Tài liệu Nghiên cứu sữa part 2

Meissner's corpuscle Tactile discs Free nerve endings Sebaceous gland Smooth Dermis Epidermis muscle Hair End bulbs of Krause Nerve ending Subcutaneous Padnian fat corpuscle wound hair Duct of Ruffini ItNMt gland ending Figure 3.8 Composite diagram of the skin in cross-section. Tactile sensations are transmitted from the variety of nerve endings, for example, the free nerve endings and the tactile discs in the epidermis, and the meissner corpuscles, end bulbs of Krause, Ruffini endings, and pacinian corpuscles in the dermis. (Reproduced with permission from ref. 35.) Swiss style strawberry yogurt were added. With the exception of a few war years, the contest has been held annually. Fifty-nine schools have fielded teams with as many as 33 participating in 1956. 5 ' 37 Several regional collegiate contests are also held each year. At the high school level, the Future Farmers of America conducts an annual state and national dairy foods evaluation contest. These have served to give thousand of students training in the recognition of dairy product defects, their causes, and control. Many other food industries have developed their *'expert" tasters resources. These experts obtained experience through the years and were charged wih the responsibility of determining the material blend or judging the quality of raw materials. They also judge the quality of finished product and identify sources of problems and suggestions for correction when the products are less than perfect. These experts include the perfumers, flavorists, brew masters, wine makers, and coffee and tea tasters. In most of these industries, such as the dairy industry, scorecards and point systems have been developed to help set standards.38 With the growth of the food industry and the expansion of product lines within companies, it has become almost impossible to have dependable expert judges of all products. It has been necessary to develop sensory evaluation systems that are more universally applicable. Sensory evaluation of foods in general with methodology appropriate for either consensus or statistically sound evaluation of foods began to develop in the 1940s and 1950s at the U.S. Army Quartermaster Food and Container Institute in Chicago.39'40 Development began also in the private sector. The Arthur D. Little Company pioneered descriptive analysis by developing a Flavor Profile Method that uses a consensus of a small group of people who are trained to the product in a way that is universally applicable. The single expert was replaced with five or six trained people.41 The University of California at Davis began to offer courses on sensory evaluation in the 1950s. The literature at that time reflects significant development in the application of sensory evaluation. Discrimination tests were developed by Boggs and Hansen,42 Girardot et al.,43 and Peryam et al.39 Ranking and hedonic scales began to be used for consumer acceptance information. Committee E-18 of the American Society for Testing Materials, the Food and Agriculture Section of the American Chemical Society, the European Chemoreception Organization, and the Sensory Evaluation Division of the Institute of Food Technologists got involved by organizing activities focusing attention on sensory evaluation and measurement of flavor and publishing information assisting the food industry in application of the new techniques.40 These methods are all applicable to dairy product evaluation. 3.2.2 Affective Testing Affective testing is acceptance testing. Its objective is to determine the degree of consumer acceptance or preference for a product. Usually it is determined relative to a product such as an existing product, or an acceptable successful product. The ideality of certain easily understood attributes can be judged by consumers using their concept of ideal as the standard. Hedonic scales are used to rate the degree of liking of products. An example of a nine-point hedonic scale is shown in Figure 3.9. There are a wide variety of hedonic or liking scales that can be and have been used. Recommended scales are balanced with an odd number of choices, with the middle choice being neutral "Neither like nor dislike." Choices above neutral are positive, with the top being "Like extremely" and the choices below neutral being negative and balanced with those above and the bottom being "Dislike extremely." The data can be treated parametrically, yielding means and standard deviations. Liking of products can be compared using the t test or analysis of variance (ANOVA). Parametric treatment assumes that data are distributed normally and that intervals on the scale are equal. There has been considerable discussion about the validity of these assumptions but the practical value of this approach continues to be demonstrated. The data can be converted to preference or ranking and analyzed binomially.40'44'45 Another affective tool is preference testing. Panelists have the opportunity in preference testing to tell which of two samples they prefer (paired comparison) or Please check a box indicating your feeling about this product. Like extremely Like very much Like moderately Like slightly Neither like nor dislike Dislike