concentrations in the 0.1 to 50 μg/L range. However, van Aardt and others (2001) found a much higher odor threshold of 167 μg/L for acetaldehyde in spring water.

Leaching of aldehydes or ketones from bottle material may also cause off-taste problems for bottled water (Song and others 2003). Acetaldehyde concentrations in mineral water due to leaching from PET bottles have been shown to be as high as108 μg/L which is well above the odor threshold of 4 μg/L (Gilli and others 1990; Mutsuga and others 2006). Bottled water has been reported to have an undesirable plastic taste which increases with increasing ozone concentration (White and others 1991; Song and others 2003). However, increasing the holding time from production of bottles until filling with ozonated water decreased the plastic taste as did the application of antioxidants to the water indicating that antioxidants protect the bottle material against oxidation by ozone (White and others 1991).

A medicinal off-flavor of mineral water has both been detected in water stored in glass and plastic bottles (Strube and others 2009). The cause of the medicinal flavor was identified to be 2-iodo-phenol and 2-iodo-4-methylphenol and which have odor thresholds 0.3 and 0.01 μg/L, respectively. The formation of iodophenol in drinking water has also been observed by Dietrich and others (1999) in the presence of phenol and iodine.

Occurrences of musty and earthy flavor problems in drinking water are often linked to the use of surface water (Bae and others 2007). Such musty and earthy odors are most often caused by the presence of the terpenoids geosmin (trans-1,10-dimethyl-trans-9-decanol) and MIB (2-methylisoborneol) (Watson 2004, 2010). The major producers of these compounds, under growth conditions where photosynthesis is possible, are cyanobacteria or some species of cyanobacteria (Watson 2004; Jüttner and Watson 2007; Watson and others 2008). Other producers of geosmin and MIB are actinomycetes, fungi (especially Penicillium) and myxobacteria though it is difficult to determine their importance for drinking water taste and odor problems due to the shortcomings of the nonspecific analytical methods which have been used for detection of these organisms in waters with taste and odor problems (Jüttner and Watson 2007; Giglio and Others 2008). Other components that may give drinking water a musty odor are 2,4,6-trichloroanisole and the methoxypyrazines 2-isopropyl-3-methoxypyrazine and 2-isobutyl-3-methoxypyrazine which are primarily produced by fungi and bacteria including actinomycetes (Jensen and others 1994; Piriou and others 2001; Watson 2004; Peter and von Gunten 2009). Geosmin, 2-isobutyl-3-methoxypyrazine, 2-isopropyl-3-methoxypyrazine, and MIB have been found to have very low odor detection thresholds in the range of 0.4 to 10 ng/L (Young and others 1996; Piriou and others 2009).

Sulfate reduction in groundwater reservoirs may result in a rotten-egg odor of the extracted drinking water due to the formation of hydrogen sulfide (Suffet and others 1996). Hydrogen sulfide has an estimated odor and taste detection threshold of about 0.05 to 0.1 mg/L (WHO 2003). However, in bottled water this flavor problem is removed by degassing and ozonation.

Young and others (1996) investigated the flavor and flavor detection threshold of different potential anthropogenic organic contaminants in still mineral water including 22 pesticides, 13 phenolic and anisole compounds, and 18 other organic compounds such as benzene, different chlorobenzenes and toluenes. Phenolic and anisole compounds were also detected at low concentrations of <0.05 to 100 μg/L, whereas chloroform and different benzenes and toluenes were detected at concentrations of 11 to >1000 μg/L. Chloroform and toluene concentrations of up to 70 and 63 μg/L have been found in bottled water, respectively (Page and others 1993), which is more than a factor of 10 below the concentration resulting in a chlorinous chemical flavor (Young and others 1996). The pesticides have little flavor in water and most pesticides investigated by Young and others (1996) could not be detected at the applied concentrations. Only the 2 pesticides carbaryl and MCPA could be detected at concentrations below which they are toxic (Young and others 1996). In conclusion these anthropogenic compounds are not likely to create bottled water flavor problems.

The odor and intensity of drinking water odorants may change with concentration, mixtures of different odorants, and temperature (Whelton and Dietrich 2004; Watson 2004). Whelton and Dietrich (2004) found that the intensity of the odorants geosmin and MIB increased with increasing temperature. Due to higher volatilization higher odor intensities was observed at 45 °C compared to 5 and 25 °C, but no difference could be detected between 5 and 25 °C, which is the most relevant temperature range for bottled water. Finally, the flavor of a compound can change with increasing concentration; for example, the flavor of 2,4,5-trichlorophenol has been found to change from fruity to antiseptic with increasing concentration (Young and others 1996).

