DETECTION OF ADULTERATION OF WITH CANOLA AND OTHER SEED OILS BY REVERSED-PHASE HIGH PERFORMANCE LIQUID CHROMATOGRAPHY

A Thesis Submitted to the College of Graduate Studies and Research in Partial Fulfillment of the Requirements for the Degree of Master of Science

in the Interdisciplinary Science Program of U ni versi ty of Saskatchewan Saskatoon

By Emmanuel Salivaras April, 1992

The author claims copyright. Use shall not be made of the material contained herein without proper acknowledgement, as indicated on the copyright page. COPYRIGHT

In presenting this thesis in partial fulfillment of the requirements for a Postgraduate degree from the University of Saskatchewan, I agree that the Libraries of this University may make it freely available for inspection. I further agree that permission for copying of· this thesis in any manner, in whole or in part, for scholarly purposes may be granted by the professor or professors who supervised my thesis work or, in their absence, by the Head of the Department or the Dean of the College in which my thesis work was done. It is understood that any copying or publication or use of this thes~s or parts thereof for financial gain shall not be allowed without my permission. It is also understood that due recognition shall be given to me and to the University of Saskatchewan in any scholarly use made of any material in my thesis.

Requests for permission to copy or to make other use of material in this thesis in whole or part should be addressed to:

Head, Department of Applied Microbiology and , University of Saskatchewan, Saskatoon, Canada S7N OWO

1 ABSTRACT

Reversed-phase high performance liquid chromatography has been shown to be very promising in the detection of olive oil adulteration with seed oils. This is especially true for seed oils with a high content in linoleic acid. The objective of this research was to evaluate the RP-HPLC method for detection of a seed oil with low linoleic acid content as the in olive oil. Twenty two authentic virgin olive oil samples were obtained and analyzed for triacyglycerol composition by RP-HPLC. Optimization of the method allowed fast separation of the triacylglycerols. Less than 30 min was needed for analysis of one sample, including its preparation. Based on this analysis two useful factors were established for the assessment of olive oil purity. First, authentic olive oil should not contain more than 1% of triacylglycerol species with equivalent carbon number (ECN) of 42 and second, the ratio of the area % of the peak that represents triacylglycerols with ECN of

46 to the area % of the peak that represents triacylglycerols with ECN of 44 should not be less than the value of 3.9. The twenty two virgin olive oil samples were then mixed with various levels (2.5-30% w /w) of canola oil, a high-oleic acid type of oil like olive oil, and analyzed by RP-HPLC. At the 7.5% level of adulteration the use of the area % of the peak for ECN of 42 could detect only 64% of the samples as adulterated. At the same level of adulteration the use of the peak ratio revealed 77% of the samples as adulterated. All samples were considered as adulterated by both factors when they were mixed with ~10% canola oil. The detection of corn, sunflower and soybean oils was feasible by both factors even at a 2.5% (w /w) admixture with olive oil. In addition, qualitative

11 results from the RP-HPLC method led to distinction between olive oil samples adulterated with canola oil from those adulterated with soybean oil. All samples were also analyzed for refractive index, absorption of UV light and fatty acid composition by GLC. These methods were inferior to RP­ HPLC as adulteration with the above mentioned seed oils below 5% was not detected. More specifically, RI analysis could detect the seed oils only when they were present in >30% in olive oil. UV analysis could detect ~5% sunflower and soybean oils, ;:::10% corn oil and ;:::12.5% canola oil in olive oil. The determination of the fatty acid composition was useful in the detection of >25% corn oil, >20% sunflower oil, >10% soybean oil and >12.5% canola oil in olive oil. Finally, a survey on commercial olive oil products was also conducted. From the RI, UV and fatty acid composition data it was concluded that all samples were properly labeled. However, based on RP-HPLC methodology as suggested in this thesis, 64% of the samples fell outside of characteristic lilnits for olive oils and thus, were considered as adulterated. This research work illustrated the advantages of the RP-HPLC technique over other fast and commonly utilized analytical procedures. The method can detect as low as 2.5% of seed oils with high linoleic acid content but olive oils adulterated with less than 10% canola oil might escape detection. The oil of new developed canola varieties as well as other inexpensive oils with similar composition to that of olive oil, should be the subject of future investigation.

111 Dedicated to ......

A solitary olive tree standing somewhere in Athens since 389 A.D ......

IV ACKNOWLEDGEMENTS

I wish to express my sincere gratitude to Professor Alan McCurdy. This thesis would not have been possible without his supervision, friendship and support. I thank the other members of my committee, Dr. G. Jones, Dr. R. Tyler and Dr. F. Sosulski for their advise.

A very special debt of thankfulness is owed to Dr. Humbert and Dr. McCurdy. When my admission to the graduate program seemed very difficult, due to administrative problems, they put their trust on me and gave me the chance. For this I am forever grateful.

The friendship and help shared with Virinder Grewal, Suresh Ramamurthi and Dr. Cheryl Tautorus, my colleagues from the "oil" lab, is appreciated.

Financial support from the College of Graduate Studies and Research is also acknowledged.

Last, but not the least, I thank my wife Andriana for everything.

v TABLE OF CONTENTS Page ABSTRACT ...... ii DEDICATED TO...... i v ACKNOWLEDGEMENTS ...... v TABLE OF CONTENTS ...... vi LIST OF TABLES ...... ix LIST OF FIGURES ...... xii 1 INTRODUCTION ...... 1 2 LITERATURE REVIEW ...... 2 2.1 General considerations on olive oil...... 2 2.2 Olive Oil designations...... 6 2.3 Olive oil adulteration ...... 10 2.4 Chemical Reaction Tests for the detection of adulteration of olive oil with seed oils ...... 13 2.5 Physical characteristics for the detection of adulteration of olive oil with seed oils ...... 16 2.5.1 Refractive Index (RI) ...... 17 2.5.2 Absorbance of Ultra-Violet (UV) Light ...... 18 2.6 Chemical characteristics for the detection of adulteration of olive oil with seed oils ...... 32 2.6.1 Iodine Value ...... 32 2.6.2 Saponification Value ...... 36 2.6.3 Bellier Number...... 39 2.7 Determination of the chemical composition for the detection of adulteration of olive oil with seed oils...... 39 2.7.1 Analysis of the sterols ...... 39 2.7.2 Analysis of the fatty acids ...... 44 2.8 Analysis of Triacylglycerols ...... 51

VI 2.8.1 Triacylglycerols ...... 51 2.8.2 Gas-liquid chromatography (glc) of triacyl- glycerols...... 53 2.8.3 Combined techniques for the analysis of triacyl- gl ycerols ...... 55 2.8.4 Triacylglycerol analysis by reversed-phase high performance liquid chromatography (RP-HPLC)...... 57 2.8.4.1 Theoretical aspects of HPLC...... 57 2.8.4.2 The development of reversed-phase high performance liquid chromatographic ana- lysis of triacylglycerols ...... 61 2.8.4.3 Quantitative analysis ...... 72 2.8.4.4 Other modes of HPLC in the analysis of triacylglycerols ...... 73 2.8.4.5 RP-HPLC as a means for detecting adulte- ration of vegetable oils...... 75 3. MATERIALS AND METHODS ...... 79 3.1 Oil samples and chemicals...... 79 3.2 Fatty acid analysis...... 82 3.3 Refractive Index...... 83 3.4 UV Absorbance...... 84 3.4.1 Direct analysis ...... 84 3.4.2 Purification through alumina...... 84 3.5 Triacylglycerol analysis by reversed-phase HPLC...... 85 3.6 Statistical methods...... ·...... 87 4. RESULTS AND DISCUSSION ...... 88 4.1 Refractive Index, UV absorption and fatty acid

vii composition of individual vegetable oils and model mixtures of olive oils and seed oils...... 88 4.1.1 Refractive Index (RI)...... 88 4.1.2 Absorbance of UV light...... 90 4.1.3 Fatty acid composition ...... 94 4.2 Triacylglycerol analysis of individual vegetable oils and model mixtures of olive oils with seed oils by RP-HPLC ...... 102 4.2.1 Preliminary studies...... 102 4.2.2 Triacylglycerol profiles and composition of olive oils, seed oils and their admixtures...... 111 4.3 Studies on commercially available olive oil products...... 128 4.3.1 Refractive Index...... 128 4.3.2 Absorbance of UV light...... 128 4.3.3 Fatty acid composition...... 132 4.3.4 Triacylglycerol profiles and composition of commercial olive oil products by RP-HPLC ...... 132 5. SUMMARY AND CONCLUSIONS ...... 140 6. REFERENCES ...... 144

Vlll LIST OF TABLES

Table 2.1. Imports of olive oil products in Canada ...... 5 Table 2.2. Imports of other olive oil products in Canada since 1988...... 6 Table 2.3. Canadian Standard for olive oil ...... 8 Table 2.4. Price comparison of oils found at supermarkets and gourmet stores in U.S. during October 1990...... 12 Table 2.5. Price comparison of oils found at supermarkets and gourmet stores in Canada during October 1991...... 13 Table 2.6. Refractive index of some vegetable oils...... 19 Table 2.7. Refractive index limits for olive oils...... 20 Table 2.8. Absorbance of crude and refined vegetable oils at UV light...... 26 Table 2.9. Ultraviolet absorbance limits for olive oils according to FAO, 1970 ...... 28 Table 2.10. Ultraviolet absorbance limits for olive oils according to IOOC, 1985...... 29 Table 2.11. Ultraviolet absorbance limits for olive oils according to the European Economic Community ...... 30 Table 2.12. Absorbance of UV light for olive oils of different origins ...... 33 Table 2.13. Iodine Values (I.V) of some vegetable oils...... 35 Table 2.14. Iodine value limits for olive oils ...... 36 Table 2.15. Saponification Values (S. V) of some vegetable oils...... 37 Table 2.16. Saponification value limits for olive oils...... 38 Table 2.17. Bellier Number (B.N .) limits for olive oils...... 40 Table 2.18. Sterol composition (%) limits of olive oil by capillary glc...... 43 Table 2.19. Fatty acid composition of some vegetable oils ...... 46 Table 2.20. International fatty acid composition limits for olive oil...... 48 Table 3.1. List of the olive oil samples with gu~ranteed purity...... 80 Table 3.2. List of the commercial olive oil samples...... 81

IX Table 4.1. Refractive indices of olive oil, seed oils and their admixtures...... 91 Table 4.2. Absorbance at 270 nm of olive oil, seed oils and their admixtures...... 95 Table 4.3. Fatty acid composition of virgin olive oil samples of guaranteed purity...... 96 Table 4.4. Fatty acid composition of olive oil, seed oils and their admixtures ...... 100 Table 4.5. Separation of triacylglycerols by RP-HPLC using various mobile phases...... 104 Table 4.6. Separation of triacylglycerols by RP-HPLC using various injection solvents ...... 106 Table 4.7. Retention time (tR), by RP-HPLC, of triacylglyce- rols eluted according to equivalent carbon number...... 109 Table 4.8. Triacylglycerol composition, of olive oil samples with guaranteed purity, as determined by direct RP-HPLC analysisa...... 115 Table 4.9. Triacylglycerol composition of seed oil samples, as deter- mined by direct RP-HPLC analysisa· ...... 117 Table 4.10. Area % of the peak for ECN 42 and ratio of 46/44 peaks for olive oil with guaranteed purity and its admixtures with different proportions of common seed oils as determined by direct RP-HPLC analysis of the triacylglycerols ...... 121 Table 4.11. Area % of the triacylglycerols peak for ECN 42 and the ratio of 46/44 peaks as determined by direct RP-HPLC analysis for olive oil with guaranteed purity and its admixtures with different proportions of canola oil...... 122 Table 4.12. Percentage of olive oil samples (n=22) mixed with canola oil that were detected as adulterated using the area% of peak for ECN 42 or the ratio of area% of peak for ECN 46 to peak for ECN 44 after direct RP-HPLC analysis of the triacylglycerols ...... 124 Table 4.13. Fatty acid composition of commercially available olive oil products ...... 133

X Table 4.14. Triacylglycerol composition of commercially available olive oil products, as determined by direct RP-HPLC analysisa ...... 136

XI LIST OF FIGURES

Figure

Figure 2.1. UV transmission spectra of a lampante olive oil in advanced state of oxidation (1), refined olive oil (2), and refined olive-residue oil (3) (adapted from Gracian, 1969)...... 22 Figure 2.2. UV absorbance spectra of oils containing conjugated dienes (1), conjugated dienes and secondary pro- ducts of oxidation (2), conjugated trienes (3), and conju- gated tetraenes (4) (adapted from Dimoulas, 1981) ...... 23 Figure 2.3. UV transmission spectra of an olive oil after no processing (1), neutralization (2), neutralization and bleaching by earth (3), neutralization and bleaching by carbon (4), neutralization, bleaching by earth and deodorization (5), and neutralization, bleaching by carbon and deodorization (6) (adapted from Gracian, 1969) ...... 25 Figure 4.1. Refractive indices (@ 20°C) of 22 virgin olive oil samples with guaranteed purity. Lowest and highest values on the y axis represent the interna- tional olive oil limits established by FAO/WHO, 1970 ...... 89 Figure 4.2. Absorbance of 22 virgin olive oil samples at 232nm. Limit for virgin olive oils is 3.50, whereas there is no specific limit for virgin lampante olive oils...... 92 Figure 4.3. Absorbance of 22 virgin olive oil samples at 270nm. Limit for virgin olive oils is 0.25, whereas for virgin lampante olive oils there is either no limit (IOOC, 1985), or a limit of 1.10 from the European Community Standards (Zygourakis, 1988)...... 93 Figure 4.4. Reversed-phase high performance liquid chroma- togram of a virgin olive oil (sample GC2) by using acetone as the mobile phase...... 105 Figure 4.5. Plot of retention time vs equivalent carbon number of trilinolenylglycerol, trilinoleylglycerol, trioleylglycerol

xii and tristearoylglycerol (in eluting order) analyzed by RP-HPLC...... 108 Figure 4.6. Reversed-phase high performance liquid chroma- togram of oleic acid and trioleylglycerol...... 110 Figure 4.7. Plot of integrator's area counts vs milligrams of trioleyl- glycerol (000) after RP-HPLC analysis ...... 112 Figure 4.8. Reverse-phase high performance liquid chromatograms of virgin olive and seed oils. Numbers represent ECN values, where ECN = carbon number - 2 x number of double bonds...... 113 Figure 4.9. Representative reversed-phase high performance liquid chromatograms of virgin olive oils. (A): Greek; (B): Spanish; (C): Italian; (D): Californian...... 114 Figure 4.10. Scattergram of the ratio of the area %of peak for ECN of 46 to area % of peak for ECN of 44 for 22 olive oil samples with guaranteed purity...... 120 Figure 4.11. Reversed-phase high performance liquid chroma- tograms of virgin olive oil (A), soybean oil (B) and sample A adulterated with 30% (w /w) sample B...... 125 Figure 4.12. Reversed-phase high performance liquid chroma­ tograms of virgin olive oil (A) and sample A aduJte­ rated with 7.5% (w/w) soybean oil (B), 30% (w/w) soybean oil (C) and 30% (w /w) canola oil (D)...... 127 Figure 4.13. Refractive indices (® 20°C) of commercially available olive oil products. Lowest and highest values on they axis represent the international olive oil limits established by FAO/WHO, 1970 ...... 129 Figure 4.14. Absorbance at 232 nm of commercially available olive oil products. Limit for virgin olive oils is 3.50, whereas there is no specific limit for products labe- led as olive oils...... 130 Figure 4.15. Absorbance at 270 nm of commercially available olive oil products. Limit for virgin olive oils is 0.25, whereas for' products labeled as olive oils is 0.90 (IOOC, 1985) ...... 131 Figure 4.16. Reversed-phase high performance liquid chroma-

xiii tograms of <:ommercially available olive oil products...... 134 Figure 4.17. Reversed -phase high performance liquid chroma­ tograms of 30% (w /w) canola in olive oil (A), 7.5% soybean in olive oil (B) and commercial olive oil sample SK3 (C)...... 139

xiv 1. INTRODUCfiON

Olive oil occupies a unique position among edible oils due to its chemical composition and organoleptic characteristics. Its production is basically restricted to the Mediterranean area where it is considered the main oil and consumed in high amounts. Virgin olive oil has a delicate flavour and aroma and, due to the limited cultivation of the olive tree, it is one of the most expensive oils internationally. Therefore, it has always been the subject of fraud by mixing it with other less expensive oils.

Producers and processors have found different ways to adulterate olive oil, but at the same time numerous methods have been developed in order to reveal the frauds. This is especially true during the las~ century in which a tremendous amount of work has been carried out by the scientific community in this area. Both specific and non-specific procedures have been proposed in order to detect certain foreign oils or a class of foreign oils in olive oil.

A great number of different oils have been used to adulterate olive oil. Among the most popular, the oils that have a similar composition to that of olive oil (refined olive and olive residue oils, teaseed and groundnut oils) and/ or a very low price (soybean, com and sunflower oils). Some of them are not likely to be used as today either because of their price or because they can be detected rather easily by available techniques. For the same reasons other oils, common or newly developed, have might become potential adulterants for olive oil. Most of the analytical methods existing today can not detect the foreign oil if it is present in ~10% in olive oil.

1 2

Chromatography has dominated in the field of oil analysis due to superior precision and accuracy as well as compatibility with sophisticated technology like computers. The determination of the fatty acid composition by gas-liquid chromatography has become a routine analytical tool for the assessment of olive oil purity. Until recently the analysis of the main compounds of an oil, i.e., the triacylglycerols, had not been used as an aid for verifying the authenticity of olive oil. Development of new liquid chromato­ graphy systems has resulted in the fast development of this technique for the separation of triacylglycerols. The utilization of this method as a way to detect olive oil adulteration with foreign oils has not been well studied.

The objective of this research was to investigate if the direct analysis of olive oil by reversed-phase high performance liquid chromatography (RP­ HPLC) is an appropriate way to detect the presence of low levels of canola oil, a high-oleic acid type of oil similar to olive oil. The advantage of the RP­ HPLC method over other fast analytical techniques as a means to detect foreign oils in olive oil, was investigated by application of the method both to model admixtures of olive oil with common seed oils and. to commercial olive oil samples. 2. LITERATURE REVIEW

2.1 General considerations on olive oil Olive oil is the oil obtained solely from the fruit of the olive tree (Olea europaea sativa ), to the exclusion of oils obtained using solvents or re­ esterification processes and any mixtures with oils of other kinds. This definition has been proposed by the International Olive Oil Council (IOOC, 1985) an organization composed of 19 nations accounting for 96% of the total world olive oil production (Kiritsakis and Markakis, 1987).

Olives grown and collected under the right conditions will yield an oil with excellent flavour and taste and most importantly this oil can be consumed as it is, without any further refining (Jacini, 1976). This is a property that distinguishes olive oil from the other edible oils, especially the· seed oils. In seed oils the model product is the refined oil, whereas in olive oil the refining process is considered detrimental (Jacini, 1976).

The production of olive oil is basically restricted to the Mediterranean region because of the weather conditions that are needed for the growing of the olive tree. About 98% of the olive oil production comes from the Mediterranean area with main producing countries Spain, Italy and Greece all belonging to the European Economic Community (EEC). Outside of the EEC, Tunisia, Morocco and Libya produce smaller but appreciable amounts. Olive oil is also produced in South America (Argentina) and the United States (California) but in very low quantities.

3 4

Olive oil is the most important in the diet of Mediterranean people. However, its consumption has been increasing in North America over the last decades. Tables 2.1 and 2.2 show the olive oil products imported to Canada since 1980. The main reasons that have led to such an increasing consumption are, its unique delicate flavour and aroma and its moderate degree of unsaturation which is considered nutritionally preferable to the high degree of saturation or unsaturation of most of the other edible oils and (Kiritsakis and Markakis, 1987).

Elevated plasma cholesterol is considered a risk factor for coronary heart (CHD). However,.high-density lipoproteins (HDL) can protect against CHD (Schaefer, 1984). The diet usually recommended for lowering cholesterol is low in total fat, saturated fatty acids, and cholesterol. Such a diet is found in countries where CHD is relatively low. Certain areas in the Mediterranean region are exceptions to this (Grundy, 1986). In Greece (especially Crete) and in southern Italy the total intake of fat is usually high but the diet is very rich in olive oil. Epidemiological studies such as the Seven Countries Study have shown that in Crete the incidence of CHD in middle-aged men was lower than the expected from their total serum cholesterol levels (Mensink and Katan, 1987). This fact could not be explained by other risk factors such as smoking or high blood pressure. A diet rich in olive oil has been shown to cause a specific fall in non-high density lipoprotein cholesterol, while leaving high density lipoprotein cholesterol and triacylglycerols unchanged (Mensink and Katan, 1987). Furthermore, a number of studies have characterized olive oil with some other beneficial physiological effects, such as being eupetic, cholagoguic, and homeostatic (Kiritsakis and Markakis, 1987). Table 2.1. Imports of olive oil products in Canadaa.

Amounts in tonnes Country of origin 1980b 1981b 1982b . 1983b 1984b 1985b 1986b 1987b 1988C 1989C 1990C

France 29 24 35 27 38 60 44 70 _d Greece 261 417 452 316 364 258 475 369 316 327 355 Italy 1097 872 1035 1615 2500 2576 2027 3393 1631 1678 2079 Portugal 106 112 147 137 186 104 157 312 194 185 242 Spain 2576 1282 1043 1513 1745 1771 1280 1822 353 262 542 u.s 227 167 1356 70 55 141 111 153 237 Other - 3 2 5 34 51 20 21 90 191 120 Total 4296 2877 4070 3683 4922 4961 4114 6140 2820 2644 3339 asource : Statistics Canada (1980-1990). bListed under the term "olive oil". CListed under the term "olive oil,virgin". dNo available data.

5 Table 2.2. Imports of other olive oil products in Canada since 1988a,b.

Amounts in tonnes Product IC Product ud

Country 1988 1989 1990 1988 1989 1990 of origin

Greece 215 210 212 _e Italy 2999 3090 3569 165 Portugal 116 Spain 1777 1342 1432 Other 66 161 209 40 93 29

Total 5173 4803 5422 206 93 29 asource: Statistics Canada (1988-1990). bsince 1988 olive oil products are listed under different categories. CListed under the term "Olive oil & its fractions, refined but not chemically modified". dListed under the term "Oils & their fractions obtained from olives, refined or not, not chemically modified, including blends". eNo available data.

6 7

Other studies have indicated that olive oil possesses very high digestibility compared to other vegetable oils (only was found as digestible) (Fedeli, 1977) and also produced very low incidences of cardiac fibrosis and malignant tumors in animal studies (Kaunitz, 1978).

2.2 Olive Oil designations. One of the main problems in olive oil marketing and research is the numerous definitions that apply to olive oil products. Such terms and descriptions lead to consumer confusion and many times are exploited for · frauds. The problem is even bigger when government regulations in different countries do not take into consideration such definitions and distinctions between olive oil products. In Canada, the Food and Drugs Act (Division of Fats and Oils) (1985) lists olive oil as "olive oil or sweet oil" and it does not even mention the market names of the several olive oil products (Table 2.3). The consumer cannot realize differences in labels such as "extra

virgin olive oil", "pure olive oil', and "100 % pure olive oil". Unfortunately this is the case even among people from the main olive oil producing countries or even olive oil producers. Therefore, consumer education is needed not only for avoiding confusion but also as a driving force for the production of better quality olive oil products.

Following are the definitions used for olive oil products according to the International Olive Oil Council (IOOC, 1985) : ''Virgin olive oil - The oil obtained from the fruit of the olive tree solely by mechanical or other physical means under conditions, particularly thermal conditions, that do not lead to alterations in the oil, and which has not undergone any treatment other than washing, decantation, centrifugation, 8

Table 2.3. Canadian Standard for olive oil

Olive Oil or Sweet Oil (a) Shall be the oil obtained from the fruit of the olive tree; (Olea europaea L ); (b) shall have a fatty acid content that is (i) oleic acid : 56.0 - 83.0% (ii) palmitic acid : 7.5 - 20.0% (iii) linoleic acid : 3.5 - 20.0% (iv) stearic acid: 0.5- 3.5% (v) palmitoleic acid : 0.3 - 3.5% (vi) linolenic acid : :::; 1.5% (vii) myristic acid : :::; 0.05%, calculated as methyl esters; (c) shall not contain more than minute amounts of arachidic, , behenic, gadoleic or lignoceric acids; (d) shall have (i) relative density : 0.910 - 0.916@ 20°C (ii) refractive index: 1.4677-1.4705@ 20°C (iii) iodine value : 75 - 94, by the Wijs test (iv) saponification value : 184- 196 in mg KOH/ g oil (v) acid value : :::; 6.6 in mg KOH/ g oil (vi) free acidity : :::; 3.3% as oleic acid (vii) peroxide value: :::; 20 meq peroxide oxygen /kg oil (viii) unsaponifiable matter: :::; 15g /kg oil (ix) Bellier index : :::; 17 in °C; (e) shall show negative results in tests for semi-siccative, olive-residue, cottonseed, teaseed or sesame oils; 9 and filtration; Virgin olive oil fit for consumption as it is includes: Virgin olive oil extra - Virgin olive oil of absolutely perfect flavour and odour having a maximum acidity, in terms of oleic acid, of 1 gram per 100 grams; Virgin olive oil fine - Virgin olive oil of absolutely perfect flavour and odour having a maximum acidity, in terms of oleic acid, of 1.5 grams per 100 grams; Virgin olive oil semi-fine (or virgin olive oil ordinary or courante) - Virgin olive oil of good flavour and odour having a maximum acidity, in terms of oleic acid, of 3 grams per 100 grams, with a tolerance margin of 10 % of the acidity indicated. Virgin olive oil not fit for consumption as it is designated virgin olive oil lampante, is an off-flavour and/ or off-smelling virgin olive oil or an oil with a maximum acidity, in terms of oleic acid, of more than 3.3 % grams per 100 grams. It is intended for refining or for technical purposes; Refined olive oil - The oil obtained from virgin olive oils by refining methods which do not lead to alterations in the initial glyceridic structure; Olive oil or pure olive oil or 100% pure olive oil - The oil consisting of a blend of virgin olive oil fit for consumption as it is and refined olive oil. Blends of virgin olive oil and refined olive oil may constitute types, the characteristics of which may be determined by mutual agreement between buyers and sellers. However, these blends must meet the quality criteria as stipulated in section 4 of this standard for pure olive oil; Olive-residue oil- The oil obtained by treating olive residue with solvents, to the exclusion of oils· obtained by re-esterification processes and any mixture with oils of other kinds. It can be classified as follows : 1 0

Crude olive-residue oil - olive-residue oil intended for refining with a view to its use in food for human consumption or for technical purposes; Refined olive-residue oil -Obtained from crude olive-residue oil by refining methods which do not lead to alterations in the initial glyceridic structure. It is intended for human consumption either as it is or else in a mixture with virgin olive oil; Refined olive-residue oil and olive oil - mixture of refined olive-residue oil and virgin olive oil fit for consumption as it is; this mixture is usually intended for domestic consumption in some producing countries".