slightly Dislike moderately Dislike very much Dislike extremely Figure 3.9 An example of the nine-point hedonic scale. The subjects indicate to what extent they like or dislike the sample by checking a box by the most correct statement. Please check a box indicating your feeling about the moistness/dryness of this product Much too moist Slightly too moist Just about right Slightly too dry Much too dry Figure 3.10 An example of a Just-about-right scale. The purpose of the judgment is to establish how close to ideal a product is in an easily understood attribute. The subject checks the box by the statement that best describes his or her feelings about the correctness of the level of that attribute. to rank more than two samples in order of preference. It is important that each sample is tasted first and last its share of the time to avoid order bias. Analysis of the paired comparison test utilizes binomial statistics. Tables are available giving the number of subjects that must prefer one sample given a certain number of participants for the preference to be significant.46 When ranking is used, tables and formulas are available showing the rank sum difference required for significantly different ranking given the number of samples compared and number of panelists used. 47 An effective tool to determine the ideality of easily understood attributes is the Just-about-right scale. This is the three- or five-point scale with "Just about right" being the middle response with balanced descriptors of the attribute extremes going up and down from ideal (Fig. 3.10). Stone suggests two methods of analyzing the data to determine if each product deviates significantly from ideal and one method to determine if the samples deviate from one another in ideality.40 One involves using the binomial table of Roessler et al. (p = 0.5, two-tailed) to determine if the number of judgments on one side of ideal is more than can be explained by chance.46 The number of nonideal judgments is n and the number on one side of ideal is found in the column under the appropriate confidence level. The appropriate type of panelist for all affective tests is a "naive" consumer, one who has no knowledge of the objective of the comparison or the technology involved in making the products. The subjects may be screened to be representative of the demographics of a certain target consumer group. Trained panelists who are used in descriptive or discrimination tests should not be used because of their analytical approach which may bias affective judgments.40 3.2.3 Discrimination Testing Discrimination testing is a very useful sensory evaluation tool that enables one to determine if a perceived difference exists between two products. Often it is preliminary to other types of testing. If no perceived difference exists, it is not necessary to determine which one is preferred or what the difference in the descriptive characteristics are.40 If a development objective is to have no perceived difference, this test can establish that the objective has been met and subsequent sensory testing may not be necessary. There are several methods that may be used to establish whether there is a perceived difference. Methods include paired-comparison, duo-trio, and triangle tests. The paired comparison test is a two-sample test with the task being to determine whether the products are the same or different, or it may be to choose which of the two samples has more of a particular attribute. When the subject is asked if the products are the same or different, it is important that half the panelists receive samples that are the same and half receive samples that are different. In interpreting the data, the number of correct choices are compared with the number of correct selections that can be explained by chance. When the assignment is to indicate which sample has more or less of a certain attribute, it is assumed that the subject recognizes that attribute in the product. It is important that the attributes be simple and easily recognizable. If the number of correct selections if greater than can be explained by chance, one can conclude that the samples are different. Interpretation involves binomial statistics. A table and formula for the significant number of correct judgments is published by Roessler et al.46 The correct table and formula would be those where the probability of being right by chance in one selection is one in two (p = 0.5). It is a one-tailed test. The tail of interest is being correct more frequently than can be explained by chance. The other tail not of interest is being wrong more frequently than can be explained by chance. Protection against a type I error (finding difference when none exists) is selected by selecting the column with the appropriate a. An a of 0.05 would allow for a 5% chance of a type I error.48 The duo-trio test was developed by Peryam and Swartz as a way to minimize the number of comparisons that have to be made.39 The subject is given a reference sample and two coded samples. One of the coded samples is the same as the reference sample. The subject is asked to indicate which sample is the same as (or different from) the reference. In variations of the test, the reference