STORAGE STABILITY OF DRUM DRIED NAVY BEAN POWDERS BY \\‘\ «it BEN?“ T ‘J " COUNTER A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Food Science 1969 ABSTRACT STORAGE STABILITY OF DRUM DRIED NAVY BEAN POWDERS by Ben T. Counter Atmosphere and retort cooked drum dried navy bean powders were investigated to determine their storage stability and to determine the nature of some of the physical and chemical changes occuring during storage. I Sensory evaluations of the powders indicated a maximum shelf life of between 90 and 120 days without the use of shelf stabilizing additives. Storage at 00 C. only increased the shelf life about 30 days over those stored at #00 C. Major changes were obtained in both the lipid and protein fractions and in the oxygen content of the headspace gas. The data indicated that headspace gas analysis might be the most promising test to use as a quality control method in the determining the rate of degradation during storage. ACKNOWLEDGMENTS The author extends sincere appreciation to Dr. C. L. Bedford for his helpful suggestions and counsel during this study. Appreciation and thanks are extended to the members of the guidance committee: Dr. L. R. Dugan and Dr. P. w. Bakker- Arkema. The author is grateful to his friends and associates who patiently discussed and directly aided in the work necessary in completing this study. Lastly, the author is especially grateful to his wife and family for their help, encouragement and sacrifices. II TABLE OF CONTENTS ACKNOWLEDGEMENTS. . . . . . . . LIST OF TABLES . . . . . . . . LIST OF FIGURES . . . . . . . . LIST OF APPENDICES . . . . . . . INTRODUCTION . . . . . . . . . REVIES OF LITERATURE . . . . . . METHODS AND MATERIALS . . . . . . Materials . . . . . . . . Beans. . . . . . . . Methods of Cooking. . . . Drum Drier . . . . . . Containers and Storage . . NethOdS O O O O O O O O 0 Physical and Chemical Evaluation Moisture Content . . . Head Space Gas Analysis . Color . . . . . . . Acid Value in Lipids . . Separation of Glycerides. Ultraviolet Absorption Spectra. Carbonyl Determination . Free Amino Groups in Lipid Fraction Biuret Protein Determination III Page II VII <.' H \Jfl'flflOOmmmU‘U‘m-F'P-F‘P-F'NH Page Colorimetric Determination of Sugars . . . 8 5-Hydroxymethylfurfural Determination. . . 8 Sensory Flavor Evaluation. . . . . . . . Selection of Panel . . . . . . . . . Preparation of Powders. . . . . . . . \O\O\O\O Method of Sample Presentation . . . . . Statistical Analysis of Test Results . . . . 10 RESULTS AND DISCUSSION . . . . . . . . . . . . 11 Physical Evaluation . . . . . . . . . . . 11 Chemical Evaluation . . . . . . . . . . . 1n Lipid Oxidation . . . . . . . . . . . 1h Non-Enzymatic Browning. . . . . . . . . 21 Sensory Flavor Evaluation . . . . . . . . . 25 General Discussion . . . . . . . . . . 26 SUMMARY AND CONCLUSIONS. . . . . . . . . . . . 27 PROPOSALS FOR FURTHER RESEARCH . . . . . . . . . 29 LITERATURE CITED . . . . . . . . . . . . . . 30 APPENDICES . . . . . . . . . . . . . . 37 IV Table LIST OF TABLES Page Free Fatty Acid Values in Lipids, in Milligrams Potassium Hydroxide per 100 Grams Powder. . . 14 Percent Absorption of Conjugated Diene and Triene Lipid Components at 242 mu . . . . . . . 16 Micrograms Free Amino per Gram of Powder. . . 21 Milligrams of Boiling Water Extracted Sugar per Gram of Powder . . . . . . . . . . . 24 Appendix A. B. LIST OF APPENDICES Page Hunter Color Difference Meter Values . . . . 37 Carbonyl Values in Micro-Moles per Gram Bean Powder . . . . . . . . . . . . 38 Sensory Flavor Panel Values . . . . . . . 39 VI LIST OF FIGURES Figure Page 1. The Effect of Storage Time and Temperature on the Oxygen in the Head Space . . . . . . 13 2. The Effect of Storage Time and Temperature on the Ultraviolet Absorption of Conjugated Lipid Components . . . . . . . . . . . 17 3. The Effect of Storage Time on the Quantity of Total Carbonyls Present. . . . . . . . . 19 4. The Effect of Storage Time on the Values of the Biuret Test . . . . . . . . . . . 23 VII INTRODUCTION The use of legume seeds as a food source in the United States has been on the decline for the last decade, according to the U.S.D.A. 1965-1966, Report Number One, of Food Con- sumption of Households in the United States. According to Muneta, (1964), one of the contributing factors to this de- cline probably can be attributed to the long soak or cook time required to prepare these products. One method of reducing preparation time and increasing the convenience of a legume seed is to produce an instant reconStituting powder by drum drying, as described by Bakker- Arkema, et al., (1968). In addition, when cooked prior to drum drying, the destruction of trypsin inhibitor and hemagglu- tinin and increase in nutritive value of these powders should help to make them a very desirable food source, (Liener, 1962i. Shelf life of these powders has not been clearly es- tablished, although Boggs, et al., (196M), reported drum dried Pinto bean powder had less than a 66 day period of flavor stability when stored at 32°C. in hermetically sealed tin cans with an atmosphere pack. Burr, et al., (1969), indicated that shelf stability of drum dried lima bean powders is less than 60 days when stored in an air pack without antioxidants. Neither of these two forementioned references determined the nature of the deterioration in these powders, but both have had success in the use of antioxidants and inert gas packs to markedly increase storage stability of powders. l 2 Because drum dried legume powders have commercial possibilities, this study was conducted to determine the shelf life of atmosphere and retort cooked Navy bean powders in an atmosphere pack at different storage temperatures. An attempt was made to determine the nature of the flavor degradation of these powders by chemical analysis of the carbohydrate, lipid, and protein fractions, and by sensory flavor evaluation. REVIEW OF LITERATURE According to Ponting, et al., (1964). there are three broad classes of chemical deterioration that occur in dried vegetables, the first two of which will be considered here. They are: (1) lipid and essential oil reactions, including mainly oxidation, isomerization, and hydrolysis, and (2) non- enzymatic browning reactions involving carbonyl containing compounds, organic acids, and nitrogenous compounds. Deterioration