Consumer Perception of Bottled Water

From this review it is clear that sales of bottled water should not show the observed increase if consumer choices of drinking water are based on rational decisions regarding quality, chemical food safety, costs, and environmental impact. The popular literature offers many speculations about the causes of the increase in bottled water consumption, such as beliefs in superior quality, safety, taste, and health benefits (Doria 2006; Ward and others 2009). However, few scientific studies have focused on why consumers buy bottled water.

A survey in England showed that a majority of interviewed people believe that bottled water has some health benefits compared to tap water but they are not clear about what these benefits are and they consider the health benefits insignificant compared to the benefits of good-quality tap water (Ward and others 2009). The study also showed that beliefs in the health benefits of drinking bottled water are unlikely to have a major influence on the consumption of bottled water for the general population, whereas factors such as convenience, cost, and flavor seem to be more important. Doria and others (2009) carried out a study using structural equation models and generalized linear models to evaluate the influence of different factors on the perceived quality of water and the consumption of tap and bottled water in England and Portugal. They found that the perceived quality of water is strongly influenced by sensory quality and less by the context and risk perception in both countries, although the risk perception had greater influence in Portugal compared to England. Furthermore, consumption of tap water was moderately well described by sensory quality and food safety risk perception, and bottled water consumption is moderately well described by a decrease in tap water consumption. This indicates that sensory quality and risk perception have at least some influence on bottled water consumption and that bottled water is an alternative to tap water and not only a convenient supplement. Taste and odor problems of tap water have been reported as one of the main reasons for consuming other sources of water (Levallois and others 1999; Dupont and others 2010). Taste and odor problems are more prone to occur in countries that use surface water as municipal water due to periodic algael or bacterial blooms. In such countries it is likely that taste and odor problems are the main drivers for buying bottled water (Levallois and others 1999). Countries using surface water often treat municipal water with chlorine, which can cause chlorinous odors. Doria and others (2009) found that chlorine flavor is negatively correlated to the perception of sensory quality and trust in food safety of drinking water. Puget and others (2010) tested the hypothesis that people who choose alternate drinking water over tap water are more sensitive to chlorine flavors compared to tap water consumers. They found similar flavor detection thresholds for consumers and none consumers of tap water. However, none consumers rated the acceptability of water containing chlorine lower than tap water consumers.

Water with off-flavors is often assumed to be of low quality and not to be wholesome (Doria and others 2009). The reason that a purely esthetic parameter strongly influences the perception of water quality might be that it is directly felt by the consumer and is therefore believed to be more trustworthy than secondary information from media or from the water sector which may have special incentives to portray water as being of good quality and beneficial to health. In some cases consumers will choose to treat municipal water with off-flavors before drinking it; in other cases they will chose an alternative water source such as bottled water (Levallois and others 1999).

Conclusion

  • Bottled water consumption has been increasing worldwide every year despite the fact that tap water quality is generally improving and that bottled water is several orders of magnitudes more expensive than tap water.
  • The chemical composition of bottled water is relatively well known, particularly for the major elements and ions, and the concentrations of different elements vary from 2 to 6 orders of magnitude for bottled water.
  • Bottled waters are generally considered safe for drinking purposes, but some contain potentially toxic elements or ion concentrations which exceed the EU or WHO limit values for drinking water.
  • The main organic contaminates of bottled water are low molecular aldehydes and originate from ozonation or leaching from bottled material which may give the water a fruity off-flavor.
  • The major cations in bottled are Ca, Mg, Na, and K; the dominating anions are HCO3, Cl, and SO42−.

Trace elements like Fe, Cu, and Zn may have a strong effect on the sensory quality of water when they are present in quantities above the flavor detection threshold. However, their flavor detection threshold is unlikely to be exceeded in most bottled drinking waters. The sensory impression of a cation can be modified both by the cation concentration and by the presence of other cations or anions, such as complexation reactions. There is a lack of sensory studies examining solutions for spring, mineral, or tap water in which the flavors of salt mixtures are not necessarily the sum of the tastes of the single salts.

Acknowledgment

Financial support from the strategic water research initiative ViVa at the Faculty of Life Sciences, Univ. of Copenhagen, is gratefully acknowledged.

Ancillary