2.3 Olive oil adulteration Among the edible oils, the product which most requires a clear identification of its purity is olive oil, which has been and will most likely continue to be adulterated (Firestone et al., 1985; 1988) mainly because of its higher price on the market. Apart from the fact that it is a fraud, adulteration of edible oils with other cheaper oils may cause health problems. The most recent example is the case in Spain and the disease now known as the Spanish Toxic Syndrome where adulteration of mainly olive oil with industrial grade rapeseed oil (denatured with aniline) (Kochhar and Rossell, 1984) caused the death of several hundred people within a year (Jimeno, 1982).

In olive oil producing countries, illegal additions to genuine olive oil would include refined olive and olive residue (solvent extracted and refined) oils. This can be accomplished by either the small producer or by large-scale industries that blend and pack olive oil. On the other hand, in countries that 1 1 import olive oil, especially in bulk amounts, common adulterants used during packaging probably would include inexpensive and highly available seed oils.

Olive oil is frequently adulterated by other vegetable oils of a lower commercial price. Tables 2.4 and 2.5 show the prices of several edible vegetable oils compared to those of olive oil products in Canada and the United States. Official methods for purity control of olive oil include criteria such as physical and chemical constants in combination with the determination of the absorbance in the ultraviolet region (UV) and the fatty acid composition by gas-liquid chromatography .(GLC) (Kapoulas and Passaloglou-Emmanouilidou, 1981).

However, the usual physical and chemical values have a very limited use. Observing the limits recorded for physical and chemical values of the majority of vegetable oils, e.g. density, refractive index, saponification value, iodine value, etc.,and those admissible for olive oil, it is easy to prepare fraudulent mixtures, even with high proportions of adulterants (Galanos and Kapoulas, 1965; Gracian, 1969). Today, some of these methods (refractive index) are still being used, especially for their simplicity, and because they can raise suspicions of the analyst.

2.4 Chemical Reaction Tests for the detection of adulteration of olive oil with seed oils For a very long time, scientists tried to solve analytical problems concerning the detection of adulteration of olive oil by means of the application of "colour reactions", with an empirical basis. References on 1 2

Table 2.4. Price comparison of oils found at supermarkets and gourmet stores in U.S. during October 1990.

Type of oil dollarsa/1

Canol a 0.95-2.85 Corn 2.53-2.85 Olive (pure) 5.06-8.23 Olive (extra virgin) 9.81-42.72 Peanut 4.75 Soybean 1.90-2.85 Sunflower 2.53-3.48

Adapted from Latta, 1991. aln Canadian funds. 1 3

Table 2.5. Price comparison of oils found at supermarkets and gourmet stores in Canada during October 1991.

Type of oil dollars/1

Canol a 1.98-3.19 Corn 2.68-2.99 Olive (pure or 100% pure) 4.48-6.69 Olive (extra virgin) 6.99-8.99 Peanut 4.00-5.99 Soybean _a Sunflower 2.48-3.99 aNot available. 1 4 these tests are rather old because of their empirical nature which has limited their use. Therefore, the following discussion is based on the exhaustive review on the subject by Gracian (1969).

In 1939 Vizern and Guillot introduced a general test for the detection of semi-drying oils. The method is based on the bromination of the unsaturated compounds of an oil, especially linoleic acid. A precipitate is formed which indicates the presence of foreign oils in olive oil. The results of the test are highly dependent on the linoleic acid content of the oil under test and one of the main problems of this method is the fact that olive oils with a rather high iodine value might form a precipitate and give erroneous results. Another limitation of this test is with olive oils containing impurities, as in the case of lampante olive oils. Such oils might also give positive reactions without being adulterated.

Bellier's reaction is another commonly used test for the detection of seed oils in olive oil. This test is based on the reaction of nitric acid and resorcin with some components of the oil which have not been identified. Olive oil gives a clear green colour whereas, seed oils -give a blue-violet colour. Introduction of seed oils in olive oil will result in the appearance of a violet or blue colouration. The main problem with this test is the positive reaction that olive oils containing impurities might give.

The Synodinos-Konstas reaction is a modification of the Hauchecorne's reaction which was related to the reaction of nitric acid with decomposition products of the autooxidation of oils as well as compounds formed during industrial manipulations. The presence of seed oils in olive 1 5 oil results in the formation of a number of colours different from the yellow colour that the virgin olive oils give. However, many non-adulterated virgin olive oils have been found to give brown colours, a problem that was eliminated by the introduction of a modification in the Hauchecorne's test from Synodinos and Konstas. According to their modification, the oil must be treated with bleaching earth prior to the reaction with the nitric acid. The method has been used extensively and theoretically it can detect any oil that has undergone any industrial process since the compounds responsible for positive reaction (oxo and hydroxy compounds) are not only produced by oxidation (which will be removed by the activated earth) but also from the application of heat. Therefore, the method has a general character and is usually found in the literature as a method for detecting foreign oils in virgin olive oil. Nevertheless, se.ed oils that have been expressed (not solvent extracted) may give a negative reaction and olive oils containing impurities rather than adulterants may give positive results.

Apart from the above mentioned general tests some oils give special reactions which can lead to their identification when incorporated in olive oil. Cottonseed oil can be detected by the Hal ph en reaction. This is a characteristic reaction for cottonseed oil since the compounds responsible for it are the cyclopropenoic acids (malvalic and sterculic) found in cottonseed oil. These compounds are also found in kapokseed oils. However, the test is unreliable for refined cottonseed oil because the cyclopropenoic acids are very susceptible to heat and/ or acid conditions.

The Fitelson test is characteristic for teaseed oils. The presence of teaseed oil in olive oil is revealed by the formation of red colour but teaseed 1 6 oil must be present at a 10 % or higher level. Some olive oils from Spain and Tunisia have been found to develop a pink-red colour during this test. These oils had in common extreme content in palmitic and linoleic acids or they were derived from olives with abnormal metabolic activity due to frosty environmental conditions. In addition, lampante olive oils of very low quality or olive oils containing unusual impurities may hide the red colour formed by the presence of teaseed oil.

Finally, sesame oil might be detected by the Villavenchia-Fabris reaction. Responsible compounds for this reaction are the phenolic compounds sesamin and sesamolin. A red colour is formed from the reaction of furfural with sesamol (formed by hydrolysis of sesamolin by hydrochloric acid). The method is very sensitive as it can detect as low as 0.5 % sesame oil in olive oil, but olive oils from certain areas such as Tunisia, Algeria and South Italy might give a slightly positive reaction.

The tests described above have been shown to be highly empirical and therefore their reliability is questionable. Moreover, it is obvious that these tests must be performed by experienced analysts. Today only a few of them are still performed by olive oil quality control laboratories. They have been abandoned and replaced by more sensitive and reproducible methods.

2.5 Physical characteristics for the detection of adulteration of olive oil with seed oils The following is a partial list of methods that have been widely applied in the detection of olive oil adulteration. Some of these methods are not commonly used today. Emphasis is given on the methods used for this 1 7 thesis. Discussion is restricted to the use of these methods for the detection of seed oils. Many of these methods have also been utilized for the detection of refined olive and olive residue oils in virgin olive oil, but this use is outside the scope of this thesis.

2.5.1 Refractive Index (RI) Newton introduced the concept of refraction at the beginning of the eighteenth century (Batsanov, 1961). According to modern theory, however, the formula defining refraction is diametrically opposite to that proposed by Newton (Batsanov, 1961):

n =velocity of light in air I velocity of light in matter

In oils and fats refractive index has been used for. following the course of the hydrogenation reaction of oils, as well as a criterion for the purity of oils (Cocks and van Rede, 1966). The use of RI as a factor for checking the identification of oils has been based on the fact that few substances have identical Ris at a specific temperature and wavelength (Olsen, 1975). The RI of organic compounds is related to density, molecular weight, and structure of the compounds (Eckey, 1954).

Following is a summary of the relationship between RI and the structure of fatty acids and acylglycerols (Formo, 1979): (a) The longer the hydrocarbon chain the higher the RI. (b) The greater the number of double bonds in the fatty acid the higher the RI. (c) Increased conjugation in the fatty acid molecule leads to a higher RI. 1 8

(d) Simple acylglycerols have a higher RI than the fatty acids that constitute them. (e) Mixed acylglycerols have RI's close to the ones of the simple acylglycerols that constitute them.

The RI of an oil is generally affected by free acidity, polymerization, and the presence of secondary groups in the molecules such as hydroxy groups (Dimoulas, 1981).

At 20°C drying oils have RI's between 1.4800 and 1.5230, the semi­ drying oils have RI's between 1.4700 and 1.4760, and non-drying oils have RI's between 1.4680 and 1,4700. Table 2.6 contains RI limits for several oils of vegetable origin. From the data of this table it is obvious that olive oil has the lowest RI among these oils and therefore adulteration of olive oil with these oils will result in an increase of its RI. However, it is well known that mixing olive oil with these seed oils even in high proportions will not raise the RI above the upper limit for olive oils. This is the result of the wide variation that oils exhibit in their RI's due to different composition. This variability depend on variety, country of origin, and way of cultivation. This is illustrated in Table 2.7 which contains data for RI limits registered for olive oils from different countries.

2.5.2 Absorption of Ultra-Violet (UV) Light In general, the application of UV spectrophotometry to fats and oils is based on observations made in the 1930's related to increased absorption of UV radiation by unsaturated fats that had undergone saponification (Eckey, 1954). However, Lewkowitch had already published data on UV spectropho- 1 9

Table 2.6. Refractive index of some vegetable oils.

Oil Type

Peanutb 1.4672 - 1.4722

CottonseedC 1.4698 - 1.4738 Oliveb 1.4677 - 1.4705 Canolad 1.4722 - 1.4742 Cornb 1.4722 - 1.4752 Sunflowerb 1.4742 - 1.4762 Soybeanc 1.4718- 1.4778 aRefractive index@ 20°C. hFood and Agriculture Organization/World Health Organization (FAO/WHO, 1969-1970). Cformo, 1979. d Vaisey-Genser and Eskin, 1982. 20

Table 2.7. Refractive index limits for olive oils.

Sample source noa

Greece 1.4680- 1.4695

Italy 1.4686 - 1.4703 Portugal 1.4681 - 1.4695 Tunisia 1.4680 - 1.4695 Spain 1.4676 - 1.4705 Algeria 1.4676 - 1.4698 Argentina 1.4685 - 1.4708 Morocco 1.4690 - 1.4700

Adapted from Gracian, 1969. aRefractive index @ 20°C. 2 1 tometry of oils since 1927 (Gracian, 1969). The first findings related to olive oils were on the differences in absorbance between virgin and refined oils (Figure 2.1). Changes in absorbance of an oil due to industrial treatments have also been well studied in an attempt to differentiate between virgin and industrially manipulated oils. A major concern in the interpretation of such data is the fact that oxidation products of the fatty adds and their esters absorb at the same wavelengths as the products derived from chemical and/ or physical treatments and particularly those products produced during the refining process of the oils (Figure 2.2).

Saturated or non-conjugated unsaturated fatty acids absorb strongly in the far UV electromagnetic radiation where measurements are difficult to perform, i.e., at 125-190 nm for saturated and at 180-210 nm for isolated multiple bonds (Rao, 1961). In the near region of UV radiation the compounds of interest that absorb strongly are mainly the fatty acids containing conjugated double bonds.

Naturally found fatty acids containing conjugated double bonds are rare in common vegetable oils. Therefore, strong absorbance of UV light is considered to be from conjugation due to processing and/ or oxidation. The main electronic transitions responsible for increased absorption in near UV of the conjugated containing fatty acids are the 1t--1t* and n--1t* (Rao, 1961; Olsen, 1975). Any multiple bond contains 1t and n electrons, resulting in the above mentioned transitions, with a shift to longer wavelengths of absorption and increased in intensity as the extent of conjugation increases (Olsen, 1975). 22

0.6 .:\ \ .. 0.5 --., ... \ ·------1 0.4 \ \ \ ---2 0.3 \ \ \ ....•.···\ \ \ ... ..\···········3 0.1 \ \ \ \ \ \ \ \ I I I \ \ ' ' ..... --, ' ' \ ' \ \. \ ··.\ \ \ \ \ \ \ .. \ \ \ ', . 01 '-., 0 \ \ .. \ . \ . ' \ .

'\·.... ·· ..

Figure 2.1. UV transmission spectra of a lampante olive oil in advanced state of oxidation (1), refined olive oil (2), and refined olive-residue oil (3) (adapted from Gracian, 1969). 23

Q) u m~ ~ 0 U) ,..0 ~ 1

1..0

0.5

A (mp)

240 260 280 300 320

Figure 2.2. UV absorbance spectra of oils containing conjugated dienes (1), conjugated dienes and secondary products of oxidation (2), conjugated trienes (3), and conjugated tetraenes (4) (adapted from Dimoulas, 1981). 24

Conjugated dienes, formed during treatment of an oil with bleaching earth (Van Den Bosch, 1973), absorb between 217 and 230 nm. Cyclic dienes absorb at longer wavelengths (230-280 nm) but the intensity is lower. However, oxidation of fatty acids also leads to a shift of bonds resulting in the formation of conjugated hydroperoxides (Nawar, 1985) which also absorb at the region of 220-235 nm with a maximum at 232 nm. Bleaching earth also catalyzes the formation of conjugated trienes and tetraenes which exhibit absorbance maxima between 270 and 315 nm (Figure 2.3). The exact wavelength that the maximum absorbance takes place is affected by the solvent used (Olsen, 1975). The formation of trienes is due to linoleic acid whereas tetraenes are formed from linolenic acid. The same products can be formed, even in the absence of bleaching earth, at elevated temperatures (>100°C), (Gracian, 1969). These changes during processing have been demonstrated for crude (cold pressed) and refined peanut and oils (Table 2.8).

Problems exist with the use of absorbance at 270 nm for the assessment of the purity of an olive oil sample, as secondary products of oxidation absorb in the same region. Again 1t--1t* and n--1t* transitions are the main reasons for increased absorbance. Particularly, carbonyl compounds such as aldehydes and ketones absorb at these wavelengths. Saturated carbonyl compounds absorb at around 275 nm (Rao, 1961). The n--1t* transition generally leads to increased absorbance at longer wavelengths but alkyl substitution will shift the absorbance band to lower wavelengths.

As the oxidation of an oil progresses an increase in peroxides is accompanied with an increase in absorbance at 232 nm. Later, peroxides 0.3- 25 0.2

0.1

-0.1

--x-- 2 ·--·•·-·· 3

-·

-··t>-·· 5

···-o-··· 6

Figure 2.3. UV transmission spectra of an olive oil after no processing (1), neutralization (2), neutralization and bleaching by earth (3), neutralization and bleaching by carbon (4), neutralization, bleaching by earth and deodorization (5), and neutralization, bleaching by carbon and deodorization (6) (adapted from Gracian, 1969). 26

Table 2.8. Absorbance of crude and refined vegetable oils of uv light.

Absorbance

Oil Type 232 nm 270 nm

Peanut crude 6.25 0.75

neutralized 6.20 0.70 bleached 3.60 3.40 deodorized 3.60 3.40

Almond crude 1.60 0.08 refined 2.25 1.80

Adapted from Dimoulas, 1981. 27 decompose and secondary products of oxidation are formed causing the absorption at 232 nm to decrease while absorbance at 270 nm increases. The interpretation of UV data is therefore difficult to use in making conclusions as to the oxidation stage or the purity of an oil, especially if only the absorbance at 232 nm is studied.

After many years of studies FAO /WHO (1970) established limits of absorbance for different olive oil products. Table 2.9 shows the FAO/WHO proposed limits for virgin, refined olive, refined olive-residue, and blends of virgin with refined olive and refined olive-residue oils. The IOOC in its 1985 recommendations includes only absorbance limits at 270 nm as a quality criterion, but more detailed information is given for different qualities of virgin olive oils (Table 2.10). The proposed specifications of the European Economic Community (Zygourakis, 1988) for olive oil products differ from the IOOC ones in the limit of the extinction at 270 nm for courante (semi­ fine), lampante and pure olive oils (Table 2.11).

The procedure for determining the absorbance of an olive oil sample of UV light has been basically the same through the last twenty years. The kind of solvent used is the main difference between the method described by Gracian in 1969, by FAO/WHO in 1970, by IOOC in 1985 and more recently by the one submitted to the EEC commission for marketing standards of olive oil and described in the Food Science Information Bulletin of the British Ministry of Agriculture, Fisheries and Food (MAFF, 1990). Cyclohexane and iso-octane are the main solvents of choice in this analytical procedure. Som;e of these methods suggest either solvent depending on their availability in a particular laboratory. Caution must be taken to work with oil concentrations 28

· Table 2.9. Ultraviolet absorbance limits for olive oils according to FAO/WHO, (1970).

Absorbance max.

Oil Type 232 nm 270 nm

Virgin 3.50 0.25 Refined olive 1.10 Refined olive-residue 6.00 2.00 Blends of virgin and refined olive 0.90 Blends of virgin and refined olive-residue 5.50 1.70 29

Table 2.10. Ultraviolet absorbance limits for olive oils according to IOOC, (1985).

Absorbance max. Oil Type 270nm

Virgin - extra 0.25

fine 0.25 semi-fine 0.30 lampante no limit Refined olive 1.10 Refined olive-residue 2.00 Blends of virgin and refined olive 0.90 Blends of virgin and refined olive-residue 1.70 30

Table 2.11. Ultraviolet absorbance limits for olive oils according to the European Economic Community.

Absorbance Oil Type 270nm

Virgin - extra max 0.25

fine max 0.25 courante max 0.25 lampante min 0.25 - max 1.10 Refined olive min 0.25- max 1.10 Refined olive-residue Blends of virgin - and refined olive min 0.25- max 1.10 Blends of virgin and refined olive-residue

Adapted from Zygourakis, 1988. 3 1 such that absorbance is a linear function, according to Beer and Lambert law (Pomeranz and Meloan, 1978a).

Any doubt about the purity of an olive oil sample which shows absorbance at 270 nm higher than the set limits can be solved by passing the oil through alumina. Alumina retains the more polar secondary oxidation products and the trienes are eluted with hexane (Dimoulas, 1981). For virgin olive oils the absorbance at 270 nm, after such purification, should never exceed a value of 0.11 (Gracian, 1969).

When data on the absorbance of an olive oil sample at 270 nm are not definite, some authors suggest the additional use of the absorption at higher wavelengths, particularly 315 nm. As mentioned earlier, conjugated tetraenes (formed either by treatment with bleaching earth or by oxidation) show their absorption maximum at 315 nm.

Galanos et al. (1968a) used the formula Rs=E315-E320/E310-E325 and concluded that detection of as little as 2-5% of refined oils in virgin olive oils is feasible. However, this technique could lead to erroneous conclusions about the purity of an olive oil sample due to experimental errors in samples that show low absorbance at the above wavelengths (Kapoulas and Andrikopoulos, 1987).

Such limitations could be overcome by the utilization of second­ derivative spectrophotometry (Kapoulas and Andrikopoulos, 1987). However, this approach also has some disadvantages as: (a) the lower level of detection of foreign oils usually ranges between 5 and 20%; (b) if crude seed oils (expressed, non-refined) are used as adulterants then the test is not 32 applicable since virgin seed oils show similarly low absorbance as do virgin olive oils (Gracian, 1969).

From the above discussion it is also obvious that the absorbance at UV light can not differentiate between adulterants since all refined oils, including refined olive oil products, show similar absorbance behaviour at UV radiation. Furthermore, if the oil sample is analyzed directly then distinction between an oxidized and an adulterated sample is not possible, especially in lampante olive oils where the set upper limits of absorption are either very high or absent (Table 2.10, 2.11).

Moreover, it has been observed that olive oils of different origin exhibit different absorbance at 232 and 270 nm. Table 2.12 includes such data reported by several scientists as summarized by Gracian (1969).

2.6 Chemical characteristics for the detection of adulteration of olive oil with seed oils 2.6.1 Iodine Value The determination of iodine value in oils and fats was first introduced by von Hubl (Baltes, 1964). This is a general reaction of substitution of the double bonds in fatty acids by halogens. The .reaction is sensitive to heat and light, and special precautions should be taken when it is performed.

The results are expressed as percent iodine absorbed whether or not the halogen used in the test is iodine (Eckey, 1954). The iodine value represents the degree of unsaturation of fats. On this basis, it could, and it has been, used as an indicator of the presence of seed oils which usually are more 33

Table 2.12. Absorbance at UV light for olive oils of different origins.

~ 1.50% acidity 1.50 - 4.0% acidity Absorbance at Absorbance at

Oil origin 232 nm 270 nm 232 nm 270 nm

Greek min 1.30 0.10 1.50 0.12 max 2.10 0.22 3.10 0.32

Italian min 1.50 0.08 1.50 0.10 max 2.49 0.32 2.90 0.36

Spanish min 1.75 0.06 1.81 0.11 max 2.53 0.22 2.95 0.32

Various countries except I tal ya min 1.40 0.10 1.55 0.15 max 2.25 0.21 3.40 0.42

Tunisia min _b 0.09 0.15 max 0.20 0.26

Adapted from Gracian, 1969. a Might include countries like Maroco, Libya, Algeria, Portugal and/ or France. bNo available data. 34 unsaturated, than olive oil. Table 2.13 gives the natural limits that have been observed in the iodine value of common vegetable oils. The sensitivity of the reaction to the factors mentioned above which could lead to erroneous quantitative results, as well as the broad variation that olive oils from different countries show (Table 2.14), are probably the main reasons for the limited use of this test in the olive oil quality control laboratories as a tool for detecting adulterations.

2.6.2 Saponification Value The saponification value, together with the iodine value, is the most important chemical constant in fat analysis (Baltes, 1964). The saponification value denotes the amount of potassium hydroxide which is needed to saponify one gram of fat or oil, i.e. to neutralize the free fatty acids and the ones combined as glycerides (Cocks and van Rede, 1966). The saponification number is a measurement of the molecular weight of lipids (Sonntag, 1979) and as a general rule, the higher the molecular weight the lower the saponifi­ cation value. The Food and Agriculture Organization has recommended limits for the saponification value of edible oils. Such data are given in Table 2.15. It is obvious from the figures of this table that it is almost impossible to detect such seed oils in olive oil on the basis of the saponification value but this constant has been used in the past as an additional tool, along with other constants, for verifying the identity of an olive oil sample.

Even if the country of origin of an olive oil sample is known (Table 2.16) a definite conclusion on the purity of an olive oil sample can not be reached. However, the test is more useful in revealing adulterations with oils containing distinct fatty acids like, coconut oil (Gracian, 1969), which has a 35

Table 2.13. Iodine Values (I.V) of some vegetable oils.

Oil Type Lv.a

Peanutb 80-106

Cottonseedc 99-113 Oliveb 75-94 Canolad 110-126 Cornb 103-128 Sunflowerb 110-143 Soybeanc 120-141 aBy Wijs method. bFAO/WHO (1969 -1970). CFormo, 1979. dvaisey-Genser and Eskin, 1982. 36

Table 2.14. Iodine Value limits for olive oils.

Sample source Lv.a

Greece 78-90

Italy 79.5-84 Portugal 76-88 Tunisia 80-92 Spain 78-88 Algeria 74.6-93 Argentina 80.4-87.4 Morocco 84-93

Adapted from Gracian, 1969. aBy Hanus, Wijs, or Hubl method. 37

Table 2.15. Saponification Values (S.V) of some vegetable oils.

Oil Type s.v.a

Peanutb 187-196

CottonseedC 189-198 Oliveb 184-196 Canolad 188-193 Cornb 187-195 Sunflowerb 188-194 Soybeanc 189-195 amg KOH/ gr oil. bFAO/WHO (1969 -1970). Cformo, 1979. d Vaisey-Genser and Eskin, 1982. 38

Table 2.16. Saponification value limits for olive oils.