of lipids is known to cause rancidity, and by-products have been shown to cause off flavors by Buttery and Teranishi, (1963), Tarladgis, et al., (1960), and Lee and Wagenknecht, (1951). Some of the lipid oxidation reactions do not require water and in fact are accelerated at very low moisture contents, according to Ponting, et al., (1964). Feustel, et al., (l96u), state that: (l) the rate of oxidation is not affected by temperature of storage to any marked degree (the rate is increa- sed only about 1.2-1.4 times for every 18°F. temperature rise); (2) packing in nitrogen greatly retards oxidative change, but has 3 little effect on rate of browning: and (3) lowering the moisture content accelerates oxidative deterioration but retards non- enzymatic browning. In connection with the moisture effect. Brunnauer, et al., (1938) and Salwin, (1959, 1961), have shown evidence that moisture content for maximum resistance to oxi— dation is at the value corresponding to a monomolecular layer of water within the internal structure of the solid, and this value varies with different materials. Koch, (l962),gives this value as 5.21% water (as is basis) for Navy Beans. Lundberg (1962), mentioned the effects of extraneous in- fluences, such as trace metals, in acting as pro-oxidants on lipids. In addition, he summarized the initial stages of the lipid auto-oxidation reaction and the effect of unsaturation. Legume seeds are high in phospholipids (White, 196A), and Koch. (1962), cites evidence that shows phosphatides have an anti- oxidant effect, perhaps by enhancing the protective action of protein on other lipids. The second class of deterioration of importance in powders is non-enzymatic browning. Hodge, (1953) has divided this into three groups: (1) carbonyl-amino reactions, includ- ing those of aldehydes, ketones, and reducing sugars with amines, amino acids, peptides, and proteins; (2) caramelization (which will not be considered here): and (3) oxidative reactions which for example, convert ascorbic acid and polyphenols into di- or poly-carbonyl compounds. Ponting, et al., (196h), stated that browning of this L, sort can be inhibited by removal or combination of carbonyl groups, and often by combination of amino groups, when these are major contributors to browning. They further stated that lowering the storage temperature is very effective in controling browning, giving a decrease of 5-7 fold for every 18°F. drop in temperature. This statement was qualified in that it is product moisture content dependent, the lower the moisture, the less the browning. METHODS AND MATERIALS Phaseolus vulgaris, Egg. Michelite, Navy beans of the 1968 crop, stored at 32°F. until time of drying, were used in this experiment. Methods of Cooking Two types of cooks were used: (1) Atmosphere cook- beans were soaked and cooked in water at 210°F. for 90 minutes. (2) Retort Cook- beans were soaked in water at 210°F. for #5 minutes, placed in a wire basket, and retorted in steam at 220°F. for 30 minutes. mass; An American 12" x 19 1/8" double drum drier, manufac- tured by the Overton Machine Company of Dowagiac, Michigan, was used. This drier was set up to simulate a single drum operation by running one drum cold (Bakker-Arkema, et a1. 1966). Drying conditions of 23 1/3 r.p.m., 4 layers, and 85 p.s.i.g. were 5 maintained for both cooks. Moisture content was 5.7% and 6.9% for the atmosphere and retort cook powders respectively. After drying, the sheets were pulverized with a Fitzpatric comminuting mill using a 0.125" screen, and were thoroughly mixed to obtain a homogenous product. Containers and Storage Three hundred grams of powder were filled into No. 2 (307 x #09) cans, and hermetically sealed with a semi-automatic double roll closing machine. The containers were then randomly distributed into three different storage cabinets with air cir- culation and maintained at 0° : 1°C., 20° 1 0.500., and #00: O.5°C. Five random samples were removed from each of these cabinets at O, 30, 60, 90, and 120 days for analysis and evalu- ation. Physical and Chemical Evaluation Moisture Content: Five 10 g. samples of each cook were taken from containers 12 hours after sealing, and the moisture content was determined using the AOAC, sixth ed. (1945), method for dried vegetables. Moisture was also deter- mined by this method on two samples of each storage temperature lot at the end of the experiment. Head Space Gas Analysis: A Beckman Model 778 Process Oxygen Analyzer, using a Beckman 76365 Polargraphic Oxygen Sensor in conjunction with a Beckman Head Space Sampler, was used to determine oxygen content of the cans. Color: Color of the powders was determined using a Hunterlab D-25 L Color Difference Meter. A yellow tile, 6 number 2814, with values of L=83.0, a= - 3.5, and b= 26.5 was used as a standard. Acid Value in Lipids: The method for grain of Baker, et. al., (1957) was used for the determination of fat acidity as an index of lipid deterioration. Forty grams of powder“were _extracted with 100 ml. of benzene in a Waring blender, filtered through Whatman No. 2 filter paper (funnels were covered with foil to prevent excessive evaporation), a 25 ml. sample pipetted into an Erlenmeyer flask, and titrated with 0.0356 N potassium hydroxide to the phenolphthalein endpoint described by Baker, et.al. (1957). Separation of Glycerides: Thin layer chromatography (TLC) of glycerides was employed using the procedure recommended by Eastman Kodak Co., (1966), to show the development of free fatty acids to further confirm the acid value titration. A 2.00 g. sample was thoroughly mixed with 4.0 ml. of 2:1 (v/v) chloroform/methanol in a 40 m1. standard taper (S. T.) cen- trifuge tube using a Scientific Products Test Tube Mixer. These tubes were then centrifuged at 2500 r.p.m. in an I.E.C. Model U centrifuge using a number 279, 12 tube, 50 ml. head. The relative centrifugal force at tip was 1570 g. A 0.05 ml. (50 ul) aliquot of the supernatant was applied to Eastman 5051 Silica Gel Sheets, and developed in 80:20:l (v/v/v) hexane/diethyl ether/acetic acid solvent for thirty _ minutes. Knowns were also spotted, and the visualizing reagent was 0.2% 2',7'-Dichlorofluorescein in ethyl alcohol, used with long wave length ultraviolet light (3660 A). 7 Ultraviolet Absorption Spectra: Conjugated diene and triene lipid components (hydroperoxides and secondary decom- position products) were determined by recording the ultraviolet absorption spectra from 220 m u. to 320 mu,, (Privett, et al., 1953, 1954). A Beckman Model DB scanning spectrophotometer with a scanning speed of 40 mu" per minute was used. A 2.00 g. sample was extracted with 10 m1. of 2:1 (v/v) chloroform/ methanol in a 40 m1. S.T. centrifuge tube and centrifuged as previously outlined. A 3 ml. aliquot was placed in the cuvette, 3 ml. of chloroform/methanol used as a blank, and maximum absorp- tion was determined to occur at 242 mu. Carbonyl Determination: 15 ml. of carbonyl free benzene, ca. 2.5 g. anhydrous sodium sulfate, and 5.00 g. of bean powder were thoroughly mixed into a 40 ml. S.T. centrifuge tube using the test tube mixer, and centrifuged as previously described under separation of glycerides. A 5.0 ml aliquot of supernatant was carefully withdrawn, and the method of Henick, et a1. (1954), was followed. Free Amino Groups in Lipid Fraction: The method of Siakotos, (1967), for the determination of free amino groups in lipids was used, with a 0.20 g. sample of powder added to 4.0 ml. of 2:1 (v/v) chloroform/methanol in a 40 m1. S.T. centrifuge tube. The samples were centrifuged as previously described, and a 1.0 ml. aliquot of supernatant analyzed. Biuret Protein Determination: A 0.5 g. sample of powder was thoroughly mixed with 10 ml. of 1:1 (v/v) 1 N. sodium hydroxide/8 M. Urea using the test tube mixer, allowed 8 to stand for thirty minutes to insure protein extraction, and then centrifuged. A 0.5 ml aliquot of supernatant was pipetted into a B & L Spectronic 20 cuvette, and 5.0 ml of Biuret rea- gent (E & M Chemicals of Brinkmann Instruments, Westbury, New York) added. After mixing on a test tube shaker, color was allowed to develope for thirty minutes, and then read at 545 mu. A standard curve was made using protein serum supplied with the Biuret reagent. Colorimetric Determination of Sugars: The method of DuBois, et al., (1956), was used. The sample was prepared as follows: A 1.00 g. sample of powder was blended with 50 ml. boiling water in a Waring blender for three minutes, and fil- tered through Whatman No. 5 filter paper into a 125 m1. Erlen~ meyer flask (funnels were covered with foil during filtration to prevent excessive evaporation). A 1.0 ml. aliquot was re- moved from the filtered sample and diluted to 100 ml. In a volumetric flask with distilled water. After thoroughly mixing, a 1.0 ml. sample was transferred to a 1.6 mm cuvette and, 1.0 ml. of 5% redistilled phenol in water, and 5.0 ml. of concentra- ted sulfuric acid were added. A standard curve was made using D-glucose. 5-Hydroxymethylfurfural Determination: This method was not used until the 90 day storage analysis period, and at this and later dates (120 day storage period) no differences were noted between the three storage temperatures. A 1.00 g. powder sample was mixed into 25 ml. distilled water in a 40 m1. centrifuge tube and centrifuged as previously described. 9 5 ml. of supernatent were carefully removed, transferred to a 40 m1. 8. T. centrifuge tube for a reaction vessel, and the method of Keeney, et al., (1959) followed for the determination of both free and total 5-Hydroxymethy1furfural (HMF). Sensory Flavor Evaluation Selection of Panel: Twenty four randomly selected persons from the Michigan State University Food Science Depart- ment were selected as an untrained flavor panel. Preparation of Powders: 375 8. of powder (75 8. from each can), 11.25 g. of salt, and 1500 g. of boiling water were thoroughly mixed in a 2000 ml. beaker, covered with foil, and placed in a 60° C. constant temperature water bath. As indi- vidual panalists arrived to taste, the three beakers from the different storage temperatures were uncovered, one ounce samples withdrawn, and the beakers recovered and placed back in the water bath to insure uniform temperature and texture for each sample. Method of Sample Presentation: A flavor difference procedure and a random method of presentation was divided into three parts to eliminate bias on the part of the judges: the first eight panelists to arrive received three randomly numbered samples, representing 0°, 20°C and 40°C stored powder and a lettered reference sample, ref, that was identical to the first sample: the second eight panelists to arrive received three num- bered samples, and a lettered reference sample, ref, that was identical to the second sample: the last eight panelists to arrive received three numbered samples, and a lettered reference sample, ref, that was identical to the third sample. (In all 10 tests, the first sample was 0°C. Storage powder, the second sample was 20°C. storage powder, and the third sample was 40°C. storage powder). All judges were instructed to compare the flavor of the numbered samples with the reference sample. In each case the difference between the numbered sample and the reference was judged on a scale of five: much worse: worse: no difference: better: and much better. Comments about any or all samples were solicited on the ballots. Statistical.Analysis 2; Test Results Analysis of variance using a factorial design (Little and Hills, 1966) including cooks, temperature, and time was used to determine significant differences in: (1) Head space gas: (2) Acid value in lipids: (3) Ultraviolet absorption spectra: (4) Carbonyl determination: (5) Free amino groups in lipids: (6) Biuret protein analysis: and (7) Water extrac- ted sugars. A correlation coefficient (Little, 1966) was deter- mined between all physical and chemical tests that showed differences at the 5% level or less. Flavor panels were analyzed individually using the above mentioned procedure, but judges, references and tempera- tures were used as factors, following the procedure of Wiley, et al., (1957). 