Sample source s.v.a

Greece 188-194 Italy 188-195 Portugal 188-192 Tunisia 188-194 Spain 184-194 Argentina 189-194 Morocco 184-189

Adapted from Gracian, 1969. amg KOH/ gm oil. 39 saponification value ranging from 248 to 264. Some of the general drawbacks of this method are that secondary reactions may take place leading to a reduction of the saponification value (Dimoulas, 1981). Depending on the lipid material tested, and because of problems with the reagents and the method, difficulties and errors may be introduced (Baltes, 1964).

2.6.3 Bellier Number The Bellier number "evolved" from the test introduced by Bellier at the end of 19th century for the detection of groundnut oil in olive oil. Later it was adopted as a new constant in fats and oils (Gracian, 1969). It has been defined as the temperature at which the precipitation of the fatty acid of an oil starts, after the oil has been saponified and exists as a solution (Baltes, 1964). The precipitation is due to fatty adds with a long carbon chain. More specifically, it has been pointed out that the precipitation is produced by fatty acids with a high melting point and that the Bellier number increases in a linear way with the total content in saturated fatty acids (Gracian, 1969). The IOOC (1985) has adopted the recommendations of FAO/WHO (1970) and has set as a limit for the Bellier number in olive oils a temperature of :::; 17°C. The observed extreme values for Bellier Number in olive oils from different producing countries are given in Table 2.17. The method is highly empirical and oil from olive fruit with altered maturation due to weather conditions or from bad quality olives might give a higher Bellier Number due to the early precipitation of compounds from the unsaponifiable fraction of the oil, thus causing false interpretations (Gracian, 1969). 40

Table 2.17. Bellier Number (B.N.) limits for olive oils.

Sample source B.N.a

Greece 4-8.4 Italy 12-16 Portugal 10-15 Tunisia 13-18 Spain 3-10 Argentina 5.2-13.8 rcx:x:b ~17

Adapted from Gracian, 1969. arn °C. brooc,1985. 41

2.7 Determination of the chemical composition for the detection of adulteration of olive oil with seed oils 2.7.1 Analysis of the sterols The analysis of the unsaponifiable matter, especially of the sterol fraction, has been well studied in olive oil. This was the result of the early discovery that olive oil consist of pure ~-sitosterol and therefore the presence of other sterols would mean the presence of foreign oils in olive oil.

Today the situation has been altered mainly because of the progress in gas liquid chromatography. The analysis of sterols in packed columns resulted in the categorization of the sterols present in olive oils as three compounds. The main one is ~-sitosterol, campesterol and stigmasterol are present in low amounts and publications in the 1970's (Tiscornia and Monacelli, 1970; Gutfinger and Letan, 1974) concluded that an increase in either of the two latter compounds above a certain point should mean the presence of seed oils. Amati et al. (1971) reported A7 stigmastenol along with the previously three mentioned compounds and they also calculated the ratio of ~-sitosterol to campesterol plus stigmasterol in olive oils from Italy and Spain, as well as for seed oils. They concluded that as little as 5-10% of seed oils such as peanut, sesame, soybean, and rapeseed could be detected by the above calculations or from the presence of sterols absent in olive oils like brassicasterol from rapeseed oils.

Until 1985 the IOOC, in its standards for olive oils, specified the limits only for the two main sterols present in olive oil. Beta-sitosterol should be ;;:::

93% and campesterol ~ 4% as percentage of the sum of ~-sitosterol, campesterol and stigmasterol determined by glc with packed columns. 42

Common seed oils are characterized by an increased conten.t on campesterol (7-39%) and stigmasterol (1-19%), (Rossell, 1991).

Both high and low erucic acid rapeseed oils contain about 10% brassicasterol in the sterol fraction, a compound found basically only in these oils, resulting in an easy detection of as low as 2.5% of these oils in olive oil (Grob et al., 1990). However, working with standard methods (IUPAC, 1987; MAFF, 1990), in order to reveal adulteration of olive oil based on sterol composition, is very time consuming (see also page 44).

Progress on high resolution gas liquid chromatography with the use of capillary columns led to separation of other phytosterols in oils, and consequently reports included more information about sterol composition (Itoh et al., 1981; Akihisa et al., 1986; Grob et al., 1990). The well recognized advantages of capillary gas chromatography over packed columns such as higher efficiency, fast separation and better quantitation (Geeraert and Sandra, 1984) resulted in the initiation of several studies for adoption of a method regarding the analysis of sterols by capillary gas chromatography internationally. In 1987 Fedeli initiated studies on the analysis of sterols in olive oil by capillary gas chromatography (Firestone, 1988). A standard procedure for determining the sterol composition in oils was developed by Fedeli and colleaques (Firestone, 1989). The procedure was adopted in Italy as a standard method under the code NGD C72-1987 (Anonymous, 1987). In 1988 limits for sterol composition in different qualities of olive oil were proposed using either a SE 52 or a SE 54 glass capillary column (Table 2.18), (Subcommittee on Vegetable Oils, 1988). A collaborative study on the same subject was carried out by a technical committee in Italy in 1989 43

Table 2.18. Sterol composition (%) limits of olive oil by capillary glc.

Sterol Virgin olive oil (max)

Cholesterol 0.5 Brassicas terol 0.1 24 Methylencholesterol 0.5 Campesterol 3.8 Campestanol 0.5 Stigmasterol 1.8 !17 Campesterol 0.1 Clerosterol 1.2 ~-Sitosterol 65.0-88.0 Sitostanol 1.8 !15 A venasterol 6.0-30.0 !15,24 Stigmastadienol 1.0 !17 Stigmastenol 0.5 !17 Avenasterol 1.1

Adapted from Subcommittee on Vegetable Oils, 1988. 44

(Subcommittee on Vegetable Oils, 1989). Similar studies have been under evaluation by the IOOC since 1989, but final standards have not been published yet from this organization.

The limitations of sterol analysis as a means to detect adulteration of olive oil with seed oils are two fold. First, sterol composition can show a wide natural variation and second, isolating the sterols from the unsaponifiable matter and preparing them for GLC is a very time consuming procedure. A full day is often needed for the analysis of 1-3 samples, based on standard methods. Only recently a fast method allowing analysis of sterols in 30-35 samples per day was published, involving the use of automated coupled LC-GLC (Grob et al., 1990). However, equipment availability is a major concern in such methodology.

A different approach in order to simplify or avoid the complicated and time consuming sample preparation for sterol analysis by GLC, is the use of high performance liquid chromatography (HPLC). As early as 1976 a technique for the separation of phytosterols by Reversed Phase-HPLC appeared in the literature (Rees et al., 1976).

Numerous reports followed on this subject (Hunter et al., 1978; Thowsen and Schroepfer, 1979; Colin et al., 1979; Hansbury and Scallen, 1980; Rodriguez and Parks, 1982; DiBussolo and Nes, 1982; Holen, 1985; Goh et al., 1989; Warner and Mounts, 1990) but in general the extraction of the unsaponifiable matter and/ or the derivatization of the sterols were not avoided in most of them. Additional problems included the poor absorption 45 of sterols in ultra-violet light, if a UV detector is utilized, and refractive index detectors have been shown to have poor sensitivity.

2. 7.2 Analysis of the fatty acids For the determination of the fatty acid composition of oils, gas liquid chromatography of the fatty acid methyl esters is the method of choice (Kinsella et al., 1975) and it has replaced the chemical methods previously used. All standard methods existing to date specify the use of packed columns (IUPAC, 1987; AOCS, 1989). However, capillary glc is now so routine, that it is difficult to get an instrument accommodating only packed columns (Rossell, 1991).

In vegetable oils, the main fatty acids are those with 18 and 16 carbon atoms. However, other fatty acids are also present in small amounts and the presence of some specific ones could be used as a way of identifying a particular oil. Since qualitatively most vegetable oils contain the above mentioned fatty acids, identification of the oils is usually based on quantitative data. As soon as 1965 there were published limits for the fatty acid composition of vegetable oils (Pallota, 1976). Analysts have used such data extensively for oil purity studies.

The fatty acid composition of olive and many other common oils is presented in Table 2.19. Olive oil is characterized by a high content in oleic acid (C18:1). The other main fatty acids in this oil are palmitic (C16:Q), linoleic (C18:2) and stearic (C18:Q). In seed oils like corn, sunflower, soybean, cottonseed, peanut and safflower the main fatty acid is linoleic acid, followed by oleic, palmitic and stearic. In soybean oil linolenic acid (C18:3) is also Table 2.19. Fatty acid composition of some vegetable oils.

Fatty acid percentage (w /w)

Oil type C16:0 C16:1 C18:0 C18:1 C18:2 C18:3 C20:0 C20:1 C22:0 C22:1 C24:0 C24:1 C14:0a- Olive _b 12.1 1.0 2.7 71.8 10.2 0.8 0.5 0.3 Corn - 12.2 0.2 2.3 36.9 45.9 0.9 0.6 0.3 0.2 - 0.2 Sunflower 0.1 6.1 0.1 5.6 19.3 67.0 0.1 0.4 0.1 0.9 - 0.3 Soybean 0.1 9.5 0.1 4.9 21.9 52.6 7.9 0.5 0.2 0.5 - 0.2 Cottonseed 0.8 23.0 0.6 2.3 15.6 55.6 0.3 0.3 0.1 0.2 Peanut 0.1 9.5 - 2.3 45.6 31.0 - 1.4 1.2 2.7 - 1.3 Safflower 0.1 6.4 0.6 2.5 11.9 73.3 0.5 0.5 Rapeseedc 0.1 2.5 0.4 0.9 11.2 12.8 8.6 0.7 7.3 - 48.1 Rapeseedd - 4.8 0.5 1.5 53.2 22.2 11.0 0.6 1.0 - 0.2 Palm 1.1 43.7 0.1 4.5 39.3 10.1 0.2 0.3 0.2

Adapted from Sheppard et al., 1978 and Van Niekerk and Burger, 1985. aNumber of carbon atoms : number of double bonds. bLess than 0.1 %. CHigh erucic acid rapeseed oil. dLow erucic acid rapeseed oil.

46 47 present in considerable amounts (7.5%).

However, such data give only general information and cannot be used in identification studies. When the analyst wants to confirm if a sample of olive oil is adulterated with seed oils the range over a particular fatty acid must be known. When the oil's growing area and country of origin are unknown, limits established internationally should be used which are usually wider than those accepted for a specific country. For example, if a sample of olive oil of Greek origin is analyzed, then the limits set for Greek olive oils could be used, which are narrower than the international limits (Table 2.20). The Greek Code of , Beverages and Objects of Common Use (1975) specifies that a linoleic acid content of less than 9% is a safe indication for the absence of seed oils. Other difficulties arise from olive oils that, due especially to climatic conditions, show abnormally high contents of linoleic acid (e.g., 30%) (Rana and Ahmed, 1981).

Adulteration of olive oil with common seed oils is often revealed by an increased content in linoleic acid. For adulteration with soybean and rapeseed oils, the content of linolenic acid also should be measured since these two oils contain considerable amounts of this fatty acid. Cottonseed oil might be detected by the use of myristic acid (C14:0) content. Cottonseed oil contains approximately 1% myristic acid whereas, all other common seed oils contain approximately 0.1 %. Contrary to the other common seed oils, peanut oil contains higher amounts of arachidic (C20:0), gadoleic (C20:1), behenic (C22:0) and lignoceric (C24:0) acids (Letan et al., 1965). These fatty acids, even in amounts lower than 2-3%, could be useful in checking the presence of peanut oil in olive oil. Behenic acid is especially high in peanut oil, up to 5%, 48

Table 2.20. International fatty acid composition limits for olive oil.

Fatty acid Percentage

C14:0a 0.0-0.1 C16:0 7.5-20.0 C16:1 0.3-3.5 C18:0 0.5-5.0 C18:1 56.0-83.0 C18:2 3.5-21.0 C18:3 0.0 -1.5 C20:0 max 0.8 C20:1 m.a. b C22:0 max 0.2 C22:1 n.d.c C24:0 max 1.0 C24:1 n.d.

Adapted from IOOC, 1985 and FAO/WHO, 1970. aNumber of carbon atoms: number of double bonds. bMinute amounts CNot detected. 49 whereas the limit for this particular fatty acid in olive oil is 0.2%. The presence of these long chain fatty acids is also the basis of empirical tests for detecting peanut oil such as the Bellier and the Evers tests (Cocks and van Rede, 1966).

Rapeseed oil was initially easy to identify because of its high content of erucic acid (C22:1). This oil, also known as high erucic acid rapeseed oil (HEAR), has a content of C22:1 ranging from 30 to 55%. This fatty acid is found mainly in plants of the genus Brassica (Ackman, 1983) and therefore, its appearance in the fatty acid chromatogram of an olive oil indicates the presence of oil from the above mentioned plants.

However, genetic alterations of rapeseed led to production of low erucic acid rapeseed (LEAR). The first such varieties, licensed in Canada in 1973, were "Torch" and "Midas", containing 0.44 and 0.02% erucic acid respectively. In 1979 the trademark "canola" was introduced which differed from LEAR only in the level of glucosinolates, undesirable odorifous compounds in rapeseed feed (Carr, 1991). The first variety licensed as canola was ''Tower" with 0.02% erucic acid (Daun, 1983).

Through the last two decades a dramatic change in the cultivation of rapeseed varieties has been observed in Canada with a switch to LEAR and canola crops (Daun, 1983). The switch from HEAR to LEAR oil resulted in an average content in C22:1 of less than 5% in 1974 and as low as 0.0 to 1% in 1980's (Daun, 1983). 50

This decrease in erucic acid was accompanied by a substantial increase in oleic acid (Vaisey-Genser and Eskin, 1982), thus bringing canol a in the family of high-oleic acid oils. For naturally found vegetable oils, such a high content in oleic acid is common only for olive oil and maybe hazelnut, teaseed and groundnut oils. Some new safflower varieties contain high amounts of oleic acid but these are also products of genetic engineering like canola.

Overall, the fatty acid composition of LEAR or canola oil is similar to olive oil with the exception of linolenic acid which has not undergone a significant change from the HEAR oil (appr. 10%), (Ackman, 1983). The linoleic acid content of canola (also relatively unchanged in HEAR and LEAR oils) can be within the limits for olive oil, and generally ranges from 18 to 30% (Vaisey-Genser and Eskin, 1982).

The situation might change in the future since the ultimate goal of breeders is 0% erucic acid, about 30% linoleic acid and less than 3% linolenic acid (Ackman, 1983). The decrease in linolenic acid is desirable because of the high susceptibility to oxidation related to this polyunsaturated fatty acid (Labuza, 1971; Prevot et al., 1990). Such a future modification will most probably make glc analysis of fatty acids more ineffective in detecting canola in olive oil since the content of linolenic acid seems to be the only possible way to reveal such an adulteration. As previously mentioned, linoleic acid is low in canola and the other fatty acids present (myristic, palmitic, palmitoleic, stearic, arachidic, gadoleic, behenic and lignoceric acid) are low enough or even lower than in olive oil to be used as indicators of adulteration. Gadoleic acid (C20:1) is quite high in HEAR oils (4-10%) but it has been dramatically 5 1 reduced in LEAR and canola oil. Fatty acids with 20 and 22 carbon atoms show a good correlation in HEAR and LEAR oils and regression equations have been used to estimate total C2o fatty acid content when C20:1 coelutes with C18:3 (Ackman, 1983).

2.8 Analysis of Triacylglycerols 2.8.1 Triacylglycerols Triacylglycerols are the main storehouse of fats in animal and plant tissues (Labuza and Erdman, 1984). They consist of glycerol, a trihydroxyl alcohol, esterified with· three fatty acids. The generic chemical structure of triacylglycerols is given below:

where R1, R2 and R3 are the same or different fatty acid moieties (Zapsalis and Beck, 1985). The glycerol carbons are numbered as 1,2,3 or designated as a,B,a' (Zapsalis and Beck, 1985). Simple triacylglycerols contain the same fatty acids in all three positions. When two or three different fa tty acids are present the triacylglycerols are known as mixed. 52

As far as the nomenclature of triacylglycerols is concerned the stereospecific number (sn) system has been most widely accepted. According to this system the middle hydroxyl group is written on the left side of the central carbon atom of the glycerol molecule. In this case the carbon atom on the top is designated as C-1 (IUP AC, 1967). If the prefix sn precedes the term glycerol then the exact positional distribution of the fatty acids is known. In simple triacylglycerols and in triacylglycerols with unknown positional distribution of the fatty acids the prefix sn is not used, e.g., the form StOP indicates a possible mixture of sn-StOP and/ or sn-OStP and/ or sn-OPSt and/ or sn-POSt and/ or sn-PStO and/ or sn-StPO (Nawar, 1985). Examples of simple and mixed triacylglycerols are given below:

tris tearoy lgl ycerol 1-stearoyl-2palmitoyl-3- myris toy1-sn -glycerol

The kind and the distribution of fatty acids in the triacylglycerol molecule is of particular interest for the analyst as it influences their separation by chromatographic methods. 53

In plants, it is believed that fatty acids are distributed in the triacylglycerol molecule according to the 1,3 random-2 random theory (VanderWal, 1960) i.e., composition at position 2 is different from positions 1 and 3 but composition at 1 and 3 is the same (Nawar, 1985). According to this theory the percentage of a triacylglycerol XYZ can be calculated from the equation (Nawar, 1985):

%sn-XYZ =[mol% X at 1,3] x [mol% Y at 2] x [mol% Z at 1,3] x 1o-4

It has also been observed that the more unsaturated fatty acids preferentially occupy the 2 position, whereas saturated fatty acids prefer the 1 and 3 positions. Some exceptions do exist, as in the case of and other unusual fats (Fedeli and ]acini, 1971).

Until recently little attention was paid to the direct use of triacylglycerols for identification purposes, mainly because of their complexity in vegetable oils and their poor and/ or difficult separation by existing techniques. The introduction of reversed-phase HPLC in the analysis of lipids, as well as newly developed stationary phases in capillary gas-liquid chromatography, have resulted in the utilization of triacylglycerol patterns for identification of vegetable oils.

2.8.2 Gas-liquid chromatography (glc) of triacylglycerols Gas chromatography on non-polar columns separates triacylglycerols strictly according to carbon number (CN). The technique is rapid and a simple dilution of the oil in an appropriate solvent is enough for introducing the sample in the chromatograph (IUPAC, 1987). However, triacylglycerol 54 composition of natural oils is usually very complex and unfortunately glc profiles alone do not reflect much information about the oil and cannot be used for differentiation between oils. Nevertheless, the method became a well established procedure especially in the field of confectionery fats (Kuksis et al., 1963; 1964).

Litchfield et al. (1967) illustrated the partial separation of triolein and tristearin (both with a CN of 54) after optimization of the operating conditions. Nevertheless, until the early 1980's, little was done with this procedure. It was known that the use of polar packings could improve resolution but such columns suffer from instability in high temperatures (Grob Jr. et al., 1980). In the late 70's and early 80's the use of capillary columns attracted attention due to shorter analysis times, better quantitation and improved separation (Geeraert and Sandra, 1984). Optimization of the method has resulted in the separation of triacylglycerols not only by CN but also according to the number of unsaturated fatty acids (NUFA separation), (Geeraert and Sandra, 1984). No information was obtained for the number of double bonds within the fatty acid (NUFA for 000, LLL and OLLn is 3) and the exact location or configuration of the double bonds.

In 1985, Geeraert and Sandra introduced a phenylmethyl-silicone ("polarizable") capillary column stable at 370°C and illustrated the highest resolution of this technique compared to simple CN or NUFA gas chromatography and RP-HPLC of vegetable oils (Geeraert and Sandra, 1985).

Chaves Das Neves and Vasconcelos (1989) using an OV-17-0H capillary column, characterized vegetable oils by a computerized pattern analysis of the 55 triacylglycerols. This is the only work on direct triacylglycerol analysis by glc that has been carried out for the detection of adulteration of olive oil.

2.8.3 Combined techniques for the analysis of triacylglycerols Most natural fats have a very complex triacylglycerol composition which cannot be completely resolved by a single analytical method (Litchfield et al., 1964). So far the most widely used technique is a combination of silver nitrate (argentation) chromatography and glc. Argentation chromatography separates triacylglycerols due to formation of a reversible complex between double bonds and silver ions (Barret et al., 1963). It was introduced by both de Vries (1962) in the form of silver nitrate column chromatography and by Barret and colleaques (1962) in using thin-layer chromatography (TLC). Both methods have been used alone or with densitometry (Chobanov et al., 1976) or with analysis of the separated triacylglycerols and/ or the fatty acids by glc (Jurriens and Kroesen, 1965; Litchfield et al., 1964; Culp et al., 1965; Litchfield, 1968) for quantitation purposes. Triacylglycerols are separated according to number of double bonds but separation of geometrical isomers is also obtained (Gegiou and Georgouli, 1983).

Argentation TLC with subsequent fatty acid and total triacylglycerol analysis by glc, with or without the determination of the fatty acid composition at 1,3 and 2 positions by enzymatic techniques (Dutta et al., 1978; Pan and Hammond, 1983), has been used for studying the triacylglycerol composition of vegetable oils, including olive oil, in detail.

Damiani and Burini (1981) used argentation TLC, oxidation techniques and glc to determine the triacylglycerol composition of olive oil. During this 56 work 21 molecular species were detected accounting for 92.5 % of the total triacylglycerols. The main species found were 000 (45.6%) and POO (20.1 %). Minor triacylglycerols present were OOL, OLO, StOO, POP, PLO, and POL with 7.2, 5.1, 3.6, 2.8, 1.9, and 1.5% respectively. Later, Descargues and Bezard (1981) used a combination of argentation tlc and glc of fatty acids and triacylglycerols to determine the triacylglycerol composition of a commercially available olive oil. Their results were not very different from previous studies (Damiani and Burini, 1981; Fedeli and Jacini, 1971). Most studies agree on the absence of trisaturated triacylglycerols from olive oil.

Similar techniques have been also used as an aid to detect adulteration of olive oil with seed oils. Galanos et al. (1968b) studied the use of argentation TLC for detecting such adulterations. As low as 2.5% of some seed oils (cottonseed, corn and sesame oils) were detected. Their method has been adapted by the Greek Code of Foods, Beverages and Objects of Common Use (1975). A combination of low temperature crystallization, argentation TLC and glc analysis of fatty acids could detect as low as 10% cottonseed oil in olive oil (Synouri-Vrettakou et al., 1984), but a long overall analytical time was needed ( -48hrs).

The main problem in methods involving TLC is quantitation. Moreover, a relatively large amount of sample is required as well as time consuming experimental procedures in order to completely separate complex lipids, especially when each spot must be identified and quantified (Privett and Erdahl, 1975). These problems have been the main stimulus for the great attention that HPLC has received as a replacement of TLC (Hammond, 1981). HPLC combines greater resolution and efficiency and better sensitivity. 57

2.8.4 Triacylglycerol analysis by reversed-phase high performance liquid chromatography (HPLC) 2.8.4.1 Theoretical aspects of HPLC Modes of separation - Chromatography is any separation technique where components of a sample partition themselves between a moving phase and a stationary phase. The moving phase can be a gas or a liquid and the stationary phase can be a solid or a liquid. A combination of gas as the mobile phase and a liquid for stationary phase, is known as gas-liquid chromatography, and a liquid mobile phase and a solid or liquid stationary phase are known as liquid-solid or liquid-liquid chromatography.

Listed below are the main modes of separation in liquid chromatography with emphasis on reversed-phase chromato-graphy. The information sited is based on discussions by Pomeranz and Meloan (1978b), Lim (1986) and Macrae (1988).

Adsorption chromatography. This separation mode is also known as liquid-solid chromatography. The solutes interact with the mobile phase and the adsorption sites of the solid phase, mainly due to polarity effects. Silica gel is the most commonly used stationary phase and by increasing the polarity of the mobile phase the retention of the solute will be decreased (normal phase). The interaction of uncharged solutes with the mobile phase is due to dispersion, hydrogen bonding and dipole effects.

Partition chromatography. In this mode the stationary phase is not a solid but a liquid supported by a solid phase, usually silica. The basis of retention of solutes by the column is solubility and the column is prepared by 58 coating the stationary phase on the support. Unfortunately column bleeding is inevitable due to solubility of the stationary phase in the solvents used. This is the main reason for the replacement of this technique by stationary phases chemically bonded on the solid support. Polar stationary phases like amino, cyano, and nitro are used in normal phase partition chromatography for the separation of relatively polar compounds.

In the reversed-phase mode of this technique the chemically bonded stationary phase is nonpolar compared to the relatively polar mobile phase. The most popular reversed-phase packing is a hydrocarbon chain chemically bonded to silica. Separation is based on hydrophobic interactions. However, some silanol groups will not be bonded with hydrocarbons and therefore, adsorption effects might occur. Since the stationary phase is hydrophobic, more polar compounds (such as free fatty acids, diacylglycerols and monoacylglycerols) will be more soluble in the relatively polar mobile phase and thus, will be eluted earlier than less polar compounds such as triacylglycerols.

Hydrocarbons ranging from one carbon atom to twenty-two carbon atoms have been used for this technique, but the Cg and C18 hydrocarbons are the most widely employed. For chromatography of lipids, and particularly for the analysis of triacylglycerols, nonaqueous organic solvents of different polarities are used as the mobile phase because of the poor solubility of triacylglycerols in water. Some of the commonly used solvents are methanol, acetonitrile, acetone and tetrah ydrofuran. 59

Parameters important in liquid chromatography also apply in HPLC, since the latter has evolved from the former performed in large glass columns. HPLC is the result of improvements, especially in the speed, of column chromatography (Johnson and Stevenson, 1978).