11 RESULTS AND DISCUSSION Physical Evaluation Moisture content was 6.9% and 5.7% for the retort and atmosphere cook respectively. There was no change in this value with storage. As the powders were both dried under the same drum conditions, some reasons for differences in moisture may be: (1) differences in moisture retention ability during drying between the two cooks: (2) differences in relative humidity on the two days of preparation: and (3) differences in solids content between the two cooks at the time of application to the drum. Salwin, (1959), states that the moisture content for maximum resistance to oxidation is the value corresponding to a monomolecular layer of water within the internal structure of the solid (Brunnauer, et al., 1938). This value varies with different materials. Ponting, et al., (1964), gave optimum values of 6.5% for potatoes, 3.5% for protein foods like meat, and near zero for fruits and high sugar type foods. Since the monomolecular layer moisture content of these beans is known to be around 5.21% (Koch, 1962) and the retort cook powder was higher in moisture than the atmosphere cook powder (which was close to the optimum value), one might have expected it to have more browning than the atmosphere cook, when in fact they were about equal when comparing Hunter Color-difference values. Tests discussed later in this text may suggest reasons for this lack of difference. 12 The retort cooked bean powders had a slightly darker (lower L value) color than the atmosphere cooked powders, but less than one full unit on the Hunter Color Difference Meter. Readings on the "a" scale seemed to increase slightly in negative values, and ”b" readings seemed to increase slightly in positive values in comparing 40°C. storage temperatures with 0°C. storage temperatures in the latter part of this experiment. However in trying to calculate this change either as hue (a/b) or chroma (\/a2 :,b2 ), no meaningful values could be found because of the neutral "a" scores giving large differences when in fact none existed, (Appendix A). Headspace oxygen analysis was the single test in this experiment that showed the largest changes with both storage temperature and time (significant at the 1% level), (Figure I). There is a difference (5% level) between the two cooks, the retort cook showing an increase in oxygen absorption over the atmosphere cook. Figure I. A logical assumption from this test would be that the use of antioxidants such as butylated hydroxyanisole (B.H.A.) and butylated hydroxytoluene (B.H.T.), as well as gas impermeable containers, may add considerably to the shelf life of these powders. The results of Boggs, et al., (1964), and Burr, et al., (1969), are in agreement with this conclusion. In the future, the use of some instrument that has more versatility than the polarographic oxygen electrode in measuring other gases would also be advantageous. This instrument should quantify such things as C02 liberation that might occur in a H m g 100 (IS-H (:0 221 90 m0 p‘H S<3 8° 0 p 55 70 We) 60 OD-: 50 E 100 (0:: {6-H .pff 90 $354 30 3“ 80 00 0+3 CS: 70 (DO) 2:: 094 50 13 0° c. 20° c. 40°C. 30 60 90 120 Days Storage, Atmosphere Cook 00 c. K 00° c. 30 60 90 120 Days Storage, Retort Cook Figure l. -- The effect of storage time and temperature on the oxygen in the head space. (Data shown represents mean values of 5 replicates) 14 non-enzymatic browning reaction (Cole, 1967), as well as loss of oxygen. Chemical Evaluations Lipid Oxidation: Fat acidity values (Table I) showed a significant loss at the 5% level with time. As one expects an increase in fat acidity with time, it would seem to suggest that the free fatty acids, instead of accumulating, were in fact degrading or becoming unavailable for this determination. In low moisture content powder as is present here, this may be the case, as one would not expect much hydrolysis to occur. This test does indicate that there was a difference between the two cooks at the 5% level, the atmosphere cook powders having higher values. Table l. -- Free Fatty Acid Values in Lipids, in mg. KOH/lOO g. Powder (data shown represent mean value of 5 replicates) Retort Cook Atmosphege Cook Time Temperature 0 3O 60 90 120 0 ‘_30 60 90 120 Days Days Days Days Days Days Days Days Days Days 0° c. 70.0 68.1 62.3 62.5 62.1 68.7 65.8 65.8 69.4 67.6 206 C. 70.0 63,5 61.9 62.3 61.6 68.7 62.8 65.3 67.4 66.2 400 c. 70.0 64.5 61.2 61.2 62.3 68.7 63.6 67.0 70.8 66.4 Mean 70.0 65.4 61.8 62.0 62.0 68.7 64.0 66.0 69.2 66.7 70.0 65.4 62.0 62.0 61.8 5% 69.2 68.7 66.7 66.0 64.0 5% 15 In comparing values calculated in this test with those given by Baker, et al., (1957), for raw beans, it becomes obvious that a large amount of hydrolysis of the lipids must occur during soaking and cooking. Koch, (1962), confirms this, suggesting that in preparation of food materials for dehydration, as was done here, derangements of the natural protective equilibrium of fats may occur, including rupture of lipid micelles and smearing of fat upon new surfaces. It is also to be noted that no large changes in values occurred in this test with changes in time and temperature, and as it is a highly empirical procedure, chances are there would be greater differences between different technicians and laboratories than there would be tetween a wholesome and degraded product. The TLC procedure used in conjunction with the above fat acidity test to confirm the accumulation of free fatty acids was not at all satisfactory. Results were variable, with some unexplainable changes in Rf values occurring between some samples and not in others. As the lipid samples spotted were a composite of all materials soluble in chloroform/methanol, and the test was sensitive only to groups or classes of glycerides (ie, mono-, di-, tri glycerides, and free fatty acids), this variability may have been due to contaminants within the solvent extraction. One other problem was that the majority of the lipids present in these legume seeds are phospholipids (White, et al., 1964) and they did not migrate from where they were spotted. If any additional TLC work is to be done on these powders, cleaning up of the lipid extract and investigation of 16 the phospholipids would be highly desirable. The phosphatides should be of particular interest because of their implication in acting as anti-oxidants, or as a first component to be oxi- dized, as was suggested by Koch, (1962). Because of the unsat- isfactory results of the TLC procedure, no results of this pro- cedure will be reported here. According to Corliss, (1968), oxidation of polyunsatur- ated fatty acids is accompanied by increased ultra-violet ab- sorption in the 232 mu range due to the formation of conjugated diene and triene hydroperoxides. The higher the absorption, the greater has been the exposure to oxygen. Both Figure II and Table 2 show there was not only a significant change in the U. V. absorption with time and temperature, but also a significant difference between cooks. Table 2. -- Percent Absorption of Conjugated Diene and Triene Lipid Components at 242 mu Retort Cook Atmosphere Cook Iflme Temperature 0 30 60 90 ‘I20 Mean 0 30 60 90 ITO Mean Days Days Days Days Days Days Days Days Days Days 0° c. 67 64 72 63 59 65.0 69 71 73 80 69 72.5 20° c. 67 68 80 73 69 71.4 69 80 84 83 67 76.7 40° c. 67 63 74 75 64 68.6 69 74 74 79 57 70.7 Mean 67.0 65.0 75.4 70.4 64.0 69.0 75.0 77.0 80.7 64.4 Time 7504 7004 6700 6500 6400 8007 77.0 7500 6900 6404 l 5% Z Temperature 71.4 68.6 65.0 5% 76.7 72.5 70.7 5% Absorption in Percent Absorption in Percent Figure 2. 17 90 8 70 --00 Co 60 --200CO 50 --uOoCo 0 30 60 90 120 Days Storage, Atmosphere Cook 90 80 --200 Co 70 ’\\\ --uog Co 60 ---0 Co 50 ' 0 30 60 90 120 Days Storage, Retort Cook --The effect of storage time and temperature on the ultraviolet absorption of conjugated diene and triene lipid components at 242 mu. 18 The change in time and temperature may be predicted (Koch, 1962, and Feustel, et al., 1964), but the change in cooks needs further study. It was one of several indications (oxygen uptake, fat acidity, and carbonyl values) that there was a change in the lipid systems due to each of two methods of cooking, and the taste panel results indicated the atmosphere cook was slightly less stable than the retort cook, by a margin of up to 30 days. The panelists uniformly rejected the atmosphere cook 40°C. storage temperature at a period of 90 days, while this phenomenon did not occur in the retort cook until 120 days of storage. There may be several possibilities that contribute to this difference in cooks: (1) an increase in the oxidation and/or hydrolysis of one cook compared to another: (2) be- cause of the higher heat of the retort cook, sulfhydryl groups may be released (as is observed in high temperature treated milk products, Coulter and Jenness, 1964) and influence the stab- ility of the lipids in this cook: and (3) the possibility that the cooking water had a higher iron concentration during the day of the atmosphere cook compared to the water On the day of the retort cook, due to the variable quality of the water available in the processing laboratory. Further confirmation that the cooks caused a change in lipid structure, when one is compared to the other, is also presented in Figure III showing the results of the carbonyl determination. There was again a highly significant difference between cooks, with a variation occuring in both saturated and 19 a 0.) U 3 o a. 0D 2 0 5 7 2 4-4 65 g o E 60 ---Atmosphere Cook :3 C 55 8 85° 0 / :3 D5 ---Retort Cook 2 g 40 f g C) 0 30 60 90 120 Days of Storage Figure 3. -- The effect of storage time on the quantity of total carbonyls present. (Data shown represents mean values of 15 replicates) 20 unsaturated carbonyls (see Appendix B) even at time zero, showing that a change in lipid and other carbonyl structures did occur between the two methods of cooking. Unfortunately, these values dropped (after reaching an initial high around 60 days) back to near "time zero" values. This eliminated this test as a method of showing storage degradation for quality control procedures. The results of the free amino group analysis in the lipid fraction, showed (as did the fat acidity and carbonyl determinations) a decrease in value with storage time. This is in further agreement with Feustel, et a1. (1964), that oxidation is somewhat insensitive to temperature. Corliss, (1968), suggests that free amino groups in lipids is involved in phospholipid oxidation, the free amino group value of cer- tain phosphatides being reduced significantly during oxidation, indicating a possible Maillard-type browning and/or interaction with carbonyl groups. However, no statistical correlation between carbonyls, peptide bonding, or free amino groups was found. 21 Table 3. -- Micrograms Free Amino per Gram of Powder (Data shown represent mean values of 5 replicates) Retort Cook Atmosphere Cook Temperature _TIme 0 30 60 9f 120 0 30 60 90 120 Days Days Days Days Days Days Days Days Days Days 0°C. 141.0 142.8 140.4 140.2 140.4 136.7 142.2 139.0 138.0 137.2 20°C. 141.0 142.1 138.4 139.7 139.3 136.7 141.8 137.9 139.5 138.7 40°C. 141.0 141.0 141.8 139.9 140.9 136.7 146.8 142.6 144.9 138.3 Mean 141.0 142.0 140.2 140.0 140.2 136.7 143.6 140.2 140.8 138.1 Atmosphere Cook 143.6 140.8 140.2 138.1 136.7 5% Non-Enzymatic Browning The water soluble protein content decreased with time of storage (Figure 4). There was also a difference between cooks with the retort cook being slightly low. The most interesting feature of the Biuret test was that the change during storage showed a positive correlation at the 1% level (r = 0.959) with the oxygen level in the head space gas anal- ysis. On the first thought one might think that there was an oxidation occuring of the peptide bonding (which the Biuret test measures) with time, but after some deliberation, one must conclude this could only be a minor contribution at best. The decrease in peptide bonds available to react with the Biuret reagent then, must come about as a result of this bonding being less available to the reagent. As a polar solvent is used in extracting the proteins (1 N. NaOH/ 8 m. urea), the 22 most plausible explanation was that the proteins were becoming less soluble in polar solvents with time. There are several possible reactions that commonly decrease the solubility of proteins in biological systems: (1) the condensation of (lipid) carbonyls with amino acids, (Braverman, 1963 and Jevons, 1964): and (2) the formation of hydrophobic lipid-protein bonds, (Corn- well and Horrocks, 1964): and (3) the formation of protein di- sulfide crosslinks (Wall, 1964). The effect of the condensation of reducing sugars and amino acids will be discussed under the boiling water extracted sugar test, and results indicated not much of this occurred in these powders. 23 25 24 23 22 --Atmosphere Cook 21 20 19 Retort Cook ................ 