The retention time (tR) of a solute in the column reflects the partitioning of the solute between the mobile and the stationary phases. Solutes that show a higher affinity for the mobile phase will participate in fewer partitioning equilibria and will elute earlier. Therefore, the tR for such components will be smaller. The retention time of an unretained solute (usually the solvent for solubilizing the sample) is denoted as tM and it is also known as dead space or void volume. No solute can elute before this time.

One of the most important equations in chromatography is that expressing the partition coefficient k. This is an expression of the equilibrium distribution of a solute between the two phases and is equal to

k=cs/cm where cs the concentration of the solute in the stationary phase and em the concentration of the solute in the mobile phase.

The capacity ratio, k', is a measure of a solute's retention and is given by the equation:

k' = tR- tM I tM 60

The unretained solute has a k' = 0. A k' of 1 means that the solute is eluted very fast whereas, higher values of k' denote a long time for the elution of the solute from the column. Low k' means fast analysis but for good separation a high k' is required. In general, a value between 1 and 10 is considered as optimum.

The separation factor, a, or relative retention or selectivity for two solutes is expressed by the following equation:

a= k'2 I k't = k2 I kt = tR2- tM I tR1 - tM

Selectivity is a measure of the ability of a column to separate the solutes 1 and 2. An a= 1 means no separation between the solutes 1 and 2.

The theoretical plate number , N, is a measure of the band spread of a peak when a solute goes through a column:

N = 16 (tR I w )2 where w is the width of the peak. In other words, N is the number of times that a solute partitions between the two phases.

The plate height, H, is defined as the distance that the solute moves within one partition:

H=LIN 6 1 where Lis the length of the column. The two last parameters are important measurements of the column efficiency.

The resolution of two adjacent peaks is given by the formula

or

Rs = 1 I 4 v' N (a- 1 I a) (k' I k' + 1) A Rs of 1 denotes about 3% overlap of the two peaks. A Rs of 1.5 indicates almost complete resolution (only 0.2% overlap).

Changes in the stationary phase, the mobile phase, the temperature and the pressure affect k' and a. Therefore, resolution and efficiency of a particular chromatographic separation can be optimized.

2.8.4.2 The development of reversed-phase high performance liquid chromatographic analysis of triacylglycerols In RP-HPLC, where a non-polar stationary phase, consisting of a hydrocarbon chemically bonded to the surface of silica particles, and a relatively polar mobile phase is utilized, triacylglycerols are separated according to molecular weight and degree of unsaturation. These two important parameters are included in the equivalent carbon number (ECN) concept (Plattner et al., 1977). The term ECN originated from the term equivalent chain length as it was used in the fatty acid analysis by glc (Miwa et al., 1960). ECN or partition number (Petersson et al., 1981) or integral partition number (Bezard and Ouedraogo, 1980) is defined as the number of 62 carbon atoms (excluding those of the glycerol) minus two times the number of double bonds (Plattner et al., 1977):

ECN = CN - 2 x DB where CN = carbon number and DB = double bonds. Each double bond has been found to be "equal" with two carbon atoms as far as the behaviour of triacylglycerols in RP-HPLC is concerned (Aitzetmuller, 1982).

One of the problems in separating triacylglycerols in RP-HPLC is the existence of "critical pairs", i.e., those. triacylglycerols with the same ECN but with differences in carbon chain length, number of double bonds and/ or positional and geometrical configuration, eg., 000, PPP, sn-POP, sn-PPO all with an ECN of 48 (Shukla, 1988).

Plattner and co-workers (1977) first observed the partial separation of critical pairs when mobile phases containing methanol were used. The same was not true for solvent systems based on acetonitrile where a baseline separation of triacylglycerols differing only by one double bond or by two carbon atoms could be obtained (simple separation according to ECN). They concluded that the use of methanol results in complex chromatograms so they discontinued the use of this solvent.

What was "difficult to interpret" for the above workers, was characterized as more informative about the triacylglycerol composition of an oil from Herslof and colleaques (1979). These workers found, however, that methanol and not acetonitrile gives simple chromatograms without 63 separation of critical pairs. These opposite results were attributed to different column packings between the two groups. The main reason for partial separation of critical pairs was due to mobile phase modifications but changes in column efficiency could also result in such separations (Petersson et al., 1981).

El-Hamdy and Perkins (1981b) introduced the concept of theoretical carbon number (TCN). The TCN for any triacylglycerol was determined from a plot of capacity factor (k') against carbon number and it was equal to actual carbon number for saturated triacylglycerols. For unsaturated triacylglycerols the TCN is also calculated from the following formula:

TCN = ECN - (Li3 Ui) where Ui is a factor determined experimentally for fatty acids and Li3 Ui is the total Ui of the individual fatty acid present in the triacylglycerol molecule. For saturated fatty acids Ui is 0.0 and ranges between 0.2 to 0.8 for elaidic, oleic and linoleic acids. For 000, POO and POP, all triacylglycerols with an ECN of 48, the TCN are 46.2, 46.8 and 47.4 respectively and therefore, their separation in RP-HPLC was possible after optimization of the operating conditions (Phillips et al., 1984a).

A different approach for the identification of peaks in triacylglycerol analysis by RP-HPLC has been based on "exact" ECN values. These can be obtained by adding two saturated triacylglycerols in the analyzed sample (Podlaha and Toregard, 1982). One of them should elute before and the other after the main part of the chromatogram. The retention times (tR) of these 64 two triacylglycerols are used to create the relationship between logarithms of tR's and carbon number. This relationship is linear and the retention times of unsaturated triacylglycerols can be used for identification of the ECN. The relationship between ECN and CN was also found to be linear for both propionitrile and acetonitrile/ acetone based mobile phases (Podlaha and Toregard, 1982). Expression of the retention times of triacylglycerols with respect to triolein has also been suggested for identification purposes (Wolff et al., 1991).

Columns. Improvements in column technology have led to an enormous growth of RP-HPLC in the area of fats and oils separations since the poor reports on the analysis of lipids by HPLC by Privett and Erdhal (1975) and Myher (1978) where it was noted that HPLC had not been applied to lipids because of column efficiency problems and lack of appropriate detectors. Today, the use of HPLC (especially on the reversed phase mode) is well established for lipid analysis (Shukla, 1988).

In the earliest report on RP-HPLC of triacylglycerols the column used was packed with a support of 35-44 J.lm (Pei et al., 1975). During this work the elution of trilaurin (ECN 36) was achieved in -8 minutes and extensive band broadening was observed. Packings with a particle size between 5-10 Jlm considerably improved the separation of triacylglycerols and decreased the total time for analysis (Vonach and Schomburg, 1978; Plattner et al., 1978; Plattner and Payne-Wahl, 1979; Payne-Wahl et al., 1979; Herslof et al., 1979; Bezard and Ouedraogo, 1980; Smith et al., 1980; Lie Ken Jie, 1980). The effect of column packings on the separation of triacylglycerols by RP-HPLC was studied by El-Hamdy and Perkins (1981a). During this work, a particle size of 65

5 J.Lm was found to be the most efficient. Due to the· smaller particle size a smaller sample could be used which in turn resulted in better resolution in less time. The longer alkyl chain (C18) bonded to silica resulted in better separation of triacylglycerols compared to octyl (C8) or methyl groups. Short columns (10 em) packed with 3 J.Lm alkyl bonded silica particles have given very high resolution of triacylglycerols with short analytical times (Dong and DiCesare, 1983; Schulte, 1982). Complete elution of olive oil triacylglycerols was achieved in 8 minutes with excellent separation of several critical pairs (Dong and DiCesare, 1983) compared to 45 minutes when a 5 J.Lm octadecyl bonded silica column was used (El-Hamdy and Perkins, 1981b).

Temperature effects. Operating at 14.5°C Jensen (1981) illustrated improved separation of triacylglycerols with the same ECN but different degree of unsaturation. However, low solubility of saturated triacylglycerols such as tristearoylglycerol (SSS) in the mobile phase due to low temperature results in band broadening. Using propionitrile instead of the mobile phase used by Jensen (acetonitrile/tetrahydrofuran/hexane) Geeraert and De Schepper (1983) achieved even better resolution of critical pairs, again at subambient temperatures (14°C).

Solubilizing agents. The effect of the injection solvent has been studied by Tsimidou and Macrae (1984). A considerable influence on the analysis of triacylglycerols was found depending on the injection solvent used. Chloroform has been used extensively as injection solvent (Peterrson et al., 1981; Jensen, 1981; Takahashi et al., 1984; Kapoulas and Andrikopoulos, 1986; Takahashi et al., 1986) but it was found to produce inferior separation when large injection volumes (10-20 J.Ll} were used. 66

Acetone gave better overall chromatograms, however poor separation was obtained for saturated triacylglycerols when acetone was used as the solubilization agent. To overcome such a problem, a mixture of acetone and tetrahydrofuran (Shukla, 1988) or acetone and chloroform (IUPAC, 1991) has been suggested.

The mobile phase. Optimization of the mobile phase depends on the particle size and the nature of the alkyl bonded chain of the column packing (El-Hamdy and Perkins, 1981a). The choice of the mobile phase depends on the kind of sample analyzed. Increased or decreased solubility of the triacylglycerols in the mobile phase will greatly affect their separation. Crystallization of saturated triacylglycerols in the column will result in a decrease of it& life (Tsimidou and Macrae, 1987). A decrease in the polarity of the mobile phase will result in an increased partition of the triacylglycerols in the mobile phase and thus, faster elution. However, the resolution will be decreased as a result of triacylglycerol overlapping.

In the earliest application of RP-HPLC for triacylglycerol analysis a mixture of methanol/water was utilized (Pei et al., 1975). However, use of this particular mobile phase does not appear in the literature after that work. Mixtures of acetone and acetonitrile in different proportions are the most widely employed mobile phase (Tsimidou and Macrae, 1984). This eluent was also suggested in a recently developed standard procedure (IUPAC, 1991).

A mixture of acetone and acetonitrile gives better resolution as compared to methanol/ acetone and the chromatograms are more informative (Herslof et al., 1979). However, this solvent combination can 67 lead to crystallization problems within the column as a result of insolubility of triacylglycerols with an ECN higher than 46 (Lie Ken Jie, 1980). Parris (1978) used tetrahydrofuran and acetonitrile or methylene chloride and acetonitrile in conjunction with an infra red (IR) detector. A combination of tetrahydrofuran and acetonitrile has also been used with ultra-violet (UV) detectors (Dong and DiCesare (1983) as well as with flame-ionization detectors (FID) (Phillips et al., 1984a; 1984b). This particular mobile phase has been shown to be an excellent eluent for high melting triacylglycerols (Thomas et al., 1988). A mixture of acetonitrile, 2-propanol and hexane has been used for separation of palm and soybean oil triacylglycerols (Herslof, 1981). A mobile phase consisting of acetonitrile, tetrahydrofuran, and hexane has been utilized for the analysis of olive oil triacylglycerols and tristearoylglycerol (Jensen, 1981). Superior separation of triacylglycerols has been obtained with propionitrile as the mobile phase (Schulte, 1981; Podlaha and Toregard, 1982; Geeraert and De Schepper, 1983; Fiebig, 1985). The use of dichloromethane with acetonitrile has also been reported as the eluent system in the analysis of olive oil (Pauls, 1983).

Acetonitrile has been used with ethanol when UV detectors were employed for analyzing lipids (Robinson and Macrae, 1984) with propionitrile and mass detectors (Myher et al., 1984), with ethanol/hexane again in conjuction with mass detectors (Herslof and Kindmark, 1985) and with chloroform and laser light scattering detection (Stolyhwo et al., 1985). A combination of acetonitrile, ethanol and isopropanol has also been used in conjuction with UV detection for the analysis of olive and seed oils (Kapoulas

and Andrikopoulos, 1986)~ 68

Finally, more complex mobile phases consisting of four or more organic and non-organic eluents have also been used in the study of triacylglycerol separation by RP-HPLC. El-Hamdy and Perkins (1981a) worked with a mobile phase consisting of methanol, acetone, acetonitrile and isopropanol. The simultaneous separation of phenolic antioxidants, toco­ pherols and triacylglycerols was obtained using water and phosphoric acid/ acetonitrile/methanol/isopropanol in the gradient mode (Andriko­ poulos et al., 1991).

From the above discussion it is obvious that the use of acetone is basically restricted to systems utilizing RI detection. Whenever another kind of detector was employed acetone is omitted from the eluting system used because of its incompatibility with these other detectors. From the commonly used solvents such as acetonitrile, acetone, tetrahydrofuran, hexane and dichloromethane only acetonitrile is transparent below 200 nm. Other factors such as cost and safety should also influence the decision for a particular eluting agent. Propionitrile, for example, is toxic and special precautions must be taken with its use (Tracey, 1986).

Detection systems. The lack of a universal detector (similar response for most compounds) (Johnson and Stevenson, 1978) in lipid analysis by HPLC has always been a problem. Ultraviolet detectors have been widely used but in general they are not considered appropriate for triacylglycerol analysis. Triacylglycerols do not have a chromophore and therefore, their absorption in UV light is low. However, the ester bonds in triacylglycerols do absorb at 220 nm and good quantitation might be possible at this wavelength (Herslof, 1981; Dong and DiCesare, 1983). Below 220 nm double bonds absorb 69 strongly and thus such wavelengths should be avoided (Shukla, 1988). The main drawbacks of UV detection in short wavelengths are the limited number of solvents that are transparent in this region of the electromagnetic radiation as well as the degree of their purity. Strongly absorbing impurities are also very common in the samples (Aitzemuller, 1982). The UV detection is based on Lambert-Beer's law, i.e., absorbance is directly related to concentration (Pomeranz and Meloan, 1978a), and therefore chromatographic conditions must be reproducible in order to obtain valid results. Working with monochromatic radiation and within the concentration where the response of the detector is linear is important (Macrae, 1988). Advantages of the UV detector include the use of gradient elution and its relatively high sensitivity.

Dissolved solutes in a mobile phase will change the refractive index. This is the basis of the RI detectors, which have been the most popular detectors for triacylglycerol analysis by RP-HPLC (Plattner et al., 1977; Bezard and Ouedraogo, 1980; Takahashi et al., 1986; IUPAC, 1991) despite their low sensitivity (20-40 JJ.g) and high interference by temperature and pressure changes (Macrae, 1988). Generally speaking, commercially available models detect as low as a 1o-7 change in refractive index (Johnson and Stevenson, 1978). Organic liquids will exhibit a change of 1o-7 in their refractive index when temperature changes are less than 1o-4oc and stability of the RI detectors is based on appropriate temperature control of both reference and sample cells (Johnson and Stevenson, 1978). Solvents such as chloroform, hexane and dichloromethane should be avoided because they possess rather high refractive indices which result in lower sensitivity (Tracey, 1986). 70

In an attempt to overcome the limitations of the RI detectors, Parris (1978) described the use of infrared (IR) detection in the analysis of triacylglycerols. Advantages of the IR detectors over RI include specificity, less need for temperature control and the use of gradient elution. However, the sensitivity of the IR detector is not better than that of the RI detector. Flexibility in the choice of the mobile phase is limited due to solvent interference and baseline drift is observed with gradient elution. Only a few reports on the use of this detector for triacylglycerol analysis appeared in the literature after the work of Parris (Payne-Wahl et al., 1981).

Flame ionization detection has also been applied in triacylglycerol analysis. The main concern with this type of detector is to evaporate the solvent prior to the detection of the solute (Johnson and Stevenson, 1978). Techniques 'for solving this problem led to the development of a number of different devices such as the the moving wire, the fused silica ring, and the woven quartz belt for the transportation of the eluted solutes (Macrae, 1988). Advantages of the FID are the use with gradient systems and the good quantitation (Phillips et al., 1984b). Nevertheless, this a destructive detector and collection of solutes for further qualitative and/ or quantitative studies is not possible (Johnson and Stevenson, 1978). In addition some difficulties with its operation do exist (Shukla, 1988).

The utilization of mass detectors (MSD) for on line use with HPLC has been suggested. The problem again is to remove the solvent without losing solutes prior to true detection (Johnson and Stevenson, 1978). The most commonly used technique is the removal of the mobile phase by evaporization after nebulization (Macrae, 1988). The tiny solute particles free 7 1 from solvents pass through a light beam and a photomultiplier tube detects light that is scattered by the solutes (Hammond and Irwin, 1988). Superior separation of butter triacylglycerols has been obtained with MSD compared to UV and RI detectors (Robinson and Macrae, 1984). Moreover, the MSD can be used with gradient elution if the solvents used are volatile at the specific evaporization temperature applied (Macrae, 1988). A drawback of the MSD is the loss of volatile components. This detector analyzes the total column eluent and its sensitivity has been found to be better than the sensitivity of the FID (Robinson et al., 1985) with a detection limit of less than 1 J.lg (Hammond and Irwin, 1988). Some of the disadvantages of the MSD are their destructive nature, the limitation in using explosive and/ or toxic solvents, their non-linear behaviour (Herslof and Kindmark, 1985) and maybe their higher cost (Shukla, 1988). Nevertheless, MSD show maximum resolution (Myher et al., 1984) and in general they have been characterized as promising universal detectors as far as reproducibility, sensitivity and quantitation are concerned (Herslof and Kindmark, 1985):

Stolyhwo and co-workers (1985) investigated the use of a detector also based on light scattering (LSD). The main difference was the utilization of a helium-neon laser as the light beam source. No baseline drift was observed when gradient elution was used. The detection limit was 1 ppm and more importantly, quantitative analysis without calibrations was in excellent agreement with data obtained by HPLC-RI detection and glc. The LSD also exhibited an excellent stability with a variety of mobile phases. 72

2.8.4.3 Quantitative analysis Because of the complexity of triacylglycerols in vegetable oils accurate quantitative determination is almost impossible by RP-HPLC. Good quantitative results could only be obtained when pure standards that had the same composition as the samples were employed (Lim, 1986). This is because of the possible different responses of the detectors to individual triacylglycerol species. Similarity of responses to all components of a sample is a desired characteristic in a universal detector. The most widely applied mode of quantitation has been based on the percentage of peak area or height as interpreted by electronic integrators. Lie Ken Jie (1980) found a very good agreement in quantitative results on coconut oil triacylglycerols obtained by RP-HPLC with RI detection and glc data. However, it is also known that saturated triacylglycerols of increasing molecular weight show a decreased peak height and area response (Lie Ken Jie, 1980). The same is true for the unsaturated triacylglycerols compared to the saturated ones (Herslof et al., 1979).

UV detectors have shown an extreme overestimation of unsaturated triacylglycerols along with an appreciable underestimation of the other compounds (Herslof and Kindmark, 1985). A linear relationship of the FID response to peak area of simple triacylglycerols found in vegetable oils has been observed by Phillips and co-workers (1984b). Based on peak areas directly, the coefficient of variation for minor components was - 7% and as low as 1% for major components. As previously mentioned, MS detectors have shown a non-linear response for triacylglycerols (Herslof and Kindmark, 1985; Robinson et al., 1985). Myher and colleaques (1984) pointed out that MS requires a better calibration than other methods since resolution 73 of individual triacylglycerols is more extended. They concluded that one of the main problems was contamination of pure standards by rearranged products. The response from the LS detector used by Stolyhwo and co­ workers (1985), even if it was not linear, was found to be the same for the majority of the triacylglycerols analyzed. A common factor of 1.7 was used for quantitation of all peaks. Subsequent tests on sunflower oil gave results in excellent agreement with data from other workers using HPLC-RI and glc techniques. Errors by using the same factor for oil samples of different origin were claimed to be less than 8%.

2.8.4.4 Other modes of HPLC in the analysis of triacylglycerols The utilization of other HPLC modes on the separation of triacylglycerols will be now discuss.ed in brief.

Normal phase HPLC. Separation of triacylglycerols in silica columns was illustrated by Plattner and Payne-Wahl (1979). Elution of triacylglycerols was in the opposite order of the RP-HPLC mode and both chain length and number of double bonds affected the chromatogram. Even if resolution was inferior to that of RP-HPLC in J.L-Bondapak C18 columns, the authors claimed that silica columns had higher capacities and that triacylglycerols were more soluble in the solvents used in normal phase HPLC. However, such applications are absent from the literature in the last ten years.

Argentation HPLC. Silver nitrate has been used in both mobile phases or column packings in order to improve separation of the unsaturated triacylglycerols. Addition of silver ions (0.01-0.2 N AgN03 or AgCl04) in mobile phases consisting of acetonitrile, tetrahydrofuran and dichloro- 74 methane (Vonach and Schomburg, 1978) considerably improved the separation of model mixtures of unsaturated triacylglycerols and vegetable oils. The reason for improved separation is the increased contribution of the double bond further than the simple two carbon effect on RP-HPLC. The most popular mode of argentation HPLC is the use of silica columns impregnated with silver ions. The method of separation in this system is identical to those already discussed regarding Ag+-TLC and Ag+-column chromatography (see page 55). Separation of positional isomers was also feasible by this technique (Hammond, 1981). Disadvantages associated with Ag+-TLC such as poor quantitation and small linear range (see page 56) can be overcome by argentation HPLC (Smith et al., 1980). Improved column technology has resulted in high-resolution separation of triacylglycerols within only twelve minutes. Quantitative analysis has also been improved and reproducibility can reach the one commonly observed in high-resolution capillary gas chromatography (Jeffrey, 1991). Argentation HPLC can be used in combination with carbon-number gas chromatography (Hammond and Irwin, 1988) or RP-HPLC (Takano and Kondoh, 1987) for complete analysis of triacylglycerols similarly to the earlier ·use of Ag+-TLC in conjunction with RP-TLC (Kuksis and Ludwig, 1966) or Ag+-TLC and carbon-number gas chromatography (Culp et al., 1965). Finally, the problems with Ag+ chromatography in general have been discussed by Shukla (1988) who pointed out that the slow growth of _this particular methodology is attributed to poor reproducibility of k' value, poor lifetime of columns, silver mirror formation on the detectors cells and the washing out effect of polar solvents on the silver. 75

2.8.4.5 RP-HPLC as a means for detecting adulteration of vegetable oils Despite the numerous reports in the literature regarding the separation of triacylglycerols by RP-HPLC in the last fifteen years little information is available on the utilization of this technique for detecting adulteration of vegetable oils.

Geeraert and De Schepper (1983) reported the use of RP-HPLC of triacylglycerols for a rapid detection of cocoa butter equivalents in cocoa butter and chocolate products. The presence of such substitutes changes the relative proportion of the triacylglycerols POP, POS, and SOS of pure cocoa butter. The method detects less than 5% of substitutes such as the palmid fraction of palm oil and shea nut stearine.

One of the earliest reports, concerning the detection of olive oil adulteration with seed oils by RP-HPLC appeared in 1986 (Kapoulas and Andrikopoulos, 1986). The chromatographic system used in that work included a UV detector and a mobile phase consisting of acetonitrile, ethanol, and isopropanol. Approximately 22-25 min were needed for total elution of the triacylglycerols. One of the most important conclusions of that work was that simple separation of triacylglycerols (incomplete separation of critical pairs) has potential for the fast detection of olive oil adulteration. The presence of 2.5% sunflower oil and 3-4% of other seed oils (soybean, corn, cottonseed) could be verified by visual inspection of the RP-HPLC chroma­ tograms. Their work also included quantitative data, the use of which could detect as low as 1.5% sunflower oil and 2% of other seed oils. One of the problems encountered with that particular system was the close elution of triacylglycerols with other non-triacylglycerol components that absorb 76 strongly and could have influenced both the qualitative and quantitative data.

In a later study the above seed oils plus rapeseed oil were detected definitely only at high levels (20%) in olive oil using special statistical techniques (Tsimidou et al., 1987). Moreover, the elution of the triacylglycerols with acetone/ acetonitrile (65/35 v /v) as the mobile phase was relatively slow with a total analytical time of approximately 40 min. Improved detection limits (5%) were obtained from the above group after crystallization of the oil samples at -22°C for twenty four hours and subsequent analysis of the liquid fraction by RP-HPLC (Tsimidou and Macrae, 1987). All seed oils analyzed (soybean, corn, cottonseed sunflower, rapeseed and groundnut oils) showed characteristic peaks in the region of the chromatogram where no significant peaks were present for olive oil. Detection of as low as 5% of soybean, corn, cottonseed and sunflower oils was possible. Detection of adulteration of olive oil with 5% rapeseed and groundnut oils was not conclusive. The additional step of the time consuming crystallization, even if it improves the lifetime of the column, is a major disadvantage for the aforementioned technique.

Casadei (1987) investigated the potential of triacylglycerol analysis by RP-HPLC for the detection of hazelnut oil in olive oil. Using the ratio of several triacylglycerol species he established the following purity criteria for olive oils: 77

OOL (46)1 I POL (46) =1.93 ± 0.14 000 (48) I POO (48) = 1.80 ± 0.12 SOO (SO) I PPO (48) = 1.50 ± 0.25 (OOL I POL) + (000 I POO) + (SOO I PPO) = 5.23 ± 0.43

Based on these limits, detection of ~ 25% hazelnut oil in olive oil was achieved.

A similar approach based on ratios of triacylglycerol species with the same ECN, might be useful in the detection of palm oil. Palm oil contains approximately 40% oleic acid, 45% palmitic acid and only 10% linoleic acid (Van Niekerk and Burger, 1985). Therefore the detection of adulteration of olive oil with palm oil would be difficult by the determination of the fatty acid composition. However, due to the high content of palmitic acid triacyl­ glycerol species such as POO and PPO predominate over 000 (all with an ECN of 48) and PLP and PLO predominate over OLO (all with an ECN of 46) (Dong and DiCesare, 1983). Therefore detection of adulteration of olive oil with palm oil could be feasible by RP-HPLC separation of the triacylglycerols. Nevertheless, no reference on this subject has appeared in the literature.