18 Percent Water Soluble Protein 17 16 0 30 6O 90 120 Figure 4. -- The effect of storage time on the values of the Biuret Test (Data shown represents mean values of 15 replicates) 24 Because of the probable decrease in availability of the peptide bonds, this test has very limited value in measuring non-enzymatic browning in these powders. The use of more sensitive and dependable protein tests are recommended in the study of future browning reactions. There was one redeeming feature of this test. It may be an indirect indicator of oxidation. However, work will have to be done to improve precision of this method before results (see Appendix A for raw data) between technicians and laboratories could be meaningfully compared. The boiling water extracted sugars were significantly lower in the atmosphere cooked powders than in the retort cooked powders, as might be expected, due to probable excessive leaching in the atmosphere cook (table 4). However there was no significant change with time or temperature which would indicate non-enzymatic browning. As this is a highly sensitive test (DuBois et al., 1956), one must conclude that sugar-amino acid condensation did not occur to any marked degree. Perhaps other more specific tests for reducing sugars (ie, Ting, 1956, as modified by Furoholmen, et al., 1964) may be more applicable to these powders. Table 4. -- Milligrams of Boiling Water Extracted Sugar per Gram of Powder. (Data shown represents mean values of 5 replicates) Retort Cook Atmosphere Cook ‘17 *ifiy"* 60 90' 120 0 30 60 90 120 Days Days Days Days Days Mean Days Days Days Days Days Mean 0° 0. 89.8 94.8 88.0 91.7 90.3 91.0 80.7 77.7 81.2 82.0 80.8 80.5 20° c. 89.8 100.3 86.2 86.0 86.8 89.9 80.7 80.0 83.9 85.4 82.3 82.5 40° c. 89.8 '93.3 88.1 83.8 85.9 88.2 80.7 76.1 80.6 83.1 82.3 80.6 Mean 89.8 96.2 87.5 87.2 87.7 89.7 80.7 78.0 81.9 83.5 81.8 81.2 25 No data is reported for the 5-hydroxymethylfurfural test as it was only applied in the last two storage periods and no differences were noted at this time. It is suggested that this test may have application in future browning reaction determination, if one were able to start at time zero. Braverman, (1963), has shown that HMF does absorb at 245 mu, and this product may also have been measured in the ultra- violet absorption spectra discussed earlier. Sensory Flavor Evaluation The sensory panel of untrained laboratory personnel was one of the variables more difficult to control in this study. As there was a lapse of 30 days between taste periods, and the panelists available in the building were not always the same, the results were somewhat variable. None the less, it can be said that the atmosphere cooked 400 C. storage sample was no longer "acceptable" at 90 days, and the retort cooked 40° C. storage sample was no longer "acceptable" at 120 days. As there was a statistically significant chemical difference (in the carbohydrate, protein and lipid tests) between the cooks in addition to the above mentioned difference in storage stability, it is a valid statement to say there was a difference in cooks, with the beans from the retort cook being slightly more stable. It must also be noted that at the 120 day storage period, the atmosphere cook sample held at 00 C. storage temperature was also showing signs of off flavors. One panelist stated that the difference between sample 1 (00 C. storage) and 26 sample 3 (40° C. storage) was only a matter of degree: both samples had an unpleasant taste. This comment reflected somewhat the other panelists opinions. For additional data, refer to Appendix C. It is recommended that in future panels only one ref- erence be used throughout the test, and have it made from a sample stored at -20°C. or lower. General Discussion It was hoped that this research project would accom- plish two objectives: (1) point out the chemical nature of product degradation, and thus suggest which systems (lipid, carbohydrate, and/or protein) contributed most to off flavors, and which controls might best be employed in maintaining quality in the powders: and (2) to find a valid quality control procedure that with further refinement could be used to monitor the quality of a given powder sample. Where possible, quantitative procedures were used in the measurements to try to show changes with time and temper- ature. However, most of the tests employed, although showing statistically significant differences with time, probably could not be used as quality control procedures to determine degrada- tion of these powders, because the changes were: (1) relatively small, as with the free fatty acid analysis, and differences between technicians and laboratories would be more significant than those between wholesome and degraded powders: (2) the powders degraded in a "bell-shaped" curve, as with the carbonyl determination, and by the time the flavor panel showed a marked 27 off flavor, the values of the variable being measured were back close to those of the new wholesome powder: or (3) the test itself had a high coefficient of variation, as with the Biuret determination. In considering the relative merits of the atmosphere and retort cooks, one must keep in mind that from a point of View of plant overhead and continuous operation, the atmosphere cook has many advantages over the retort cook. These advan- tages probably more than offset the disadvantages of being lower in carbohydrate, higher in carbonyls, and slightly less stable with time. SUMMARY AND CONCLUSIONS This research project showed there were differences in the powders caused by the two methods of cooking. The retort powders not only showed significantly different lipid and protein characteristics, but also a slight trend in increased storage stability over the atmosphere cook. Oxidation did occur and in significant amounts, as was shown by the oxygen headspace gas, fat acidity, ultra- violet absorption, carbonyl, and free amino group values. This would implicate the lipid system as one of the prime factors in storage degradation, and judging from the change shown in all these chemical tests with time, there is where one should concentrate to stabilize the powders in storage. In this vein, the use of anti-oxidants and inert gas packs should be thoroughly explored if these powders are to be marketed commercially. 28 Because the lipid fraction is probably involved in storage degradation, further studies should also be directed in this area. The exact nature of the lipid degradation should be investigated, with added emphasis placed on the phospholipids. Since the protein-peptide bonding showed changes with time (probably due to decreased solubility in polar solvents) and correlation with oxygen uptake, the use of more sensitive techniques such as different forms of chromatography and infra-red spectrophotometry should be employed to elucidate their role in flavor changes. During the storage time of 120 days reducing sugar-amino acid condensation did not seem to play a major role in off flavors. With the use of anti-oxidants and particularly inert gas packs to increase shelf life, this type of browning may become a problem. The use of sulfur dioxide in dehydrated products controls this (Eskew, 1967, and Hendel, 1955), and its use in these legume powders should be considered. Attention should also be directed towards improving and developing new methods to monitor storage degradation. The one method in this study that showed the most promise is that of head space gas analysis, both to monitor oxidation and as a quality control procedure. One type of instrument avail- able that is more sensitive than the oxygen electrode used in this experiment is a gas partitioner (Moeller, 1967). This is an inexpensive gas chromatograph which employees a single thermal conductivity detector and short columns specific for 29 the gasses in which one is interested. This instrument could be set up to monitor carbon dioxide content, as well as oxygen, and thus a relative indication on non-enzymatic browning would also be present, (Cole, 1967). The use of distilled water in soaking and cooking these powders in the future is also recommended to eliminate the possibility of pro-oxidant metals being present in one sample and not in another. This is one variable in this study that was not taken into account, and may have very marked effects on the two different cooks involved. PROPOSALS FOR FURTHER RESEARCH New research on these bean powders should include: (1) The absolute confirmation of the optimum moisture content (mono-molecular layer) at which these powders are most stable. (2) The study of the use of antioxidants and inert gas packs to increase shelf life of these powders. (3) The use of a gas partitioner in head space gas analysis to determine quantitatively oxygen and carbon dioxide contents of these powders during storage. (4) A detailed study of the lipids, including phosphatides, and their role in bean powder degradation. (5) The use of more sensitive techniques in deter- mining changes in these powders. 30 LITERATURE CITED Association of Official Agricultural Chemists, 1945. Meth- ods 2:.Analysis, 6th ed., Washington, D. C. Baker, D., M. H. Neustadt, and L. Zeleny, 1957. Application of the fat acidity test as an index of grain deterioration. Cereal Chem. 34:226-233. Bakker-Arkema, F. W., C. L. Bedford, R. J. Patterson, M. Palnitkar, B. B. McConnell, and H. K. Burr, 1968. Production of instant legume powders. Michigan State University Agricultural Experiment Station, Memo. May, 1968. Bakker-Arkema, F. W., C. L. Bedford, R. J. Patterson, B. B. McConnell, M. Palnitkar, and C. W. Hall, 1966. Drum and spray drying characteristics of precooked bean powders. Presented at 8th Dry Bean Research Conference, Bellaire, Michigan, August, 1966. Batzer, O. F., M. Sribney, D. M. Doty, and B. S. Schweigert, 1957. Production of carbonyl compounds during irradiation of meats and meat fats. J. Agr. Food Chem. 5:700—703. Boggs, M. M., H. J. Morris, and D. W. Venstrom, 1964. Stability studies with cooked legume powders 1. Flavor judging procedure. Food Tech. 18:1618-1621. Braverman, J. B. S., 1963. Introduction to the Biochemistry of Foods. Elsevier Publishing Co., N. Y. pp. 306-310. 10. ll. 12. 13. 14. 15. 31 Brunnauer, 3., P. H. Emmett, and E. Teller, 1938. Adsorp- tion of gases in multimolecular layers. J. Am. Chem. Soc. 60:309—319. Burr, H. K., M. M. Boggs, H. J. Morris, and D. W. Venstrom, 1969. Stability studies with cooked legume powders 2. Influence of various factors on flavor of lima bean powder. Food Tech. 23:842-844. Buttery, R. G. and R. Teranishi, 1963. Measurement of fat autooxidation and browning aldehydes in food vapors by direct vapor injection gas-liquid chromatography. Agric. Food Chem. 11:504-507. Cole, S. J., 1967. The Maillard reaction in food products carbon dioxide production. J. Food Sci. 32:245-250. Corliss, G. A., 1968. Phospholipid oxidation in emulsions. Ph. D. Thesis, Michigan State University. Cornwell, D. G., and L. A. Horrocks, 1964. Protein-Lipid complexes In: Proteins and their Reactions. ed. by H. W. Schultz and A. F. Anglemier. Avi Publishing Co., Inc. Westport, Conn. pp. 122-123. Coulter, S. T., and R. Jenness, 1964. Dry milk products In: Food Dehydration vol. 11 ed. by W. B. VanArsdel and M. J. Copley. Avi Publishing Co., Inc. Westport, Conn. Dubois, M., K. A. Gilles, J. K. Hamilton, P. A. Bebers, and 16. 17. 18. 19. 20. 21. 32 F. Smith, 1956. Colorimetric method for determination of sugars and related substances. Anal. Chem. 28:350-356. Eastman Chromagram System, 1966. Separation of Glycerides. Distillation Products Industries, Eastman Kodak Co., Rochester, N. Y. Analytical Procedure 1065C-212. Eskew, R. K., 1967. Potato Flakes In: Potato Proces- sing, 2nd ed. ed. by W. F. Talburt and O. Smith. Avi Publishing Co., Inc. Westport, Conn. pp. 334-335. Feustel, I. C., C. E. Hendel, and M. E. Juilly, 1964. Potatoes In: Food Dehydration vol. 11. ed. by W. B. VanArsdel and M. J. Copley. Avi Publishing Co., Inc. Westport, Conn. pp. 353-355. Forss, D. A., 1967 Origin of flavors in lids In: The Chemistry and Physiology 9; Flavors. ed. by H. W. Schultz, E. A. Day, and L. M. Libbey. Avi Publishing Co., Inc. Westport, Conn. p. 509. Furuholmen, A. M., J. D. Winefordner, F. W. Knapp, and R. A. Dennison, 1964. The quantitative analysis of glucose and fructose in potatoes. J. Agr. Food Chem. 12:109-112. Hendel, C. E., H. K. Burr, and M. M. BOggs, 1955. Non- enzymatic browning and sulfite flavor in dehydrated white potato. Effects of sulfite level and method of sulfite application. Food Tech. 9:627-629. 22. 23. 24. 25. 26. 27. 28. 