Flor and Martin (1989) established a limit of 0.5% trilinoleylglycerol for olive oils and illustrated the possible detection of polyunsaturated vegetable oils. Recently, Cortesi and coworkers (1990) described a method for detecting foreign oils in olive oil. The triacylglycerol composition of an olive oil sample is calculated based on fatty acid composition as determined by glc.

1Number in brackets denotes ECN. 78

These data are then compared to those from the analysis of the oil by RP­ HPLC. Deviations between the two sets of data are used to establish limits useful for detecting other oils in olive oil. By this technique the presence of 5% sunflower oil in olive oil was detected. This method, in conjuction with a standard method for the analysis of triacylglycerols by RP-HPLC (IUP AC, 1991), is currently under study by the International Olive Oil Council (IOOC, 1990; 1991). 3. MATERIALS AND METHODS

3.1 Oil samples and chemicals Genuine olive oil samples were obtained from a number of different sources. Samples Gl-10 were all of Greek origin. Samples Sl-3 were Spanish olive oils and I1 and I2 were Italian olive oils. They were all donated by Eleourgiki SYNPE (Athens, Greece). Samples GC1 and GC2 were obtained from a producer, after personal communication, from the island of Crete, Greece and they were produced from olives grown locally.

Samples CA1-5 were of California origin purchased from Nick

Sciabica & Sons, Modesto, CA, U.S.A. A description of the olive oils with guaranteed purity is given in Table 3.1.

Commercially available olive oil products were purchased in several retail stores in Saskatoon (samples SK1-3) and others (samples

1489-1566) were provided by Health & Welfare Canada, Winnipeg, Manitoba. Labelling information for these samples is given in Table 3.2.

Corn, sunflower and soybean oils were purchased in Saskatoon whereas refined, bleached and deodorized canola oil was dona ted by CSP Foods Ltd., Altona, Manitoba.

All solvents were of Omnisolv grade as supplied by BDH Inc., Saskatoon, Saskatchewan. Oleic acid (BDH Inc.) was of technical grade. Alumina (80-200 mesh) was supplied by Fisher Scientific, Edmonton,

79 80

Table 3.1. List of the olive oil samples with guaranteed purity.

Olive Oil Sample Category

Gl Greek Virgin G2 Greek Virgin G3 Greek Virgin G4 Greek Lampante GS Greek Virgin G6 Greek Virgin G7 Greek Virgin GB Greek Virgin G9 Greek Virgin G10 Greek Virgin GC1 Greek Virgin GC2 Greek Virgin 51 Spanish Virgin 52 Spanish Virgin 53 Spanish Virgin 11 Italian Virgin 12 Italian Virgin CAl California Virgin CA2 California Virgin CA3 California Virgin CA4 California Virgin CAS California Virgin 8 1

Table 3.2. List of the commercial olive oil samples.

Sample Label information

SK1 olive oil SK2 extra virgin olive oil SK3 light olive oil 1489 100% pure olive oil 1490 100% pure olive oil 1491 100% pure olive oil 1492 olive oil 1493 pure olive oil 1494 pure olive oil 1495 extra virgin olive oil 1496 extra virgin olive oil 1551 pure olive oil 1565 pure olive oil 1566 pure olive oil 82

Alberta. Fatty acid methyl ester and triacylglycerol standards were purchased from Supelco Canada Ltd., Oakville, Ontario and Sigma Chemical Company, St. Louis, MO, USA respectively.

3.2 Fatty acid analysis Methyl esters of fatty acids of all samples were prepared through acid methanolysis as described by Hitchcock and Hammond (1978). Approximately 50 mg of sample was mixed with 2 ml methylation reagent consisting of sulfuric acid, toluene and methanol in a 1:10:20 (v /v) ratio in an 8 ml screw cap culture tube. The mixture was heated to 105°C for 30 min and allowed to cool at room temperature. Two ml each of water and hexane were added and mixed with the use of a vortex (Scientific Industries Inc., Bohemia, NY, U.S.A.). The hexane layer was taken up by a pasteu·r pipet and further dried with sodium sulfate.

The fatty acid methyl esters were analyzed by both packed and capillary gas chromatography. The packed column was of stainless steel containing GP 3% SP-2310/2% SP-2300 on 100/120 Chromosorb W AW (Supelco Inc., Bellefonte, PA, U.S.A) in a dual column Hewlett Packard (model 5750) gas chromatograph with a flame ionization detector (FID). Temperature programming was used with an initial oven temperature of 150°C (1 min), followed by an increase of 8°C·min-1 to a final temperature of 240°C (4 min). The carrier gas was helium with a flow rate of 40 ml·min-1. The capillary column utilized was a 25m x 0.22 mm i.d. BPX 70 (0.25 J.tm film) (Rose Scientific, Edmonton, Alberta) connected to a Varian 3400 gas chromatograph (Varian Canada Inc., Georgetown, Ontario) with a flame ionization detector. Initial temperature was 160°C. The 83 temperature was initially increased to 170°C at 1 °C·min-1 and then to 220°C at 4°C·min-1. Finally a temperature of 240°C was reached at 10°C·min-1 and held there for 3 min. The head pressure was 20 psi, the split flow was 60 ml·min-1 and the split ratio was 80:1.

Quantitation was based on peak area % multiplied by appropriate relative response factors. Response factors were determined by analyzing fatty acid standards and dividing the amount of a particular fatty acid by its corresponding peak area. Relative response factors were established by dividing the response factor of individual fatty acids with the response factor of an internal standard - .heptadecanoic acid -which was analyzed along with the fatty acid standards. Fatty acid composition of all samples was expressed as weight percentages.

3.3 Refractive Index Refractive indices of all samples were determined (in triplicate) by an Abbe' refractometer according to IUPAC (1987) method 2.102. The calibration of the refractometer was checked with distilled water (RI = 1.330@ 40°C). Hexane was used for cleaning the prism of the instrument between each determination. All determinations were carried out at 40°C using a circulating water bath (Haake, Berlin, West Germany). The following formula was applied for converting results to 20°C : not = nott + (tt - t) F where no is the refractive index , t1 is 40°C, t is 20°C and F is equal to 0.00036. 84

3.4 UV Absorbance 3.4.1 Direct analysis The determination of the absorbance of UV radiation was carried out according to a method described by Gracian (1969). Approximately 100 mg of sample was weighed into a 10 ml volumetric flask and made to volume with cyclohexane. Absorbance at 232 and 270 nm was measured three times by either a Spectronic 21 (Bausch and Lomb, U.S.A) or a Spectronic 1201 (Milton Roy, U.S.A) using 1 em quartz cuvettes against cyclohexane as the blank. If the absorbance of a sample was outside of the range 0.1-0.8 the measurement was repeated with a more concentrated or diluted solution of the sample (MAFF, 1990). Results were expressed as 1% (w /v) oil solutions. Cloudy samples were filtered before analysis (Gracian, 1969) through a 12.5 em Whatman filter paper (no. 1) (Whatman Ltd., Maidstone, England).

3.4.2 Purification through alumina Virgin olive oil samples showing absorbance at 270 nm above 0.25 (set limit for characterizing an olive oil as virgin) were purified with alumina. Ten grams of sample were dissolved in 100 ml of hexane and the solution was passed through a glass chromatographic column containing a slurry of 30 g of activated alumina in hexane (MAFF, 1990). Polar secondary products of autoxidation were retained, whereas the eluent containing the unretained hydroperoxides was collected in a round bottom flask. The solvent was then evaporated with a rotary evaporator at 25°C. The residue was used for determination of its absorbance at 270 nm as described in 3.4.1. 85

3.5 Triacylglycerol analysis by reversed-phase HPLC Triacylglycerol profiles of all samples were obtained by using a Beckman model 342 HPLC system (Beckman Instruments Inc., Berkeley, CA, U.S.A) containing a 20J.1l injection loop, a micro-guard ODS-55 (40 x 4.6 mm ID) pre-column (Bio-Rad Laboratories, Richmond, CA, U.S.A) and a reversed-phase Supelcosil LC-18 (5 micron) (25cm x 4.6mm ID) column (Supelco Inc., Bellefonte, P A, U.S.A). An Altex model 156 refractive index (RI) detector (Altex Scientific Inc., Berkeley, CA, U.S.A) was used. Parameters of the Hewlett Packard 3390 integrator were as follows: attenuation 7, threshold 8, peak width 0.16 and chart speed 0.5.

All samples were 7.5% (w /v) solutions in acetone or chloroform and were filtered prior to injection through a 0.45Jlm nylon filter. Solvents and combinations of solvents including acetone, acetonitrile, tetrahydrofuran and hexane were degassed by filtration and agitation under vacaum through a millipore filtration device (Millipore Corporation, Bedford, MA, U.S.A) equipped with a 0.45 Jlm nylon filter (MSI, Westborough, MA, U.S.A) and used as the mobile phase.

The mobile phase was pumped isocratically at a flow rate of 1 ml per minute. Each day the mobile phase was allowed to pass through the system (including the reference cell of the detector) for at least two hours or until a stable baseline was obtained for approximately one hour.

Separation factors (a,) were calculated using the formula

a, = tR2 - tM I t&l - tM 86 where tR1 is the retention time of peak 1, tM is the retention time of the injection solvent and tR2 is the retention time of the adjacent peak 2.

Identification of peaks was based on the retention time of pure triacylglycerol standards (trilinolenylglycerol, trilinoleylglycerol, trioleyl­ glycerol and tristearoylglycerol). These retention times were plotted against equivalent carbon numbers, where ECN = actual carbon number (without glycerol) minus 2 times the number of double bonds per triacylglycerol molecule (Shukla, 1988) and the resulting standard curve was used for calculation of the ECN of the peaks of the oil samples.

Quantitation was based directly on area % since separated shoulders and main peaks represented complex mixtures of triacylglycerols having similar ECNs but with differences in carbon number, degree of saturation, and positional and geometrical configuration. Although simple separation of the triacylglycerols differing just by one double bond or two carbon atoms was desirable, partial separation of critical triacylglycerol pairs was inevitable due to optimization of the method. Therefore, area % of shoulders were included with the area % of main peak and results are reported as area % of the individual ECN. The total area of all triacylglycerols was assumed to be equal to 100%. Samples were run for approximately twenty minutes except in preliminary studies where individual samples were allowed to run for at least thirty minutes in order to check the presence of any peaks representing triacylglycerols with high ECN (above ECN of 52). 87

3.6 Statistical methods The analysis of variance (ANOV A) and the regression analysis for the RP-HPLC of the olive oil samples with guaranteed purity and their admixtures with canola oil, were performed using a computer program

(Stat View 512+ ™). Statistical tables were taken from Hogg and Tanis

(1988). 4. RESULTS AND DISCUSSION

4.1 Refractive Index, UV absorption and fatty acid composition of individual vegetable oils and model mixtures of olive oils and seed oils Traditional analytical techniques such as refractive index, absorption in UV light, and fatty acid composition were selected due to their simplicity and popularity in the quality control of olive oil. Industrial manipulations will affect the UV characteristics of olive oil, and therefore specific limits exist for the different olive oil categories. On the other hand, no differentiation of olive oil categories with respect to refractive index, fatty acid and triacylglycerol composition is possible since no effect is expected from the commonly performed alkali refining process. The same is not true for the physical refining process where an effec~ in the glyceride structure of olive oil has been observed. This is the reason why the application of this process to olive oil is doubtful (Amelotti, 1987).

4.1.1 Refractive Index (RI) All 22 virgin olive oil samples with guaranteed purity exhibited RI values within the international accepted limits for olive oil (1.4677-1.4705) (FAO/WHO, 1970) with an average value of 1.4682 (±0.0002). These data were plotted in Figure 4.1 where it can be seen that the samples showed relatively little variation and the upper set limit was well above the RI of any of the 22 samples. A representative sample was then selected to form the basis for adulteration with canola and other seed oils. Initially, 10% (w /w) of canola oil and 5% (w /w) of corn, sunflower or soybean oil were mixed with the olive oil sample. However, the change in the RI of the olive oil was insignificant

88 89

1.4704 >< 1.4701 Cl) "C 1.4698 c: 1.4695

Cl) 1.4692 > 1.4689 1.4686 1.4683 1.4680 1.4677 0 2 4 6 8 1 0 12 14 1 6 1 8 20 22

Olive Oil Samples

Figure 4.1. Refractive indices (@ 20°C) of 22 virgin olive oil samples with guaranteed purity. Lowest and highest values on the y axis represent the international olive oil limits established by FAO/WHO, 1970. 90 and therefore, larger amounts of the seed oils were incorporated in to the olive oil sample.

45, 35 and 30% (w /w) of corn, soybean and sunflower oils, respectively, resulted in RI values approaching or slightly exceeding the upper limit (1.4705) for olive oils (Table 4.1).

Adulteration with canola oil was more difficult to detect by this method since a mixture of 60% (w /w) canola in olive oil possessed a RI that remained below the upper limit (Table 4.1). These results further support stated concerns about the validity of the upper established limit (1.4705) for the RI in olive oils.

4.1.2 Absorbance of UV light Figures 4.2 and 4.3 depict the UV absorbance at two characteristic wavelengths for the 22 olive oil samples. The absorbance at 232 nm reflects the content of conjugated hydro-peroxides formed either by oxidation or industrial manipulations whereas, the absorbance at 270 nm indicates the content of conjugated trienes and polar compounds such as ketones and aldehydes also formed by the above mentioned factors.

The majority of the samples showed absorbance in accordance with established limits for olive oils (FAO/WHO, 1970; IOOC, 1985). Three of the Californian samples exhibited higher absorbance than the limits at both 232 (limit: 3.50) and 270 nm (limit: 0.25) (FAO/WHO, 1970). One Greek sample exhibited higher absorbance than the limit at 270 nm. Nevertheless, all four samples were considered virgin since their absorbance after 9 1

Table 4.1. Refractive indices of olive oil, seed oils and their admixtures.

Oil sample

Oliveb 1.4682 ± 0.0002C Canol a 1.4717 ± 0.0001 Corn 1.4736 ± 0.0001 Soybean 1.4743 ± 0.0001 Sunflower 1.4751 ± 0.0001 10%d Canola 1.4687 ± 0.0002 30% Canola 1.4696 ± 0.0001 60% Canola 1.4704 ± 0.0001 5% Corn 1.4687 ± 0.0001 45% Corn 1.4707 ± 0.0001 5% Soybean 1.4687 ± 0.0002 35% Soybean 1.4706 ± 0.0001 5% Sunflower 1.4689 ± 0.0001 30% Sunflower 1.4705 ± 0.0001 Olive limitse 1.4677- 1.4705

aRefractive index at 20°C. bsample GC2. CMean of three determinations± standard deviation. dwfw mixture with the olive oil sample. eAdapted from FAO/WHO, 1970. 92

...... 4.50 E • • s:::: • • C\1 3.50 • virgin olive oils M ..._..C\1 virgin lampante • ... • • z s:::: 2.50 ... • 0 • • • • • ...c. ... • a.. 0 1.50 - tn • .c

Figure 4.2. Absorbance of 22 virgin olive oil samples at 232 nm. Limit for virgin olive oils is 3.50, whereas there is no specific limit for virgin lampante olive oils. 93

0.45 ...-.. • virgin olive oils E . c J: • virgin lampante 0 ,..... 0.35 -% ._.,C\1 I ... 0.25 • c ... • 0 • . :IE ...c. :IE ~- .... • ~ ~- ~ 0 0.15 ... t/) ... .c . z • <(

0.05 I I I I I I I I I I " I 0 2 4 6 8 1 0 12 14 16 18 20 22 Oil Samples

Figure 4.3. Absorbance of 22 vir&in olive oil samples at 270 nm. Limit for virgin olive oils is 0.25, whereas for virgin lampante olive oils there is either no limit (IOOC, 1985), or a limit of 1.10 from the European Community Standards (Zygourakis, 1988). 94 purification with alumina was less than 0.11 at 270 nm . Alumina holds the secondary polar products of oxidation and thus the subsequent UV analysis reflects only the content of conjugated hydroperoxides (see section 2.5.2, page 31).

The amount of a seed oil that could be incorporated into a virgin olive oil sample without detection by UV absorbance analysis depended upon the extent of oxidation of these oils. A virgin olive oil with minimal oxidation (e.g., with very low absorbance at 270 nm) could be adulterated with a higher amount of "non oxidized" canola or other seed oil without the resulting mixture approaching the upper limit at 270 nm for virgin olive oils (0.25) (FAO/WHO, 1970; IOOC, 1985).

Table 4.2 includes data on UV absorbance at 270 nm of a representative virgin olive oil sample and its admixtures with several proportions of seed oils. It is apparent that adulteration with canola oil at a 10% (w /w) level would not have been detected by this method. On the other hand, the other common seed oils showed very high absorbance at 270 nm, even at a level of 10% (w /w) in the olive oil sample. It is also obvious that it would not be possible to distinguish between a poor quality olive oil and one adulterated with seed oils through direct UV analysis.

4.1.3 Fatty acid composition The fatty acid composition of the 22 virgin olive oil samples is shown in Table 4.3. Myristic acid (C14:0) was absent from all samples. Palmitic acid (C16:0) ranged from 9.0% to 15.9%. Oleic acid (C18:1), the main fatty acid of olive oil, was quite high in all samples, ranging from 70.2% to 78.5%. Linoleic 95

Table 4.2. Absorbance at 270 nm of olive oil, seed oils and their admixtures.

Oil sample Absorption at 270 nm -

Olivea 0.18 ± 0.007b 5%C Sunflower 0.26 ± 0.006 10% Sunflower 0.32 ± 0.009 5% Corn 0.21 ± 0.005 10% Corn 0.27± 0.004 5% Soybean 0.26 ± 0.003 10% Soybean 0.33 ± 0.004 5% Canola 0.22± 0.002 10% Canola 0.24 ± 0.006 12.5% Canola 0.27± 0.002 15% Canola 0.29 ± 0.008 Virgin olive limitd 0.25

asample GC2. bMean of three determinations± standard deviation. Cw /w mixture with the olive oil sample. dAdapted from FAO/WHO, 1970 and IOOC, 1985. Table 4.3. Fatty acid composition of virgin olive oil samples of guaranteed purity.

% Fattr acid methrl esters (w /w)a Olive Oil Sample C14:0b C16:0 C16:1 C18:0 C18:1 C18:2 C18:3 C20:0 C20:1 C22:0 C22:1 C24:0 C24:1

Gl n.d.c 10.8 0.5 2.9 77.9 6.1 0.8 0.5 0.3 0.2 n.d. 0.1 n.d. G2 n.d. 10.1 0.5 3.0 77.3 7.1 0.9 0.5 0.3 0.2 n.d. 0.1 n.d. G3 n.d. 9.6 0.3 2.7 74.1 11.6 0.8 0.4 0.4 0.1 n.d. trd n.d. G4 n.d. 12.9 0.9 2.1 70.7 11.9 0.8 0.4 0.3 0.1 n.d. tr n.d. G5 n.d. 12.2 0.7 2.9 75.1 7.6 0.7 0.4 0.3 0.2 n.d. tr n.d. G6 n.d. 11.7 0.7 3.2 75.3 7.6 0.7 0.4 0.3 0.1 n.d. n.d. n.d. G7 n.d. 12.2 0.8 3.1 73.8 8.3 0.8 0.5 0.3 0.1 n.d. 0.1 n.d. G8 n.d. 11.5 0.8 2.3 75.0 8.8 0.7 0.4 0.3 0.1 n.d. tr n.d. G9 n.d. 9.3 0.5 3.6 78.5 6.6 0.5 0.5 0.3 0.2 n.d. 0.1 n.d. GlO n.d. 11.3 0.6 3.1 73.9 9.4 0.6 0.5 0.3 0.2 n.d. 0.1 n.d. GCl n.d. 11.3 0.6 3.5 73.2 9.5 0.8 0.5 0.3 0.2 n.d. tr n.d. GC2 n.d. 10.7 0.6 2.9 76.9 7.0 0.8 0.5 0.3 0.1 n.d. tr n.d. 51 n.d. 12.0 1.0 4.5 75.8 5.2 0.8 0.4 0.2 0.1 n.d. n.d. n.d. 52 n.d. 14.1 1.3 2.0 70.3 10.8 0.7 0.4 0.3 0.1 n.d. 0.1 n.d. 53 n.d. 11.4 0.7 2.9 75.9 7.5 0.7 0.5 0.3 0.1 n.d. tr n.d. 11 ri.d. 12.6 0.6 2.2 74.6 8.4 0.7 0.4 0.3 0.1 n.d. n.d. n.d. 12 n.d. 13.1 0.8 2.7 72.3 9.5 0.8 0.4 0.3 0.1 n.d. tr n.d. CAl n.d. 11.2 0.6 2.6 75.0 8.5 1.2 0.3 0.3 0.1 n.d. tr tr

96 Table 4.3-contd.

% Fatty acid methyl esters (w /w) Olive Oil Sample C14:0 C16:0 C16:1 C18:0 C18:1 C18:2 C18:3 C20:0 C20:1 C22:0 C22:1 C24:0 C24:1

CA2 n.d. 9.0 0.5 3.1 77.7 8.1 0.9 0.3 0.3 0.1 n.d. tr n.d. CA3 n.d. 14.3 1.0 2.5 72.8 7.2 1.2 0.4 0.3 0.1 n.d. 0.1 n.d. CA4 n.d. 14.8 1.0 2.7 73.1 6.4 1.1 0.4 0.3 0.1 n.d. 0.1 n.d. CAS n.d. 15.9 0.9 2.5 70.2 8.3 1.3 0.4 0.3 0.1 n.d. 0.1 n.d.

Mean - 11.9 0.7 2.9 74.5 8.3 0.8 0.4 0.3 0.1 - 0.1 StDev - 1.78 0.23 0.56 2.40 1.71 0.19 0.06 0.04 0.03 - 0.02 Olive oil max.e 0.1 20.0 3.5 5.0 83.0 21.0 1.5 0.8 m.a.f 0.2 n.d. 1.0 n.d.

aMean of three determinations. hNumber of carbon atoms : number of double bonds. eN ot detected. dTrace. eAdapted from IOOC, 1985 and FAO/WHO, 1970. fMinute amounts. For sample abbreviations see Table 3.1.

97 98 acid (CJ8:2) was present in moderate amounts with a minimum of 5.2% for the Spanish 51 sample and a maximum of 11.9% for the Greek G4 sample. The maximum value of 21% for linoleic acid, which is considered normal for olive oils according to the IOOC recommendations (1985), was not approached by any sample. Such an extreme value has been most probably established due to the unusually high content of linoleic acid found in some olive oils of North African origin (Rana and Ahmed, 1981). Climatic conditions and/ or delayed harvesting affect the fatty acid composition of the olive fruit with mainly an increase in linoleic acid at the expense of palmitic acid (Kiritsakis and Markakis, 1987). The linolenic acid (C18:3) content of all samples was within the limits (0.0-1.5%) (IOOC, 1985). The lowest value determined was 0.5% for the Greek sample G9, whereas the maximum value of 1.3% was found in the Californian sample CAS. The Californian samples had the highest ~ontent of this particular fatty acid, ranging from 0.9 to 1.3%. All other samples had less than 1.0% linolenic acid content.

With respect to the content of higher molecular weight fatty acids, all samples were characterized as "normal" according to international proposals (FAO/WHO, 1970; IOOC, 1985). However, gadoleic acid (C20:1) was present in all samples in a rather constant amount (-0.3%) but no specific limit for this particular fatty acid has been proposed. The only official reference for the content of gadoleic acid in olive oils is found in the recommendations of the FAO/WHO (1970) where it is stated that gadoleic acid might be present in "minute amounts". Values as high as 0.5% appear in several publications (Fedeli, 1977; 1983). 99

Since adulteration of olive oil with common seed oils is expected to increase the linoleic acid content an olive oil sample with a low linoleic acid content (sample GC2) was selected for preparing mixtures with seed oils. This was done to illustrate a difficult case where the olive oil sample can withstand the introduction of a considerable amount of a seed oil before its linoleic acid content falls outside of acceptable limits.

Sample GC2 was mixed with several levels of canola, corn, sunflower and soybean oils and the fatty acid composition was determined (Table 4.4). The canola oil sample had a high oleic acid content and a relatively low linolenic acid content. Its linoleic acid content was very close to the normal limits of olive oil. The presence of canola oil in the virgin olive oil sample could be detected only when incorporated at a level of 15% (w /w) level. At a level of 15% (w /w), the linolenic acid slightly passed the upper limit (1.5%) for olive oil. It was not feasible to detect adulteration with canola on the basis of linoleic acid content.

Sunflower and corn oil additions were even more difficult to detect as incorporation of levels as high as 20% and 25% (w /w), respectively, yielded values for linoleic acid content (20.2 and 20.0 respectively) that still fell below the international limit for olive oil (21.0 % w /w) (IOOC, 1985). Soybean oil could be detected more readily since a 10% (w /w) level of adulteration yielded a linolenic acid content of 1.5%, the upper limit for olive oil (IOOC, 1985). Soybean oil could be detected on the basis of linoleic acid content only if incorporated at a ~30% level in the olive oil sample.

No other single fatty acid was useful in the detection of such admixture Table 4.4. Fatty acid composition of olive oil, seed oils and their admixtures.