33 Henic, A. S., M. F. Benca, and J. H. Mitchell, 1954. Esti- mating carbonyl compounds in rancid fats and foods. J. Am. 011 Chem. SOC. 31'88“9lo Hodge, J. E., 1953. Chemistry of browning reactions in model systems. J. Agr. Food Chem. 1:928-943. Hodge, J. E., 1967. Nonenzymatic browning reactions In: The Chemistry and Physiology 22 Flavors. ed. by H. W. Schultz, E. A. Day, and L. M. Libbey. Avi Publishing Co., Inc. Westport, Conn. pp. 465-491. Jevons, F. R., 1964. Protein-carbohydrate complexes In: Proteins and their Reactions. ed. by H. W. Schultz and A. F. Anglemier Avi Publishing Co., Inc. Westport, Conn. p. 161. Kenney, M. and R. Bassette, 1959. Detection of intermediate compounds in the early stages of browning reaction in milk products. J. Dairy Sci. 42:945-960. Koch, R. B., 1962. Dehydrated foods and model systems In: Lipids and their Oxidation. ed. by H. W. SchultZ, E. A. Day, and R. O. Sinnhuber. Avi Publishing Co., Inc. Westport, Conn. pp. 230-251. Lee, F. A. and A. C. Wagenknecht, 1951. On the development of off flavors during storage of frozen raw peas. Food Reasearch. 16:239-244. 29. 30. 31.. 32. 33. 34. 36. 34 Liener, I. E., 1962. Toxic factors in edible legumes and their elimination. 'Amer. J. Clin. Nutrition. 11:281. Little, T. A., 1966. Correlation and regression. A supplement to Experimental methods for extension workers. University of California, Agricultural Extension Service, Davis. Little, T. A., and F. J. Hills, 1966, Experimental methods for extension workers. University of California, Agri- cultural Extension Service, Davis. Lindberg, W. 0., 1962. Mechanisms. In: Lipids and their Oxidation. ed. by H. W. SchultZ, E. A. DaV, and R. O. Sinnhuber. Avi Publishing Co., Inc. Westport, Conn. pp. 31-500 Moeller, T. W., 1967. Design and development of an iso- static test method for determining gas permeability rates of packages. M. S. Thesis, Michigan State Univer- SitY' Pp. 31-u7e Muneta, p., 1964. The cooking time of dry beans after extended storage. Food tech. 18:1240-1241. Ponting, J. D., W. L. Stanley, and M. J. Copley, 1964. Fruit and vegetable juices. In: Food Dehydration, vol. II. ed. by W. B. VanArsdel and M. J. Copley. Avi Publish~ ing Co., Inc. Westport, Conn. pp. 544-562. Privett, O. 8., W. O. Lundberg, and C. Nickel, 1953. Prep- 37. 38. 39. 40. 41. 42. 43. 35 aration of peroxide concentrates from autooxidized fatty acid esters. J. Am. Oil Chem. Soc. 30:17-21. Privett, O. 3., C. Mickell. W. E. Tolberg, R. F. Paschke. D. H. Wheeler, and W. O. Lundberg, 1954. Evidence for hydroperoxide formation in the autoxidation of methyl linolenate. J. Am. Oil Chem. Soc. 31:23-27. Salwin, H., 1959. Defining minimum moisture content for dehydrated foods. Food Tech. 13:594-595. Salwin, H.. 1961. The role of moisture in deterioration reactions of dehydrated foods. In: Freeze-Drying 33 Foods ed. by F. R. Fisher. Natl. Acad. Sci.- Natl. Research Coun- cil, Washington, D. C. (Published, 1962) pp. 64-65. Siakotas, A. N., 1967 Rapid determination of lipids con- taining free amino groups with trinitrobenzene sulfonic acid reagent. Lipids. 2:87-88. Tarladgis, B. G., B. M. Watts, and M. T. Younathan, 1960. A distillation method for the quantitative determination of manonaldehyde in rancid foods. J. Am. 011 Chem. Soc. 37:44-48. Ting, S. V., 1956. Rapid colorimetric methods for simul- taneous determination of total reducing sugars and fructose in citrus juices. J. Agr. Food Chem. 4:263-266. United States Department of Agriculture, 1965. Food con- 44. 45. 46. 36 sumption of households in the United States, spring 1965. Household Food Consumption Survey 1965-1966, Report No. 1 PP. 46, 203. Wall, J. S. 1964. Cereal Proteins In: Proteins and their Reactions, ed. by H. W. Schultz and A. F. Anglemier. Avi Publishing Co., Inc. Westport, Conn. p. 336. White, A., P. Handler, and E. L. Smith, 1964. Principles g£_Biochemistry, 3rd ed. McGraw Hill Book Co., N. Y. pp. 463- 468. Wiley, R. C., A. M. Briant, I. S. Fagerson, E. F. Murphy, and J. H. Sabry, 1957. The Northeast Regional approach to collaborative panel testing. In: The Methodology g: Sen- sory Testing, presented at 17th annual meeting of Institute of Food Technologists. Garrard Press, Champaign, Ill. pp 0 “’3'“? o APPEN DI C ES APPENDIX A HUNTER COLOR DIFFERENCE METER VALUES 37 Hunter Color Difference Meter Values * Storage Storage Retort Cook Atmosphere Cook Time Temperature L Value a Value b Value L Value a Value b Value 0 Days 0° c. 78.0 -o.3 14.3 78.7 -0.5 13.6 20° c. 78.0 -0.3 14.3 78.7 -0.5 13.6 40° c. 78.0 -0.3 14.3 78.7 -0.5 13.6 30 Days 0° c. 78.2 -0.1 14.4 78.6 -0.5 13.6 20° c. 78.1 -o.2 14.5 78.8 -0.5 13.7 40° c. 78.3 -0.2 14.8 78.8 -0.5 13.7 60 Days 0° c. 78.0 -0.0 14.5 78.7 -0.6 13.8 20° c. 78.2 -0.0 14.5 78.9 -0.6 13.9 40° c. 78.2 -0.2 15.0 78.8 -0.8 14.2 90 Days 0° c. . 77.8 -o.2 14.4 78.6 -0.8 13.6 20° c. 78.1 -o.3 14.4 79.0 -0.8 13.7 40° c. 77.8 -0.3 15.1 78.8 -0.7 14.1 120 Days 0° c. 78.0 -0.5 14.2 78.8 -1.0 13.5 20° c. 78.1 -0.5 14.3 78.8 -1.1 13.7 40° c. 78.1 -0.6 15.0 78.0 -1.1 14.1 *Data represent mean values of 5 replicates APPENDIX B SATURATED, UNSATURATED, AND TOTAL CARBONYLS 38 Carbonyl Values in Micro-Moles per Gram Bean Powder Storage Retort Cook** Atmosphere Cook** Storage Tempera- Unsat- Sat- Unsat- Sat- Time* ture urated urated Total urated urated Total 0 Days 8.78 30.97 39.75 5.70 38.81 43.51 30 Days 0° 0. 7.92 33.11 41.03 12.73 31.97 44.69 20° c. 2.36 52.63 54.99 16.88 46.39 63.27 40° c. 3.50 50.17 53.67 13.54 29.35 42.89 60 Days 0° c. 9.27 32.00 41.27 14.75 55.72 70.47 20° c. 9.03 36.67 45.69 13.36 54.28 67.64 40° c. 8.12 36.89 45.01 11.28 50.73 62.00 90 Days 0° c. 5.31 32.84 38.15 11.57 39.01 50.58 20° c. 9.35 29.28 38.63 9.95 44.06 54.01 40° c. 10.74 21.04 31.78 11.78 35.88 47.66 120 Days 0° c. 12.18 30.43 42.61 13.24 36.97 50.21 20° c. 12.89 34.56 47.45 16.35 35.57 51.92 40° c. 13.86 26.05 39.91 17.24 37.80 55.04 Data Represent Mean Values of 5 Replicates *Denotes significance at 57. level **Denotes significance at 17. level APPENDIX C FLAVOR PANEL SUMMARY Flavor Panel Summary 39 ‘Time Cook 30 Days 60 Days 90 Days 120 Days Atmosphere 3.04#:0.34. 2.8 I 0.38 3.0l*#:_0.38 2.8** i 0.38 Retort 3.12 +0.27 2.94**+ 0.37 2.76* + 0.37 2.73M+ 0.42 Temperature 30 Days 60 Days 90 Days 120 Days .Atmos- Retort Atmos- Retort Atmos- Retort Atmos- Retort Rhere Bhere Ebere ‘Bhere 0° c. 3.45* 3.25 3.12 3.54** 3.16** 3.62** 3.16** 3.62** 20° c. 2.95 3.16 2.62 2.66* 3.08 2.62* 3.008 2.70 40° c. 2.704 2.83 2.63 2.62** 2.12** 2.50* 2.25** 1.87** * Denotes significance at 5% level **Denotes significance at 1% level MICHIGAN STATE UNIV RSITY Ll E BRARIES 9 3 1293 03046 856