% Fatty acid meth~l esters (w /w)a

Oil Sample C14:0b C16:0 C16:1 C18:0 C18:1 C18:2 C18:3 C20:0 C20:1 C22:0 C22:1 C24:0 C24:1

Olivec n.d.d 10.7 0.6 2.9 76.9 7.0 0.8 0.5 0.3 0.1 n.d. tre n.d.· Canola 0.1 4.5 0.2 2.2 61.6 22.1 6.2 0.6 1.5 0.3 0.3 0.2 0.2 2.S%f Canola n.d. 10.5 0.6 2.9 76.5 7.4 0.9 0.5 0.3 0.1 n.d. tr n.d. 5% Canola n.d. 10.4 0.6 2.9 76.1 7.8 1.1 0.5 0.4 0.1 n.d. tr n.d. 7.5% Canola n.d. 10.2 0.6 2.8 75.8 8.1 1.2 0.5 0.4 0.1 n.d. tr n.d. 10% Canola n.d. 10.1 0.6 2.8 75.4 8.5 1.3 0.5 0.4 0.1 n.d. tr n.d. 12.5% Canola n.d. 9.9 0.6 2.8 75.0 8.9 1.5 0.5 0.5 0.1 n.d. tr n.d. 15% Canola n.d. 9.8 0.5 2.8 74.6 9.3 1.6 0.5 0.5 0.1 n.d. tr n.d. 20% Canola n.d. 9.5 0.5 2.8 73.8 10.0 1.9 0.5 0.5 0.1 0.1 tr n.d. 50% Canola 0.1 7.6 0.4 2.6 69.3 14.6 3.5 0.6 0.9 0.2 0.2 0.1 0.1 Corn n.d. 9.7 0.1 1.7 27.8 59.0 0.8 0.4 0.3 0.1 n.d. 0.1 n.d. 2.5% Corn n.d. 10.7 0.6 2.9 75.7 8.3 0.8 0.5 0.3 0.1 n.d. tr n.d. 25% Corn n.d. 10.5 0.5 2.6 64.6 20.0 0.8 0.5 0.3 0.1 n.d. tr n.d. 30% Corn n.d. 10.4 0.5 2.5 62.2 22.6 0.8 0.5 0.3 0.1 n.d. tr n.d. Sunflower 0.1 5.9 n.d. 5.5 14.2 72.8 0.2 0.3 0.1 0.7 n.d. 0.1 n.d. 2.5% Sunflower n.d. 10.6 0.6 3.0 75.3 8.6 0.8 0.5 0.3 0.1 n.d. tr n.d. 20% Sunflower n.d. 9.7 0.5 3.4 64.4 20.2 0.7 0.5 0.3 0.2 n.d. tr n.d. 25% Sunflower n.d. 9.5 0.5 3.6 61.2 23.5 0.7 0.5 0.3 0.3 n.d. tr n.d. Soybean 0.1 10.7 0.1 3.7 19.9 56.7 8.0 0.3 0.2 0.3 n.d. 0.1 n.d.

100 Table 4.4-contd.

% Fatt~ acid methyl esters (w/w)

Oil Sample C14:0 C16:0 C16:1 C18:0 C18:1 C18:2 C18:3 C20:0 C20:1 C22:0 C22:1 C24:0 C24:1

2.5% Soybean n.d. 10.7 0.6 2.9 75.5 8.2 1.0 0.5 0.3 0.1 n.d. tr n.d. 7.5% Soybean n.d. 10.7 0.6 3.0 72.6 10.7 1.3 0.5 0.3 0.1 n.d. tr n.d. 10% Soybean n.d. 10.7 0.6 3.0 71.2 12.0 1.5 0.5 0.3 0.1 n.d. tr n.d. 12.5% Soybean n.d. 10.7 0.5 3.0 69.8 13.2 1.7 0.5 0.3 0.1 n.d. tr n.d. 30% Soybean n.d. 10.7 0.5 3.1 59.8 21.9 3.0 0.4 0.3 0.2 n.d. tr n.d. Olive oil maxg 0.1 20.0 3.5 5.0 83.0 21.0 1.5 0.8 m.ah 0.2 n.d. 1.0 n.d. aMean of three determinations. bNumber of carbon atoms : number of double bonds. csample GC2 dNot detected. eTrace. fw/w admixture with the virgin olive oil sample. gAdapted from IOOC, 1985 and FAO/WHO, 1970. hMinute amounts.

10 1 102 with the exceptions of gadoleic and erucic (C22:1) acids which were present at 1.5% and 0.3% respectively in canola oil. However, these fatty acids were present in useful amounts only when canola was incorporated at such a level that its detection was already possible from the increased content in linolenic acid. Moreover, the use of the linolenic acid limit is preferable since a definite upper value has been proposed, in contrast to the specifications for gadoleic and erucic acids, which are more uncertain.

Meaningful values for linoleic acid (25% corn and 20% sunflower oils in olive oil) and linolenic acid (12.5% canola and 10% soybean oils in olive oil) (Table 4.4) would conclusively reveal adulteration if an olive oil with precisely known origin was analyzed since limits in fatty acid composition for certain geogr,aphical areas are narrower. This is the case in private industry or cooperative organizations which receive bulk amounts of olive oil of known fatty acid variation from specific areas. However, in Canada such extreme values must be considered normal since only the country of production is usually known.

4.2 Triacylglycerol analysis of individual vegetable oils and model mixtures of olive oils with seed oils by RP-HPLC

4.2.1 Preliminary studies Mobile phase selection. A number of different mobile phases were studied in order to optimize analysis time, separation and other factors such as complexity and cost of the solvent system. A mixture of acetone and acetonitrile (93:7 v /v) was finally used for the main study. Such a mobile 103 phase gave fast elution of triacylglycerols with good separation, and due to the high content of acetone in the mixture, was the least expensive (Table 4.5).

Simple separation of the triacylglycerols according to equivalent carbon number (ECN) (without separation of critical pairs) was preferable (Kapoulas and Andrikopoulos, 1986) but such an attempt (by decreasing polarity of the mobile phase) resulted in fairly crowded chromatograms with a risk of overlap of triacylglycerols with non-triacylglycerol constituents (Figure 4.4).

Injection solvent. Two injection solvents were also studied. Acetone gave in all cases sharper peaks with the only disadvantage being an inability to solubilize trisaturated triacylglycerols. However, such triacylglycerols were absent from the oils utilized in this study. In identifying the eluted triacylglycerol peaks according to their retention times, tristearoylglycerol was used with chloroform as the solubilizing agent.

The nature of the injection solvent was important in subsequent studies on the adulteration of olive oil with seed oils (especially canola oil) since by using chloroform the peak representing triacylglycerols with an ECN of 42 was not detected. By using acetone the same peak accounted for 1% of the total area of the triacylglycerols in a sample of olive oil adulterated with 10% (w /w) canola oil. This resulted in the detection of lower amounts of seed oils in olive oils since the area percentage of the peak for ECN of 42 was one of the useful factors in revealing such adulteration, as will be illustrated later. The comparison of the two injection solvents is summarized in Table 4.6.

Identification of the triacylglycerol peaks. The identification of the Table 4.5. Separation of triacylglycerols by RP-HPLC using various mobile phases.

Retention time for ECN 50 (tR in min) Mobile Phase a a Comments

ACNb /THFC- 57/43 16.5 1.06 expensive

Acetone 12 _d cheap Acetone/THF- 95/5 12 Acetone/ ACN- 95/5 14 1.05 Acetone/ ACN- 90/10 19 1.06 Acetone/ ACN- 93/7 <15 1.06 Acetone/ ACN /Hexane-90/7 /3 15 1.05 complex

Acetone/ ACN /Hexane-90/8/2 16 1.06 II

Acetone/ ACN /Hexane-88/10/2 18 1.06 II

aseparation factor (tRB-tM/tRA-tM) for the two major peaks for ECN of 48. b Acetonitrile. CTetr ah ydrofuran. d No separation.

104 105

triacylglycerols ~

0 4 8 12 16 Time (min)

Figure 4.4. Reversed-phase high performance liquid chromatogram of a virgin olive oil (sample GC2) by using acetone as the mobile phase. Table 4.6. Separation of triacylglycerols by RP-HPLC using various injection solvents.

Area% of peak for

Injection Solvent Advantages Disadvantages ECN 42 in sample xa

Chloroform dissolves peak broadening no integration saturated TAGsb

Acetone sharp peaks saturated TAGs 1.0 not soluble asample is a mixture of a virgin olive oil with 10% (w lw) canola oil. Mobile phase: ACN I THF-57 I 43 (v lv). hTriacylgl ycerols.

106 107 peaks was based on retention times. Triacylglycerol standards (trilinolenylglycerol, trilinoleylglycerol, trioleylglycerol and tristea­ roylglycerol) were analyzed and their retention times were plotted against ECN values (Figure 4.5). The resulting line was used to identify triacylglycerol peaks of natural oils by comparison of their retention times. A drastic increase in the retention time of triacylglycerols with high ECN was observed, presumably due to the lower solubility of such triacylglycerols in the mobile phase.

Data on calculated and experimental retention times of triacylglycerols are given in Table 4.7. There have been reports in the literature stating that elution of tristearoylglycerol with mixtures of acetone and acetonitrile as the mobile phase was a problem (Lie Ken Jie, 1980). Such a difficulty was not found during the injection of tristearoylglycerol in our system. This particular triacylglycerol having an ECN of 54 was eluted in approximately 23 minutes.

Free fatty acid effect. Injections of oleic acid solutions in the HPLC system showed that no interference from the free fatty acids of a sample occurred during RP-HPLC triacylglycerol analysis (Figure 4.6). Oleic acid eluted in about 6 min with the conditions used in the main study. Under the same conditions, even trilinolenylglycerol, with an ECN of 36, and which is absent from the oils utilized in this study, eluted in 7.25 minutes.

Linearity. Finally the linearity of the chromatographic system was checked by injecting known amounts of trioleylglycerol. The amount of trioleylglycerol in mg was plotted against area counts of the integrator (Figure 108

24.00 ....IJ 22.00 ./ ..-...... 20.00 ..... = ...... ·-.._,E ..... 18.00 ..... c:J ...... ·· E ... 16.00 ...... E-- .... ·- ..... 14.00 ...... d" =Q ...... - ...... ·- 12.00 ...... -=c:J ...... c:J 10.00 ...... a "' 8.00 ...... 6.00 -1--r---r---.----r----.,.------.----r.-,r-..--~-r---r---r---,..--,r---w 36 38 40 42 44 46 48 50 52 54 56

Equivalent Carbon Number (ECN)

Figure 4.5. Plot of retention time vs equivalent carbon number of trilinolenylglycerol, trilinoleylglycerol, trioleylglycerol and tristearoylglycerol (in eluting order) analyzed by RP-HPLC. Table 4.7. Retention time (tR), by RP-HPLC, of triacylglycerols eluted according to equivalent carbon number.

Sunflower ECNof tR Olive oil Canola oil oil triacylglycerol of standard Calculated tR found found found

36 (LnLnLn) 7.23 40 8.60 8.69 8.67 42 (LLL) 9.34 9.56 9.31 44 10.70 10.88 10.58 46 12.10 12.11 11.88 48 (QCX)) 13.49 13.63 13.56 50 16.80 16.66 16.34 54 (StStSt) 23.00

Abbreviations: Ln: linolenic acid; L: linoleic acid; 0: oleic acid; St: stearic acid.

109 110

~ ~ .....-4 Q) 0 :> M .....-4 Q) 0 u {f) ~ .....-4 b1) .....-4 ~ Q) .....-4 '"d 0 •...-! •...-! u M (lj ~ ...... u Q) .....-4 0

0 4 8 12 16 Time (min)

Figure 4.6. Reversed-phase high performance liquid chromatogram of oleic acid and trioleylgl ycerol. 1 1 1

4.7). A good correlation was found between these two parameters, as shown by an R2 value of 0.984.

4.2.2 Triacylglycerol profiles and composition of olive oils, seed oils and their admixtures After standardization of the conditions in the RP-HPLC system, individual samples of virgin olive, canola, corn, sunflower and soybean oil were run for approximately 30 min in order to obtain the RP-HPLC profiles (Figure 4.8). No reproducible peaks subsequent to the peak representing an ECN of 52 were found for any of these oils, and therefore all subsequent runs had a stop time of 18 min.

Twenty two virgin olive oil samples with guaranteed purity were analyzed and despite their differences in fa tty acid com position had very similar triacylglycerol profiles (Figure 4.9). The quantitative data for these oils are given in Table 4.8.

Triacylglycerols with an ECN of 40 were absent from all olive oils. Triacylglycerols with an ECN of 42 were present only in trace amounts (<0.5%) with the exception of two Californian samples (CAl and CAS) which had higher amounts (0.8 and 0.9% respectively) of these particular triacylglycerols. According to Descargues and Bezard (1981) such triacylglyce­ rols in olive oil are of the type OLLn, PLLn and LLL. Since all seed oils were characterized by a large peak representing triacylglycerols with an ECN of 42 (Figure 4.8) a first decision rule was established for detecting adulteration of

olive oil i.e., olive oils should have ~1%. of triacylglycerols representing ECN of42. 112

1.40e+8

1.20e+8 R"2 = 0.984 ....1:/l = 1.00e+8 =0 u 8.00e+7

~ ~ 6.00e+7 '- < 4.00e+7

2.00e+7 0.40 0.60 0.80 1.00 1.20 1.40 1.60 000 mg

Figure 4.7. Plot of integrator's area counts vs milligrams of trioleylglycerol (000) after RP-HPLC analysis. -IS

-16 1 1 3 olive oil Jjl!i

-12 -1-t

c 1 -16 "> g

corn oil ll, -18

40 ~i 50 52

42 44 c "> ~ 46 sunflower oil

48

42 44

40 soybean oil

46 48

canola oil 42

16 Time (min)

Figure 4.8. Reverse-phase high performance liquid chromatograms of virgin olive and seed oils. Numbers represent ECN values, where ECN = carbon number - 2 x number of double bonds. 114 48 48 A B ...... c 46 c 46 Q) Q) .2! > 0 0 U) U)

) 4 8 12 16 0 4 8 12 16 48 48 c D ...... c 46 c Q) Q) > > 46 0 0 W. (/)

44

0 4 8 12 16 0 4 8 12 16

Time (min) Time (min)

Figure 4.9. Representative reversed-phase high performance liquid chromatograms of virgin olive oils. (A): Greek; (B): Spanish; (C): Italian; (D): Californian. 115

Table 4.8. Triacylglycerol composition of olive oil samples with guaranteed purity, as determined by direct RP-HPLC analysisa.

Area % of peak with ECNb

Sample 42 44 46 48 50 52

Gl trC 3.5 13.6 76.5 6.5 tr G2 tr 3.9 16.9 73.2 6.0 tr G3 tr 5.8 24.8 65.0 4.3 tr G4 tr 5.8 26.1 65.2 3.0 tr GS tr 3.7 18.7 72.5 5.2 tr G6 tr 3.7 18.3 72.2 5.7 tr G7 tr 4.8 19.7 70.5 5.0 tr G8 tr 5.0 22.4 69.5 3.2 tr G9 tr 3.3 15.4 75.0 6.3 tr GlO tr 4.8 21.1 68.9 5.3 tr GCl tr 4.5 20.6 66.9 7.1 0.8 GC2 tr 3.8 17.5 72.1 5.8 0.9 51 tr 2.9 14.1 74.0 9.0 tr 52 tr 5.3 25.9 66.0 2.8 tr 53 tr 3.8 17.9 72.2 5.5 0.7 Il tr 4.0 19.5 72.5 4.1 tr 12 tr 4.8 22.6 68.2 4.4 tr CAl 0.8 4.8 19.8 70.2 4.2 tr CA2 tr 4.0 18.8 71.6 5.6 tr CA3 tr 3.6 16.9 73.8 5.7 tr CA4 tr 3.3 15.3 75.6 5.9 tr CAS 0.9 4.0 19.3 70.9 4.9 tr

Mean <0.5 4.2 19.3 71.0 5.2 <0.5 Min tr 2.9 13.6 65.0 2.8 tr Max 0.9 5.8 26.1 76.5 9.0 0.9 StDev 0.8 3.5 3.3 1.4

aMean of three determinations. hECN = carbon number - 2 x number of double bonds. c

The area % of the peak containing triacylglycerols with an ECN of 44 such as OLL and PLL (Descargues and Bezard, 1981) is quite low in olive oils, with a maximum of 5.8%. Peaks containing triacylglycerols with ECN of 46 and 48 were the main ones in all olive oil samples. They were found to account for as high as 26.1 and 76.5% respectively. These two peaks contain mainly triacylglycerols of the type LOO, PLO (both with an ECN of 46) 000 and POO (both with an ECN of 48) (Dong and DiCesare, 1983). The peak containing triacylglycerols with an ECN of 50 was the third main peak in olive oils. This peak is expected to contain triacylglycerols of the type StOO and StPO (Dong and DiCesare, 1983). Finally, the peak representing triacylglycerols with an ECN of 52 such as A1oo and PAO (Descargues and Bezard, 1981) was present in trace amounts.

·All the seed oils had considerable amounts of triacylglycerols with an ECN of 42. Canola oil had only 8.1% of such triacylglycerols, whereas very high amounts were found in sunflower, soybean and corn oils (Table 4.9).

In contrast to olive oils, the main peaks in corn, sunflower and soybean oils were the ones representing triacylglycerols with ECN 42, 44 and 46. This is not the case, however, with canola oil. This seed oil shows a triacylglycerol composition more similar to olive oil since the main peak in canola is the one containing triacylglycerols with ECN of 48, just like olive oil. The second main peak is the one for ECN of 46, again similar to olive oil. The peak for ECN of 44 was quite high in canola oil but not as high as in the other common seed oils analyzed (Table 4.9).

1A: arachidic acid. Table 4.9. Triacylglycerol composition of seed oil samples, as determined by direct RP-HPLC analysis a.

Area % of peak with ECNb Seed Oil Sample 38 40 42 44 46 48 50 52

Canola _c 2.1 8.1 19.7 28.2 37.3 3.8 0.9

Corn - 1.0 23.5 39.3 26.2 10.0 trd tr

Sunflower - 0.5 39.0 35.1 19.5 4.8 1.0

Soybean 1.6 9.9 27.7 31.2 18.6 9.3 1.7 aMean of three determinations. bEeN = carbon number - 2 x number of double bonds. CN ot detected. d<0.5%.

11 7 118

The peak representing an ECN of 52 was either absent (in sunflower and soybean) or present (in canola and corn) in approximately the same amount as in olive oil, and thus it was of no use in revealing admixtures of these seed oils with olive oils.

Finally, triacylglycerols with ECN of 40 were present in all seed oils but in considerable amounts only in soybean oil (Table 4.9). Moreover, soybean oil was the only oil that contained triacylglycerols with ECN of 38 (-1.5%). This peak includes basically triacylglycerols of the type LLnLn (Geeraert and De Schepper, 1983).

The area % of the peaks for ECN of 48 and 46 alone were not useful in revealing adulteration of olive oil. Therefore, all possible ratios between the area % of the eluting peaks were calculated in order to find a ratio that would be helpful in detecting adulteration of olive oil, especially with canola oil, because of the low content of the latter of triacylglycerols with ECN of 42.

With the olive oil samples a good correlation was found between the area % of the peaks for ECN 44 and 46 (R=0.941), 44 and 48 (R=-0.921), and 46 and 48 (R=-0.951). That is, samples with a high content of triacylglycerols with ECN 44 also had higher amounts of triacylglycerols with ECN 46 whereas the same samples had a lower content in triacylglycerols with ECN 48. The lower the variability in the content of a particular group of triacylglycerols (with the same ECN) the more useful this "peak" is for incorporation into a ratio indicating the purity of an olive oil sample. The area % of the peak for

ECN 48 was the least variable (Table 4.8) but its ratio with the area % of the peak for ECN 44 was more variable (significant at 95%) than the ratio of the 11 9 area % of the peak for ECN 46 to that for ECN 44 (coefficients of variation 22.5 and 6.5 respectively).

The latter ratio was thus considered to be a fingerprint for olive oils. It was found to range between the values 3.9 and 5.1 (Figure 4.10) and therefore, any olive oil sample with a value for this ratio of lower than 3.9 could be considered as adulterated, since incorporation of seed oils in olive oil would result in a reduction of this ratio due to the higher content of triacylglycerols with an ECN of 44 in all seed oils.

The genuine olive oil samples were then mixed with 2.5%-10% (w/w) corn and sunflower oils, and with 2.5% to 30% (w /w) soybean and canola oils (in 2.5% increments). The area percentage of the peak corresponding to the triacylglycerols with ECN 40 and 42 and the rati_o of the area percentage of the peak for ECN 46 to that for ECN 44 were obtained by direct RP-HPLC analysis. Results for corn, sunflower, and soybean oils mixtures with a genuine olive oil sample are presented in Table 4.10. As shown in this table, both factors would indicate the adulteration of olive oil with these seed oils when the latter were present even at 2.5% (w /w) in the genuine olive oil sample (% peak 42 >1% and ratio 46/44 peaks < 3.9). At a 5% level of adulteration, the peak representing an ECN of 40 starts to appear in the olive oil sample adulterated with soybean oil reflecting the presence of that peak in soybean oil in appreciable amounts (-10%).

On the other hand, using the same factors, adulteration of the genuine olive oil samples with canola oil at less than 7.5% (w /w) proportion could not be detected(% peak 42 <1% and ratio 46/44 peaks >3.9), (Table 4.11). The peak 120

5.61..------r 5.

0

0 0 ·~.~·~.·.~w.·~.·~.·.·~.·.~~·.~.v~.•. , •.•.•.•.•.-.v•.···•·•.v.·.~~·~.·.•.v.·~.~·.·~.·.•.-.•.•.w.v-...·~.·.~·~.~·~.-~--~··"··w~.-.·•·~·Y"-''•.-.·.·~.w.w~.-.·~.v.y.-.w.w.w.·.v~.·.•.•.•.•.v.v.•.v.v.v.•.·~············~.•.•.v.·~ ...... •.•.•.•.•.•.•..... - 2 (J 3. 0 3.~------Observations

Figure 4.10. Scattergram of the ratio of the area % of peak for ECN of 46 to area% of peak for ECN of 44 for 22 olive oil samples with guaranteed purity. 121

Table 4.10. Area% of the peaks for ECN 40 and 42 and ratio of 46/44 peaks for olive oil with guaranteed purity· and its admixtures with different proportions of common seed oils as determined by direct RP-HPLC analysis of the triacylglycerols.

%Area of

Oil Sample Peak 40 Peak 42 Ratio of peaks 46 I 44

Olive a _b <0.5 4.6 (0.08)C Corn 1.0 (0.15) 23.5 (0.19) 0.7 (0.00) Sunflower 0.5 (0.04) 39.0 (0.70) 0.6 (0.00) Soybean 9.9 (0.13) 27.7 (0.03) 0.6 (0.00) Corn 2.5%d 1.1 (0.06) 3.8 (0.08) Corn 5% 1.5 (0.13) 3.3 (0.08) Sunflower 2.5% 1.4 (0.18) 3.8 (0.05) Sunflower 5% 2.4 (0.07) 3.2 (0.07) Soybean 2.5% 1.5 (0.17) 3.7 (0.27) Soybean 5% 0.4 (0.02) 1.9 (0.20) 3.4 (0.19) Soybean 7.5% 0.7 (0.00) 2.3 (0.03) 3.3 (0.16) Soybean 30% 3.0 (0.12) 9.0 (0.09) 1.4 (0.06) a A representative virgin olive oil (sample GC2 from Table 4.8). b Not detected. c Mean of three determinations (standard deviation). d w /w admixture with the olive oil sample. 122

Table 4.11. Area % of the triacylglycerols peaks for ECN 40 and 42 and the ratio of 46/44 peaks as determined by direct RP-HPLC analysis for olive oil with guaranteed purity and its admixtures with different proportions of canola oil.

%Area of

Oil Sample Peak 40 Peak 42 Ratio of peaks 46/44

Olive a _b <0.5 4.6 (0.38)C

Canola 2.1 (0.03) 8.1 (0.05) 1.4 (0.01)

2.5%d canola <0.5 4.1 (0.38)

5% canola 0.7 (0.46) 3.9 (0.35)

7.5% canola 1.1 (0.28) 3.7 (0.26)

30% canola 0.7 (0.03) 2.7 (0.19) 2.5 (0.07)

aAverage of 22 olive oil samples (from Table 4.8 and Figure 4.10). bNot detected. CMean of three determinations (standard deviation). dw /w admixture with the olive oil samples. 123 for ECN 40 was not useful as a tool for detecting the adulteration of the olive oil samples with canola oil because it begins to appear in the chromatograms only when olive oil was mixed with 30% (w/w) or more of canola oil (Table 4.11).

However, the use of the peak ratio (46/ 44) proved to be consistently more useful than the area % of the peak for ECN 42 in detecting the presence of canola oil in the olive oil samples. This is shown in Table 4.12 which contains the percentage of the samples detected as adulterated by using both factors. At the low levels of adulteration (2.5%, 5%, and 7.5% w /w canola in olive oil) the calculation of the ratio 46/44 could reveal a larger number of the adulterated samples compared to the use of the area% of the peak for ECN 42.

Qualitatively, the peak corresponding to an ECN of 42 appears to have various characteristics depending on the foreign oil that has been used as the adulterant. Sunflower and corn oil have a very sharp peak with an ECN of 42 due to the high content of LLL in these oils (Aitzetmuller, 1982; El-Hamdy and Perkins, 1981b), (Figure 4.8). Canola oil also shows a sharp peak for ECN of 42 even though the main triacylglycerol in this peak is not LLL but OLLn (Prevot et al., 1990) (Figure 4.8). Soybean oil has a peak with an ECN of 42 that has a distinct shoulder (due to the critical isomers OLLn and PLLn) (Phillips et al., 1984b) which is reflected in the virgin olive oil samples adulterated with soybean oil (Figure 4.11).

The RP-HPLC technique could also be useful in potentially identifying whether soybean or canola oil had been used as an adulterant. The peak with 124

Table 4.12. Percentage of olive oil samples (n=22) mixed with canola oil that were detected as adulterated using the area% of peak for ECN 42 or the ratio of area% of peak for ECN 46 to peak for ECN 44 after direct RP-HPLC analysis of the triacylglycerols.

% (w/w) of %of samples detected as adulterated using canola oil area % ratio of in olive oils Area% of peak 42 46/44peaks

2.5 5 18

5 18 32

7.5 64 77

10 100 100 48 125

46

A

0 4 8 12 16 2 44 s:: -Q,) > 40 -0

0 4 8 12 16 .... 48 .:: "> -0 46 c Ill 42 44

0 s 12 16 Time, (min)

Figure 4.11. Reversed-phase high performance liquid chromatograms of virgin olive oil (A), soybean oil (B) and sample A adulterated with 30% (w /w)

sample B. 126 an ECN of 40 is common for soybean and canola oils and has been found to consist of triacylglycerols containing linolenic acid (mainly LLLn) (Takahashi et al., 1984). Although the canola and soybean oils used in this study had a similar linolenic acid content, 6.8 and 8.4 respectively (Table 4.4), the peak with an ECN of 40 accounts for almost 10% of the total area of the triacylglycerols present in soybean oil and only 2.0% in canola (Table 4.9). This property reflects a characteristic triacylglycerol composition specific for each kind of natural vegetable oil.

Mixtures of virgin olive oil with 30% (w /w) of either canola or soybean oil give distinct chromatograms (Figure 4.12). In the sample adulterated with 30% (w /w) soybean oil, the peak with an ECN of 42 had an area of almost 10% and the peak with an ECN of 40 had an area of 3% (Table 4.10). Furthermore, the peak height of the peak with an ECN of 42 was greater than that of the peak with an ECN of 44 and both exhibited distinct shoulders (due to the separation of the critical pairs PLLn and OLLn for peak 42 and PLL for the peak 44) (Phillips et al., 1984b). On the other hand, the mixture of 30% (w /w) canola in virgin olive oil gives a peak with an ECN of 42 equal to 2.7 area % and a peak with an ECN of 40 of 0.7 area% (Table 4.11). The peak for an ECN of 44 does not appear to have a separated shoulder in this sample and the peak height of the peak for an ECN of 42 is only about half that of the peak for an ECN of 44. A comparison of the chromatograms yielded by a level of 7.5% soybean oil or 30% (w /w) canola oil in virgin olive oil show similar size peaks with ECNs of 40 and 42. However, the sample that contains canola oil can be differentiated by a larger peak with an ECN of 44, and a lack of the distinct shoulders that appear in the peaks for ECNs of 42 and 44 in the virgin olive oil sample adulterated with soybean oil. 127

48

-4-1 48 s:: ~ QJ s:: QJ > 46 > -0 ~ 0 46 A B (I)

50

0 4 8 12 16 0 4 8 12 16 .., 48 46 48 s:: ..... ~ c:: ~ > > 0 - 46 Q Vl c D Vl 42 44

44

0 4 8 12 16 0 4 8 12 16 Time, (min) Time, (min)

Figure 4.12. Reversed-phase high_ performance liquid chromatograms of virgin olive oil (A) and sample A adulterated with 7.5% (w /w) soybean oil (B), 30% (w /w) soybean oil (C) and 30% (w /w) canola oil (D). 128

4.3 Studies on commercially available olive oil products A comparison between the methods used in 4.1 and 4.2 was also carried out by application to commercial olive oil products. The purpose of this approach was two fold: First, to investigate the potential of RP-HPLC as a method to detect adulteration of "real" olive oil samples and, second, to give an indication of the purity of such products available to Canadian consumers.

4.3.1 Refractive Index Data on the RI of 14 olive oil products are given in Figure 4.13. All samples analyzed fell within the international limits (1.4677-1.4705) (FAO/WHO, 1970). The RI ranged from 1.4:679 to 1.4692.

4.3.2 Absorbance of UV light All samples showed good quality characteristics as far as their absorbance at 232 nm is concerned. Absorbance at this particular wavelength ranged between 0.87 and 1.33, well below even from the 3.50 limit · (FAO/WHO, 1970) for virgin olive oils (Figure 4.14). For products labeled as olive oil, no specific limit exists.

Three out of 14 samples were labeled as "virgin olive oils". Their absorbance at 270 nm was rather high (0.22-0.26), (Figure 4.15). One of the samples slightly passed the limit of 0.25 but its absorbance after purification through alumina was lower than 0.11 and therefore, was still regarded as properly labelled (IOOC, 1985).

The 11 samples labelled as "olivt~ oils" had absorbance below the limiting value of 0.90 (FAO/WHO, 1970; l()OC, 1985) at 270 nm (Figure 4.15). 129

1.4704 >< 1.4701 Cl) -c 1.4698 c 1.4695

Cl) 1.4692 ...> 1.4689 0 ! as 1.4686 • • lo...... 1.4683 Cl) t ! ! ! ! • a: 1.4680 ! 1.4677 0 2 4 6 8 1 0 12 14 Olive Oil Samples

Figure 4.13. Refractive indices (@ 20°C) of commercially available olive oil products. Lowest and highest values on they axis represent the international olive oil limits established by FAO/WHO, 1970. 130

1.40 ...... • olive oils E 0 virgin olive oils c 1.30 • N ("') 1.20 ....._.N • I z :E

c 1.10 • :E 0 ... • ...a. 1.00 J... 2 0 0 tn 0.90 I .c < 0.80 0 2 4 6 8 1 0 12 1 4 Oil Samples

Figure 4.14. Absorbance at 232nm of commercially available olive oil products. Limit for virgin olive oils is 3.50 (FAO/WHO, 1970), whereas there is no specific limit for products labeled as olive oils. 13 1

...-.. 0.85 E - c: . • • olive oils 0 • • • ...... 0.65 ... 0 virgin olive oils N • ~ . •

c: 0.45 0 .... . • • Q. 0.25 t\ 0 0 u -U) • .c < 0.05 I I 1 I . I I . I 0 2 4 6 8 1 0 12 14 Oil Samples

Figure 4.15. Absorbance at 270 nm of commercially available olive oil products. Limit for virgin olive oils is 0.25, whereas for products labeled as olive oils it is 0.90 (FAO/WHO, 1970; IOOC, 1985). 132

One of them had low enough absorbance to be characterized as virgin olive oil (0.21).

4.3.3 Fatty acid composition The fatty acid composition of the 14 commercial olive oil products is given in Table 4.13. No sample was found to deviate from naturally acceptable values for olive oils. Both linoleic (C18:2) and linolenic (C18:3) acid content, which is particularly useful in detecting seed oils, did not raise any suspicions about the purity of these products. Linoleic acid ranged from as low as 5.5% to 15.5% a value well below the maximum allowed for olive oils (21.0%) (IOOC, 1985). Linolenic acid ranged from 0.4 to 1.2 and never approached the upper set limit of 1.5% (IOOC, 1985) (Table 4.13).

No other particular fatty acid content was considered abnormal. Gadoleic acid (C20:1) was present in all samples in amounts between 0.2 to 0.3%. Olive oil limits call for only minute amounts for this fatty acid (FAO/WHO, 1970). However, such values were very common in all samples of olive oil with guaranteed purity as described in section 4.1.3.

4.3.4 Triacylglycerol profiles and composition of commercial olive oil products by RP-HPLC The RP-HPLC profiles of the commercial olive oils are shown in Figure 4.16 as compared to the profile of an authentic olive oil sample (sample GC2).

Quantitative data on the triacylglycerol composition and the ratio of the area% of peaks for ECN of 46 to 44 for the commercial olive oils are shown in Table 4.14. This data provides insight as to the commercial oil Table 4.13. Fatty acid composition of commercially available olive oil products.

Fatty acid composition (wt.%)a

Olive Oil Type C14:0b C16:0 C16:1 c18:0 C18:1 C18:2 C18:3 C20:0 C20:1 C22:0 C22:1 C24:0 C24:1

1489/100% Pure n.d.c 11.6 1.0 3.2 71.9 10.8 0.7 0.4 0.3 0.1 n.d. 0.1 n.d. 1490/100% Pure n.d. 12.1 0.8 3.3 66.6 15.5 1.2 0.4 0.3 0.2 n.d. 0.1 n.d. 1491/100% Pure n.d. 13.0 1.3 2.3 70.5 11.5 0.7 0.4 0.2 0.1 n.d. 0.1 n.d. 1492 n.d. 10.4 0.7 3.6 76.2 7.0 0.5 0.4 0.3 0.1 n.d. n.d. n.d. 1493/Pure n.d. 13.7 1.9 2.1 75.3 5.5 0.8 0.3 0.3 0.1 n.d. n.d. n.d. 1494/Pure n.d. 10.6 0.7 3.6 75.2 8.4 0.5 0.4 0.3 0.2 n.d. 0.1 n.d. 1495/Extra Virgin n.d. 11.1 0.8 3.3 77.2 5.8 0.8 0.3 0.3 0.1 n.d. 0.1 n.d. 1496/Extra Virgin n.d. 10.8 0.7 3.0 75.0 8.8 0.7 0.5 0.3 0.2 n.d. 0.1 n.d. 1551/Pure n.d. 10.5 0.7 3.6 74.9 8.6 0.5 0.4 0.2 0.2 n.d. 0.1 n.d. 1565/Pure n.d. 10.2 0.6 3.6 77.5 6.9 0.4 0.4 0.3 0.1 n.d. 0.1 n.d. 1566/Pure n.d. 8.8 0.5 2.9 75.3 11.5 0.5 0.3 0.2 0.1 n.d. n.d. n.d. SK1 n.d. 11.4 0.8 2.9 72.3 11.1 0.6 0.4 0.3 0.1 n.d. 0.1 n.d. SK2/Extra Virgin n.d. 11.1 0.7 3.0 74.5 8.9 0.7 0.5 0.3 0.2 n.d. 0.1 n.d. SK3/Light n.d. 10.6 0.7 3.3 72.0 11.4 1.1 0.4 0.2 0.2 n.d. 0.1 n.d. Olive oil maxd 0.1 20.0 3.5 5.0 83.0 21.0 1.5 0.8 m.ae 0.2 n.d. 1.0 n.d. aMean of three determinations. bNumber of carbon atoms : number of double bonds. CNot detected. dAdapted from IOOC, 1985 and FAO/WHO, 1970. eMinute amounts. 133 48 48 fi ,.. 134 46 c:; 1493 -:, I I > pure olive oil :;... 1489 I /c'J'J 'h 100% pure I ~II olive oil ~ 46

0 4 8 12 16 0 4 8 12 16

48 48 ,.. ,.. ~ 1494 s 46 1490 s 2: > pure olive oil c 100% pure 0 46 v; olive oil Vl

50 50 L 0 4 8 12 16 0 4 8 12 16

48 48 ...... 1493 ,.. 46 1491 ,... -:.1 ;:; extra virgin > 100% pure 2: c olive oil c olive oil Vl Vl 46

50 44 50 44

t~ 0 4 8 12 16 0 4 8 12 16

48 48

....,.. 1492 46 1496 ;:; olive oil "E > 46 CJ extra virgin c > oliYe oil Vl 0 Vl

50

0 4 8 1:! 16 0 4 8 12 16 Time (min) Time (min) 48 48 1551 46 SK1 I pure olive oil olive oil 46 135

44 so vJL 0 4 8 12 16 0 4 8 12 16

48 48 SK2 1565 extra virgin pure olive oil 46 46 olive oil

50 44 50

4 8 12 16 0 4 8 12 16

46 48 48 1566 SK3 pure olive oil i: i: ~ light olive oil ~> > 0 46 II) 0II)

44

n 4 8 12 16 0 4 8 12 16

48

GC2 46 virgin olive oil

50

0 4 8 12 16 Time (min)

Figure 4.16. Reversed-phase high performance liquid chromatograms of commercially available olive oil products. 136

Table 4.14. Triacylglycerol composition of commercially available olive oil products, as determined by direct RP-HPLC analysis3 •

Area% of peak with ECNb

Olive oil type 40 42 Ratio 46/44

1489/100% pure _c 1.2 3.4 1490/100% pure 1.2 3.5 2.2 1491/100% pure 0.9 3.6 1492 <0.5 4.2 1493/pure <0.5 5.0 1494/pure 1.0 3.4 1495/ extra virgin 5.8 1496/ extra virgin 0.9 4.3 1551/pure 1.1 3.6 1565/pure 4.2 1566/pure 2.0 3.4 SK1 1.0 3.8 SK2/ extra virgin 0.8 3.4 SK3/light 0.7 2.6 3.8

aMean of three determinations. bEeN = carbon number - 2 x number of double bonds. CN ot detected. 137 authenticity that was not apparent from the data of the other analytical methods.

For example five samples of commercial olive oils exceeded the limiting value (1 %) for area % of peak for ECN 42 ranging from 1.1 to 3.5. These samples were considered as adulterated based on results discussed in section 4.2.2.

Based upon the ratio of the area% of peaks for ECN of 46 to 44, an even higher number of samples appeared to be adulterated. Nine samples were assumed to be adulterated since their ratio of 46/44 ranged between 3.8 and 2.2, values below the proposed minimum of 3.9 for olive oils (see section 4.2.2). Confirmation of these results could be based on the additional determination of the sterol composition.

Two of the samples showed a high content of triacylglycerols with an ECN of 40 which should have been absent in olive oil. Sample 1490 had 1.2% of such triacylglycerols and sample SK3 had 0.7%. The appearance of a peak for ECN of 40 along with a high content of triacylglycerols with an ECN of 42 (3.5 and 2.6%, respectively, for the above samples) suggests that these two olive oils might be adulterated with soybean or canola oils, two seed oils with a characteristically high content of triacylglycerols with an ECN of 40. Interestingly, sample 1490 possessed the highest UV absorption (at both 232 and 270 nm), RI and linoleic (C18:2) and linolenic (C18:3) acid content (see sections 4.2.1, 4.2.2 and 4.2.3). 138

In Figure 4.17, the triacylglycerol profile of sample SK3 is compared to the profiles of an olive oil sample with guaranteed purity (sample GC2) mixed with 7.5% (w /w) soybean oil and 30% (w /w) canola oil. These two levels of adulteration were selected on the basis of quantitative data. All three samples contain 0.7% of triacylglycerols with ECN of 40. The content of triacylglycerols with ECN of 42 was 2.6, 2.3 and 2.7% for the samples SK3, 7.5% soybean and 30% canola in olive oil, respectively. By inspection of these profiles, it was concluded that sample SK3 is presumably adulterated with approximately 7.5% soybean oil, as the qualitative similarities in peaks for ECN of 42 and 44 indicate. 46 48

139

A

44

0 4 3 12 16 48

..... c: (I) ...... > 46 0 Uj B

0 4 8 12 16

48

..... c: (I) > 0 c -CJ) 46

0 4 8 12 16 Time (min)

Figure 4.17. Reversed-phase high performance liquid chromatograms of 30% (w /w) canola in olive oil (A), 7.5% soybean in olive oil (B) and commercial olive oil sample SK3 (C). 5. SUMMARY AND CONCLUSIONS

Several analytical techniques were employed to determine their effectiveness in detecting adulteration of olive oil with canola and other seed oils.

The determination of refractive index was able to detect the presence of the seed oils used in this study only when incorporated into olive oil at levels of 30% (w /w) or higher. Canola oil escaped detection even at a 60% (w /w) level in olive oil. The wide limits that have been proposed for the refractive index of olive oils contributed to the poor performance of this technjque. It should be emphasized at this point that the refractive index of 22 genuine olive oil samples with different origin was found to be quite invariable with a mean value of 1.4682 ± 0.0002 whereas, the upper limit for olive oils has been proposed to be as high as 1.4705.

The analysis of UV light absorbance detected ~5% (w /w) sunflower and soybean oil, ~10% (w /w) corn oil and ~12.5% (w /w) canola oil. Therefore, it was characterized as a more reliable method than the RI analysis for detecting seed oils in olive oil. However, only general conclusions could be drawn for the quality of an olive oil by this method. Measurements of the absorbance of an olive oil sample cannot differentiate between a poor quality virgin olive oil (absorbance limit for virgin olive oils at 270 nm: 0.25), a product labelled olive oil (limit: 0.90) and a sample adulterated with seed oils.

The results of fatty acid analysis could not be used to detect low levels of seed oils in olive oil but could detect lower levels of adulteration than the

140 141 results from refractive index. The content of linoleic (Cts:2) and linolenic

(Cts:3) acids were the only useful fatty acids for revealing adulteration of olive oil samples with seed oils.

Admixtures of 20% (w/w) sunflower and 25% (w/w) corn oil with olive oil found to have a linoleic acid content below the international accepted limit (21 %) for olive oils. On the basis of linoleic acid content, the method could detect the adulteration of olive oil with ~30% (w /w) soybean oil, whereas it was not feasible to detect adulteration with canola by checking the content of this particular fatty acid.

Soybean and canola oil were detected at lower levels in olive oil by using the content of linolenic acid. Admixtures of 10% (w /w) soybean and 12.5% (w /w) canola oil possessed a linolenic acid content of 1.5% which the upper limit for olive oils.

The RP-HPLC analysis of 22 olive oil samples led to the utilization of two characteristic factors for olive oils. Genuine olive oil samples contained only traces of triacylglycerols with ECN of 42 (max. 1 %) and therefore, any sample with >1% of triacylglycerols with an ECN of 42 could be considered as adulterated. The ratio of the area % of peak for ECN of 46 to area % of peak for ECN of 44 was considered as a fingerprint for olive oils and the mean of this ratio was 4.6 ± 0.3. The lowest value for this ratio in genuine olive oils was 3.9 and therefore, any olive oil sample with a value for this ratio of lower than 3.9 could be considered as adulterated. Incorporation of seed oils in olive oil would result in a reduction of this ratio due to the higher content of triacylglycerols with an ECN of 44 in all seed oils analyzed. 142

The RP-HPLC technique was superior to the previously discussed methods as it was found to be capable of detecting as low as 2.5% (w /w) of corn, sunflower and soybean oils, mainly because of the high content in these oils of triacylglycerols with an ECN of 42.

Due to the low content of triacylglycerols with an ECN of 42, canola oil was not detected by the RP-HPLC method when was present at less than 7.5% (w /w) in olive oil. At 2.5 and 5% (w /w) canola in olive oil the area % of the peak representing triacylglycerols with an ECN of 42 was lower than the limit (1 %) for olive oils. The calculation of the ratio of the area % of peak for ECN of 46 to area% of peak for ECN of 44 was also able to detect only higher than 7.5% (w /w) canola oil in olive oil, but it could reveal a larger number of the adulterated samples compared to the use of the area% of the peak for ECN 42.

When olive oil products available at the Canadian market were surveyed, the samples were found to be properly.labelled (not adulterated) according to analysis of the refractive index, the absorption of UV light, and their fatty acid composition. Utilization of the RP-HPLC method, however, indicated that 36% of the samples were adulterated based on the area % of the peak for ECN of 42, and based on the calculation of the proposed area % ratio of peaks for ECN 46 to 44, 64% of the commercial samples were considered adulterated.

From this study, it is apparent that the RP-HPLC analysis is a powerful method for detecting low levels of seed oils with a high content of linoleic acid such as corn, sunflower and soybean oils. However, the analyst should be aware that the detection of vegetable oils with a higher content of oleic 143 acid, such as canola oil, is more difficult by the direct analysis of an olive oil sample by RP-HPLC. Further studies are needed to examine other natural oils, products of biotechnology (low-linolenic acid canola varieties, high-oleic acid safflower and sunflower oils), and products of chemical treatments (hydrogenation, in teres terifica tion). 6. REFERENCES

Ackman, R.G. 1983. Chemical composition of rapeseed oil. In "High and Low Erucic Acid Rapeseed Oils", Kramer, J.K.G., Sauer, F.D. and Pigden, W.J. (eds), Academic Press, Toronto. pp 85-129.

Aitzetmuller, K. 1982. Recent progress in the high performance liquid chromatography of lipids. Prog. Lipid Res., 21:171-193.

Akihisa, T., Ghosh, P., Thakur, S., Rosenstein, F.U. and Matsumoto, T. 1986. Sterol composition of seeds and mature plants of family Cucurbitaceae. J. Am. Oil Chern. Soc., 63:653-658.

Amati, A., Zanirato, F.C. and Ferri, G. 1971. Determination of sterols in Italian Virgin olive oils-Application to purity control. Riv. !tal. Sostanze Grasse, 48:39-44.

Amelotti, G. 1987. Interesterification effect by physical refining of some olive oils. Riv. Ital. Sostanze Grasse, 64:223-226.

Andrikopoulos, N.K., Brueschweiler, H., Felber, H. and Taeschler, Ch. 1991. HPLC analysis of phenolic antioxidants, tocopherols and triglycerides. J. Am. Oil Chern. Soc., 68:359-364.

Anonymous, 1987. Determi~ation of sterol composition by capillary gas chromatography. Riv. !tal. Sostanze Grasse, 64:553-560.

AOCS 1989. Official Methods and Recommended Practices of the American Oil Chemists' Society, 4th ed., Champaign, Illinois.

Baltes, J. 1964. Classical chemical methods in fat analysis. In "Analysis and Characterization of Oils, Fats and Fat Products",Vol.1, Boekenhoogen, H. A. (ed), Interscience Publishers, New York. pp 1-58.

Barrett, C.B., Dallas, M.S.J. and Padley, F.B. 1962. The separation of glycerides by thin-layer chromatography on silica impregnated with silver nitrate. Chern. Ind., 1050-1051.

Barrett, C.B., Dallas, M.S.J. and Padley, F.B. 1963. The quantitative analysis of triglyceride mixtures by thin-layer chromatography on silica impregnated with silver nitrate. J. Am. Oil Chern. Soc., 40:580-584.

Batsanov, S.S. 1961. The history of refractometry. In ''Refracto- metry and Chemical Structure", Consultants Bureau Enterprises Inc., New York. pp 1-12.

144 145 Bezard, J.A. and Ouedraogo, M.A. 1980. Fractionation of oil triacylglycerols by reversed-phase high-performance liquid chromatography. J. Chroma­ togr., 188:279-293.

Canada Food and Drugs Act 1985. General standard for Vegetable fats and oils-Sections B.09.001 and B.09.003.

Carr, R.A. 1991. Development of deep-frying fats. Food Tech., 45:95-96.

Casadei, E. 1987. First results on detection of adulterated olive oil products with hazel-nut and/ or esterified oils by HPLC of triglycerides. Riv. Ital. Sostanze Grasse, 64:373-376.

Chaves Das Neves, H.J. and Vasconcelos, A.M.P. 1989. Characterization of fatty oils by pattern recognition of triglyceride profiles. HRC, 12:226-229.

Chobanov, D., Tarandjiska, R. and Chobanova, R. 1976. Rapid densitometric determination of triglyceride groups by argentation thin layer chromatography. J. Am. Oil Chern. Soc., 53:48-51.

Cocks, L.V. and van Rede, C. 1966. Oils, fats, fatty acids, and fatty alcohols. In "Laboratory Handbook for Oil and Fat Analysts", Academic Press, London. pp 80-189.

Colin, H., Guiochon, G. and Siouffi, A. 1979. Comparison of various systems for the separation of free sterols by high performance liquid chromatography. Anal. Chern., 51:1661-1666.

Cortesi, N., Rovellini, P. and Fedeli, E. 1990. Triglycerides of natural oils. Riv. Ital. Sostanze Grasse, 67:69-73.

Culp, T.W., Harlow, R.D., Litchfield, C. and Reiser, R. 1965. Analysis of triglycerides by consecutive chromatographic techniques. J. Am. Oil Chern. Soc., 42:974-978.

Damiani, P. and Burini, G. 1981. Determination of the triglyceride composition of olive oil by a multistep procedure. J. Agric. Food Chern., 28:1232-1236.

Daun, J.K. 1983. In "High and Low Erucic Acid Rapeseed Oils", Kramer, J.K.G., Sauer, F.D. and Pigden, W.J. (eds), Academic Press, Toronto. pp 175-180

de Vries, B. 1962. Quantitative separations of lipid materials by column chromatography on silica impregnated with silver nitrate. Chern. Ind., 1049-1050. 146 Descargues, G. and Bezard, J. 1981. Triacylglycerol structure of a consumer­ available virgin olive oil. Riv. Hal. Sostanze Grasse, 58:613-619.

DiBussolo, J.M. and Nes, W.R. 1982. Structural elucidation of sterols by reversed-phase liquid chromatography: I. Assignment of retention coefficients to various groups. J. Chromatogr. Sc., 20:193-202.

Dimoulas, K.A. 1981. General methods of analysis for fats and oils. In "Quality Control of Foods of Plant Origin-Part II", Publishing Organization of Instructive Books, Athens, Greece. pp 150-157.

Dong, M.W. and DiCesare, J.L. 1983. Improved separation of natural oil triglycerides by liquid chromatography using columns packed with 3J.Lm particles. J. Am. Oil Chern. Soc., 60:788-791.

Dutta, J., Das, A.K. and Saha, S. 1978. Enzymatic reactions on thin-layer chro-matographic plates-!. Lipolysis of triglycerides and separation of products on a single plate. J. Chromatogr., 154:39-50.

Eckey, E.W. 1954. Physical properties of fats. In "Vegetable Fats and Oils", American Chemical Society (ed), Reinhold Publishing Company, New York. pp 71-133.

El-Hamdy, A.H. and Perkins, E.G. 1981a. High performance reversed-phase chromatography of natural triglyceride mixtures. J. Am. Oil Chern. Soc., 58:49-53.

El-Hamdy, A.H. and Perkins, E.G. 1981b. High performance reversed-phase chromatography of natural triglyceride mixtures: critical pair separation. J. Am. Oil Chern. Soc., 58:867-872.

FAO/WHO 1969a. Recommended International Standard for Arachis Oil. Food Standards Program, CAC/RS 21, Rome, Italy.

FAO/WHO 1969b. Recommended International Standard for Maize Oil. Food Standards Program, CAC/RS 25, Rome, Italy.

FAO/WHO 1969c. Recommended International Standard for Sunflowerseed Oil. Food Standards Program, CAC/RS 23, Rome, Italy.

FAO/WHO 1970. Recommended International Standard for Olive Oil, Virgin and Refined, and for Refined Olive-Residue Oil. Food Standards Program, CAC/RS 33, Rome, Italy.

Fedeli, E. 1977. Lipids of olives. Prog. Chern. Fats and other Lipids, 15:57-74. 147 Fedeli, E. 1983. Miscellaneous exotic oils. J. Am. Oil Chern. Soc., 60:404-406.

Fedeli, E. and Jacini, G. 1971. Lipid composition of vegetable oils. Prog. Chern. Fats and other Lipids, 9:335-382.

Fiebig, H.J. 1985. HPLC separation of triglycerides. Fette Seifen Anstrichmittel, 87:53-57.

Firestone, D. 1988. General Referee reports on oils and fats. J. Assoc. Off. Anal. Chern., 71:76-79.

Firestone, D. 1989. General Referee reports on oils and fats. J. Assoc. Off. Anal. Chern., 72:80-83.

Firestone, D., Carson, K.L. and Reina, R.J. 1988. Update on control of olive oil adulteration and misbranding in the United States. J. Am. Oil Chern. Soc., 65:788-792.

Firestone, D., Summers, J.L., Reina, R.J. and Adams, W.S. 1985. Detection of adulterated and misbranded olive oil products. J. Am. Oil Chern. Soc., 62:1558-1562.

Flor, R.V. and Martin, B.D. 1989. Establishing adulteration of olive oil triglycerides analysis (abstract). J. Am. Oil Chern. Soc., 66:431.

Formo, M. W. 1979. Physical properties of fats and fatty acids. In "Bailey's Industrial Oil and Fat Products" Vol. 1., Swern, D. (ed), John Wiley and Sons, New York. pp 177-232.

Galanos, D.S. and Kapoulas, V.M. 1965. Detection of adulteration of plant oils with special reference to olive oil. J. Am. Oil Chern. Soc., 42:815- 816.

Galanos, D.S., Kapoulas, V.M. and Voudouris, E.C. 1968a. Application de la spectrophotometrie ultra-violette dans la region des 315 J.lm au controle des huiles: Detection de la falsification de l'huile d'olive par les huiles des grignons. Rev. Franc. Corps Gras, 15:291-300.

Galanos, D.S., Kapoulas, V.M. and Voudouris, E.C. 1968b. Detection of adulteration of olive oil by argentation thin layer chromatography. J. Am. Oil Chern. Soc., 45:825-829.

Geeraert, E. and DeSchepper, D. 1983. Structure elucidation of triglycerides by chromatographic techniques-Part 2: RP-HPLC of triglycerides and brominated triglycerides. HRC & CC, 6:123-132. 148 Geeraert, E. and Sandra, P. 1984. On the potential of CGC in triglyceride analysis. HRC & CC, 7:431-432. . ·

Geeraert, E. and Sandra, P. 1985. Capillary GC of triglycerides in fats and oils using a high temperature phenylmethylsilicone stationary phase-Part I. HRC & CC, 8:415-422.

Gegiou, D. and Georgouli, M. 1983. A rapid argentation tlc method for detection of reesterified oils in olive and olive-residue oils. J. Am. Oil Chern. Soc., 60:833-835.

Goh, E.H., Colles, S.M. and Otte, K.D. 1989. HPLC analysis of desmosterol, 7- dehydrocholesterol, and cholesterol. Lipids, 24:652-655.

Gracian, J.T. 1969. The chemistry and analysis of olive oil. In "Analysis and Characterization of Oils, Fats and Fat Products", Vol. 2, Boekenoogen, H.A. (ed), Interscience Publishers, London. pp 315-606.

Greek Code of Food, Beverages and Objects of Common Use 1975. Official Methods of Analysis for Foods and Beverages. General Chemical State Laboratory, Government Publications, B' /245.

Grob Jr., K., Neukom, H.P. and Battaglia, R. 1980. Triglyceride analysis with glass capillary gas chromatography. J. Am. Oil Chern. Soc., 57:282-286.

Grob, K., Lanfranchi, M. and Mariani, C. 1990. Evaluation of olive oils through the fatty alcohols, the sterols and their esters by coupled LC­ GC. J. Am. Oil Chern. Soc., 67:626-634.

Grundy, M.D. 1986. Comparison of monounsaturated fatty acids and carbohydrates for lowering plasma cholesterol. N. Engl. J. Med., 314:745-748.

Gutfinger, T. and Letan, A. 1974. Studies of unsaponifiables in several vegetable oils. Lipids, 9:658-663.

Hammond, E.W. 1981. The analysis of lipids: a personal approach to an analytical service. Chern. Ind., 710-715.

Hammond, E.W. and Irwin, J.W. 1988. The analysis of lipids by HPLC. In "HPLC in Food Analysis", Macrae, R. (ed), Academic Press, London. pp 95-132.

Hansbury, E. and Scallen, T.J. 1980. The separation of sterol intermediates in cholesterol biosynthesis by high pressure liquid chromatography. J. Lipid Res., 21:921-929. 149 Herslof, B. and Kindmark, G. 1985. HPLC of triglycerides with gradient elution and mass detection. Lipids, 20:783-790.

Herslof, B., Podlaha, 0. and Toregard, B. 1979. HPLC of triglycerides. J. Am. Oil Chern. Soc., 56:864-866.

Herslof, B.G. 1981. HPLC of triglycerides using UV detection. HRC & CC, 4:471-473.

Hitchcock, C. and Hammond, E. W. 1978. The determination of lipids in foods. In "Developments in Food Analysis Techniques-2", King, R. D. (ed), Applied Science Publishers Ltd., London. pp 185-224.

Hogg, R.V. and Tanis, E.A. 1988. Statistical tables. In "Probability and Statistical Inference", Macmillan Publishing Company, New York, U.S.A. pp 616-620.

Holen, B. 1985. Rapid separation of free sterols by reversed-phase high performance liquid chromatography. J. Am. Oil Chern. Soc., 62:1344- 1346.

Hunter, I.R., Walden, M.K. and Heftmann, E. 1978. Liquid column chroma­ tography of free sterols. J. Chromatogr., 153:57-61.

IOOC 1985. International trade standard applying to olive oils and olive­ residue oils. Document COI/T.15/NO 1.

IOOC 1990. Meeting of chemists to study methods of analysis for olive oils and olive pomace oils. Document T.20/no. 17-4; 18- 1,2,3 and 4.

IOOC 1991. Meeting of chemists to study methods of analysis for olive oils and olive pomace oils. Document T.20/no. 19-1, 2,4 and 5.

Itoh, T., Yoshida, K., Yatsu, T., Tamura, T. and Matsumoto, T. 1981. Triterpene alcohols and sterols of Spanish olive oil. J. Am. Oil Chern. Soc., 58:545-550.

IUPAC 1967. International Union of Biochemistry. J. Bioi. Chern., 242:4845- 4849.

IUP AC 1987. Standard Methods for the Analysis of Oils, Fats and Deriva­ tives. Applied Chemistry Division, Commission on Oils, Fats and Derivatives, 7th ed., Blackwell Scientific Publications, Oxford, England. 150 IUPAC 1991. Determination of triglycerides in vegetable oils in terms of their partition numbers by high performance liquid chromatography­ method 2.324. Pure Appl. Chern., 63:1178-1182.

Jacini, G. 1976. Problems in olive oil research. In "Lipids: Technology", Vol. 2, Paoletti, R., Jacini, G. and Porcellati, G. (eds), Raven Press, New York. pp 283-292.

Jeffrey, B.S. 1991. Silver-complexation liquid chromatography for fast, high­ resolution separations of triacylglycerols. J. Am. Oil Chern. Soc., 68:289- 293.

Jensen, G.W. 1981. Improved separation of triglycerides at low temperature by reversed-phase liquid chromatography. J. Chromatogr., 204:407-411.

Jimeno, S.A. 1982. The Spanish toxic syndrome. Trends Anal. Chern., 1:4-6.

Johnson, E. and Stevenson, B. 1978. Basic Liquid Chromatography. Varian Associates Inc., Palo Alto, California. pp 1-50; 270-317.

Jurriens, G. and Kroesen, A.C.J. 1965. Determination of glyceride composition of several solid and liquid fats. J. Am. Oil Chern. Soc., 42:9-14.

Kapoulas, V.M. and Andrikopoulos, N.K. 1986. Detection of olive oil adulteration with linoleic acid-rich oils by reversed-phase high­ performance liquid chromatography. J. Chromatogr., 366:311-320.

Kapoulas, V.M. and Andrikopoulos, N.K. 1987. Detection of virgin olive oil adulteration with refined oils by second-derivative spectrophotometry. Food Chern., 23:183-192.

Kapoulas, V .M. and Passaloglou-Emmanouilidou, S. 1981. Detection of adulteration of olive oil with seed oils by a combination of column and gas liquid chromatography. J. Am. Oil Chern. Soc., 58:694-697.

Kaunitz, H. 1978. Toxic effects of polyunsaturated vegetable oils. In "The Pharmacological Effect of Lipids", Kabara, J. (ed), American Oil Chemists' Society, Champaign, Illinois, U.S.A.

Kinsella, J.E., Posati, L., Weihrauch, J. and Anderson, B. 1975. Lipids in foods: problems and procedures in collating data. CRC Crit. Rev. in Food Tech., 5:299-324.

Kiritsakis, A. and Markakis, P. 1987. Olive oil: a review. Adv. Food Res., 31:453-482. 15 1 Kochhar, S.P. and Rossell, J.B. 1984. The Spanish toxic oil syndrome. Nutr. Food Sc., 90:14-15.

Kuksis, A. and Ludwig, J. 1966. Fractionation of triglyceride mixtures by preparative gas chromatography. Lipids, 1:202-208.

Kuksis, A., McCarthy, M.J. and Beveridge, J.M.R. 1963. Quantitative gas liquid chromatographic analysis of butterfat triglycerides. J. Am. Oil Chern. Soc., 40:530-535.

Kuksis, A., McCarthy, M.J. and Beveridge, J.M.R. 1964. Triglyceride composition of native and rearranged butter and coconut oils. J. Am. Oil Chern. Soc., 41:201-205.

Labuza, T.P. 1971. Kinetics of lipid oxidation in foods. CRC Crit. Rev. in Food Tech., 2:355-405.

Labuza, T.P. and Erdman, 1984. Fats, fatty acids, and cholesterol. In "Food Science and Nutritional Health: An Introduction", West Publishing Company, St. Paul, U.S. pp 65-74.

Latta, S. 1991. Gourmet oils in the 1990s. INFORM, 2:98-108.

Letan, A., Turzynski, B. and Raveh, A. 1965. Detection of peanuts in adulterated tehina and halva. J. Assoc. Off. Anal. Chern., 48:897-901.

Lie Ken Jie M.S.F. 1980. A quantitative treatment of saturated triglycerides by reversed-phase high-performance liquid chromatography. J. Chroma­ togr., 188:457-462.

Lim, C.K. 1986. Introduction to HPLC of small molecules. In "HPLC of Small Molecules", Lim, C.K. (ed), IRL Press Limited,Oxford, England. pp 1-5.

Litchfield, C. 1968. Triglyceride analysis by consecutive liquid-liquid partition and gas-liquid chromatography. Lipids, 3:170-177.

Litchfield, C., Farquhar, M. and Reiser, R. 1964. Analysis of triglycerides by consecutive chromatographic techniques. J. Am. Oil Chern. Soc., 41:588- 592.

Litchfield, C., Harlow, R.D. and Reiser, R. 1967. Gas-liquid chromatography of triglyceride mixtures containing both odd and even carbon number fatty acids. Lipids, 2:363-370.

Macrae, R. 1988. Theory and practice of HPLC. In "HPLC in Food Analysis", Macrae, R. (ed), Academic Press, London. pp 2-61. 152 MAFF 1990. Methods of Analysis for Olive Oil. Information Bulletin, No. 110, Food Science Division, Ministry of Agriculture, Fisheries and Food to Public Analysts, Norwitch, U.K. pp 39-43.

Mensink, R.P. and Katan, M.B. 1987. Eff«~ct of monounsaturated fatty acids versus complex carbohydrates on high-density lipoproteins in healthy men and women. Lancet, 122-125.

Miwa, T.K., Mikolajczak, K.L., Earle, F.R. and Wolff, I.A. 1960. Characteriza­ tion of fatty acids. Anal. Chern., 32:1~739-1742.

Myher, J.J., Kuksis, A., Marai, L. and Manganaro, F. 1984. Quantitation of natural triacylglycerols by reversed phase liquid chromatography with direct liquid inlet mass spectrometry. J. Chromatogr., 283:289-301.

Myher, J.L. 1978. Separation and determination of the structure of acylglyce­ rols and their ether analogues. In "Fatty Acids and Glycerides", Kuksis, A. (ed), Plenum Press, New York. pp 123-188.

Nawar, W.W. 1985. Lipids. In "", Fennema, O.R. (ed), Marcel Dekker Inc., New York. pp 139-244.

Olsen, E.D. 1975. Ultraviolet and visibl~~ spectrophotometry. In "Modern Optical Methods of Analysis", SumLmersgill, R.H. and LaBarbera, M. (eds), Mc-Graw-Hill Inc., U.S. pp 83-154; 411-430.

Pallota, U. 1976. Analytic problems i.n the ascertainment of olive oil genuineness. In "Lipids: Technology'', Vol. 2, Paoletti, R., Jacini, G. and Porcellati, G. (eds), Raven Press, New York. pp 387-393.

Pan, W.P. and Hammond, E.G. 1983. Stereospecific analysis of triglycerides of Glycine max, Glycine soya, Avena sativa and Avena sterilis strains. Lipids, 18:882-888.

·Parris, N.A. 1978. Non-aqueous reversed-phase chromatography of glyce­ rides using infrared detection. J. Chromatogr., 149:615-624.

Pauls, R.E. 1983. A time normalization study of the separation of_ olive oil triglycerides. J. Am. Oil Chern. Soc., 60:819-822.

Payne-Wahl, K., Plattner, R.D. and Spence~r, G.F. 1979. Separation of tetra-, penta-, and hexaacyl triglycerides by high performance liquid chroma­ tography. Lipids, 14:601-605.

Payne-Wahl, K., Spencer, G.F., Plattner, R.I). and Butterfield, R.O. 1981. High performance liquid chromatographic method for quantitation of free 153 acids, mono-, di- and triglycerides using an infrared detector. J. Chro­ matogr., 209:61-66.

Pei, P., Henly, R.S. and Ramachandran, S. 1975. New application of high pressure reversed-phase liquid chromatography in lipids. Lipids, - 10:152-156.

Petersson, B., Podlaha, 0. and Toregard, B. 1981. HPLC separation of natural oil triglycerides into fractions with the same carbon number and numbers of double bonds. J. Am. Oil Chern. Soc., 58:1005-1009.

Phillips, F.C., Erdahl, W.L., Nadenicek, J.D., Nutter, L.J., Schmit, J.A. and Privett, O.S. 1984a. Analysis of triglyceride species by high performance liquid chromatography via a flame ionization detector. Lipids, 19:142-150.

Phillips, F.C., Erdahl, W.L., Schmit, J.A. and Privett, O.S. 1984b. Quantitative analysis of triglyceride species of vegetable oils by high performance liquid chromatography via a flame ionization detector. Lipids, 19:880- 887.

Plattner, R.D. and Payne-Wahl, K. 1979. Separation of triglycerides by chain length and degree of unsaturation on silica HPLC columns. Lipids, 14:152-153.'

Plattner, R.D., Spencer, G.F. and Kleiman, R. 1977. Triglyceride separation by reversed-phase high-performance liquid chromatography. J. Am. Oil Chern. Soc., 54:511-515.

Plattner, R.D., Wade, K. and Kleiman, R. 1978. Direct analysis of trivernolin by high-performance liquid chromatography. J. Am. Oil Chern. Soc., 55:381-382.

Podlaha, 0. and Toregard, B. 1982. A system for identification of triglycerides in reversed phase HPLC chromatograms based on equivalent carbon numbers. HRC & CC, 5:553-558.

Pomeranz, Y. and Meloan C.E. 1978a. The Visble and Ultraviolet Regions. In "Food Analysis: Theory and Practice", AVI Publishing Company, Inc., Connecticut, U.S. pp 257-289.

Pomeranz, Y. and Meloan C.E. 1978b. Column Chromatography. In "Food Analysis: Theory and Practice", AVI Publishing Company, Inc., Connecticut, U.S. pp 257-289. 15 4 Prevot, A., Perrin, J.L., Laclaverie, G., Auge, Ph. and Coustille, J.L. 1990. A new variety of low-linolenic rapeseed oil; Characteristics and room­ odor tests. J. Am. Oil Chern. Soc., 67:161-164.

Privett, O.S. and Erdahl, W.L. 1975. Liquid chromatography of lipids. In "Analysis of Lipids and Lipoproteins", Perkins, E.G. (ed), American Oil Chemists' Society, Champaign, Illinois. pp 123-138.

Rana, M.S. and Ahmed, A.A. 1981. Characteristics and composition of Libyan olive oil. J. Am. Oil Chern. Soc., 58:630-631.

Rao, C.N.R. 1961. Simple and conjugated molecules. In "Ultraviolet and Visible Spectroscopy", Butterworth & Company Ltd., London. pp 15-38.

Rees, H.H., Donnahey, P.L. and Goodwin, T.W. 1976. Separation of C27, C28 and C29 sterols by reversed-phase high -performance liquid chromato­ graphy on small particles. J. Chromatogr., 116:281-291.

Robinson, J.L. and Macrae, R. 1984. Comparison of detection systems for the high-performance liquid chromatographic analysis of complex triglyce­ ride mixtures. J. Chromatogr., 303:386-390.

Robinson, J .L., Tsimidou, M. and Macrae, R. 1985. Evaluation of the mass detector for quantitative detection of triglycerides and fatty acid methyl esters. J. Chromatogr., 324:35-51.

Rodriguez, R.J. and Parks, L.W. 1982. Application of high-performance liquid chromatography separation of free sterols to the screening of yeast sterol mutants. Anal. Biochem., 119:200-204.

Rossell, J.B. 1991. Vegetable oils and fats. In "Analysis of Oilseeds, Fats and Fatty Foods", Rossell, J.B. and Pritchard, J.L.R. (eds), Elsevier Science Publishers Ltd., Essex, England. pp 261-327.

Schaefer, E.J. 1984. Clinical, biochemical and genetic features in familial disorders of high density lipoprotein deficiency. Arteriosclerosis, 4:303- 322.

Schulte, E. 1981. Separation of triglycerides according to chain length and degree of saturation by HPLC. Fette Seifen Anstrichmittel, 83:289-291.

Schulte, E. 1982. Detection of olive oil adulteration by HPLC. Lebensmit­ telchemie und Gerichtliche Chemie, 36:88-89. 155 Sheppard, A.J., Iverson, J.L. and Weihrauch, J.L. 1978. Composition of selected dietary fats, oils, margarines, and butter. In "Fatty Acids and Glycerides", Kuksis, A. (ed), Plenum Press, New York. pp 341-361.

Shukla, V.K. 1988. Recent advances in the high performance liquid chroma­ tography of lipids. Prog. Lipid Res., 27:5-38.

Smith, E.C., Jones, A.D. and Hammond, E.W. 1980. Investigation of the use of argentation high-performance liquid chromatography for the analysis of triglycerides. J. Chromatogr., 188:205-212.

Sonntag, N. 0. V. 1979. Structure and composition of fats and oils. In "Bailey's Industrial Oil and Fat Products" Vol. 1., Swern, D. (ed), John Wiley and Sons, New York. pp 1-98.

Statistics Canada 1980-1990. Oils and Fats. Ottawa. Supply and Services Canada.

Stolyhwo, A., Colin, H. and Guiochon, G. 1985. Analysis of triglycerides in oils and fats by liquid chromatography with the laser light scattering detector. Anal. Chern., 57:1342-1354.

Subcommittee on Vegetable Oils 1988. Proposed limits for sterol composition of olive oils. Riv. Ital. Sostanze Grasse, 65:39-40.

Subcommittee on Vegetable Oils 1989. Results of a collaborative study on gas chromatographic analysis of sterols by capillary columns. Riv. Ital. Sostanze Grasse, 66:531-538.

Synouri-Vrettakou, S., Komaitis, M.E. and Voudouris, E.C. 1984. Triglyceride composition of olive oil, cottonseed oil and their mixtures by low temperature crystallization and gas liquid chromatography. J. Am. Oil Chern. Soc., 61:1051-1056.

Takahashi, K., Hirano, T. and Zama, K. 1984. A new concept for determining triglyceride composition of fats and oils by liquid chromatography. J. Am. Oil Chern. Soc., 61:1226-1229.

Takahashi, K., Hirano, T., Egi, M., Hatano, M. and Zama, K. 1986. Suppleme­ ntary consideration of the triglyceride matrix model on reverse phase high performance liquid chromatography. J. Am. Oil Chern. Soc., 63:1543-1546.

Takano, S. and Kondoh, Y. 1987. Triglyceride analysis by combined argenta­ tion/nonaqueous reversed phase high performance liquid chromato­ graphy. J. Am. Oil Chern. Soc., 64:380-383. 15 6 Thomas, K.C., Magnuson, B., McCurdy, A.R. and GrootWassink, J.W.D. 1988. Enzymatic interesterification of canola oil. Can. Inst. Food Sci. Techno!. J., 21:167-173.

Thowsen, J.R. and Schroepfer Jr., G.J. 1979. Sterol synthesis-High pressure liquid chromatography of C27 sterol precursors of cholesterol. J. Lipid Res., 20:681-685.

Tiscornia, E. and Monacelli, R. 1970. Purity of olive oil by determination of the sterol composition. Scienza Alimentazione, 16:423-442.

Tracey, B. 1986. Lipids. In "HPLC of Small Molecules", Lim, C.K. (ed), IRL Press Limited, Oxford, England. pp 87-94.

Tsimidou, M. and Macrae, R. 1984. Influence of injection solvent on the re­ versed-phase chromatography of triglycerides. J. Chromatogr., 285:178- 181.

Tsimidou, M. and Macrae, R. 1987. Detection and quantitative determination of adulteration of olive oil-Fractional crystallisation as a preliminary stage prior to HPLC. International Analyst, 1:29-34.

Tsimidou, M., Macrae, R. and Wilson, I. 1987. Authentication of virgin olive oils using principal component analysis of triglyceride and fatty acid profiles: Part 2-detection of adulteration with other vegetable oils. Food Chern., 25:251-258.

Vaisey-Genser, M. and Eskin, M.N.A. 1982. Canola Oil-Properties and Perfor­ mance. Canola Council of Canada, no. 60, Winnipeg, Manitoba.

Van Den Bosch, G. 1973. Bleaching of vegetable oils: I. Conversions in soybean oil, triolein and trilinolein. J. Am. Oil Chern. Soc., 50:421-423.

Van Niekerk, P.J. and Burger, E.C. 1985. The estimation of the composition of edible oil mixtures. J. Am. Oil Chern. Soc., 62:531-538.

Vander Wal, R.J. 1960. Calculation of the distribution of the saturated and unsaturated acyl groups in fats, from pancreatic lipase hydrolysis data. J. Am. Oil Chern. Soc., 37:18-20.

Vonach, B. and Schomburg, G. 1978. High -performance liquid chromatogra­ phy with Ag+ complexation in the mobile phase. J. Chromatogr., 149:417-430. 157 Warner, K. and Mounts, T.L. 1990. Analysis of tocopherols and phytosterols in vegetable oils by HPLC with evaporative light-scattering detection. J. Am. Oil Chern. Soc., 67:827-831.

Wolff, J.P., Mordret, F.X. and Dieffenbacher, A. 1991. Determination of triglycerides in vegetable oils in terms of their partition numbers by high performance liquid chromatography. Pure Appl. Chern., 63:1173- 1177.

Zapsalis, C. and Beck, R.A. 1985. Lipids: chemistry, structure, and occurrence. In "Food Chemistry and Nutritional Biochemistry", John Wiley & Sons Inc., New York, U.S. pp 415-418.

Zygourakis, C. 1988. Personal communication. Elaiourgiki SYNPE, Athens, Greece.