sample may be removed after it is tasted to force the use of memory for comparison. Reliance on memory decreases the sensitivity of the test. The same sample may be used as the reference through the entire test, or each sample may take its turn as the reference. It is important that the order of tasting the two samples be rotated so that each sample is tasted immediately after the reference with equal frequency. The data are evaluated using the same formula and tables as for paired comparisons.46 The probability of being correct on one decision is one in two (p = 0.5) and interest is in one tail (being right more frequently than can be explained by chance). The most frequently used discrimination test is the triangle test. It was initially developed by a beer company.49 In this test, the panelist is presented three coded samples. Two are the same and one is different. The panelist evaluates all three and determines which one is different or which two are most alike. This test requires more tasting than the others. Three pairs are compared in making the judgment. Again binomial statistics are used to evaluate the results. The probability of being right by chance (p) in one selection is one in three and it is a one-tailed test (the probability of being wrong more frequently than is explained by chance is the tail that is not of interest).40 The table and formula provided by Roessler et al. are used to determine when the frequency of correct selection exceeds chance.46 Subjects for discrimination tests should like the product, be familiar with the test procedure, have frequent practice with the test, have a record of exceeding chance in choosing correctly in previous tests, and have no specific knowledge about the samples.40 The number of panelists used should be no more than 40 and may be as few as 12 to 15. Too many panelists will result in significant differences when the differences are very subtle and of no practical importance. Too few will allow for a large type II error (finding no difference when difference exists).30'48 It is important to guard against unintended differences. For example, it is easy to have slight temperature, serving amount, piece shape or size, or color differences that are not intended. Panelists are playing a game and will look for any clues that will reveal the different sample. If a conclusion is reached, due to inadvertent hints that samples are different when they are not, the results can be misleading and expensive. Further development or costly consumer or descriptive testing may be mandated. 3.2.4 Descriptive Analysis Descriptive analysis is the process of developing a total sensory description of a product. In its complete form it involves identifying each flavor, aroma, and textural quality detectable in the product and quantifying each. The time sequence of the detection of the qualities can also be included in the profile. Affective judgments as to the desirability of the sensory qualities are generally not a part of descriptive analysis. It is important that the panel members are highly trained to recognize all of the qualities of the product and to use a standardized terminology to describe them. Developing and proving a descriptive panel requires skill on the part of the leaders, and dedication, time, patience, and attention to detail on the part of panel leaders and panelists. 30 ' 40 Several methods of descriptive analysis have been developed. Three that represent the development of descriptive analysis and slightly different philosophies are the Flavor Profile, Texture Profile, and Quantitative Descriptive Analysis (QDA). The Flavor Profile method was developed by Arthur D. Little, Inc. in the late 1940s. A small panel of four to six trained judges analyze a product's perceived aroma and flavor qualities, and their order of detection, intensity, and aftertaste. They also assess the degree to which various flavor or aroma characteristics fit together and their appropriateness in the product and call this characteristic amplitude.41'50 Prospective panelists are screened for their ability to detect and discriminate tastes and odors. Their interest and availability and ability to work with a group are assessed in a personal interview. Selected panelists are trained with product examples that represent the extremes of the different qualities that may be encountered. Product is made with a variety of ingredients and processes to produce a wide variety of product. In the actual evaluation session, trained panelists first evaluate a product individually while seated together around a table. The results are reported to the panel leader who leads a discussion that results in a consensus profile. More than one sample can be profiled in a session but they are done one at a time without tasting back and forth. Once a panel is trained, profiles can be obtained easily. 10 ' 40 General Foods developed the Texture Profile method to do for texture analysis what the Flavor Profile method had done for flavor and aroma. 51 " 53 It was different from flavor profiling in that the terminology for different texture qualities was standardized (Table 3.2). The anchors used to standardize the scales were also predefined. Odd numbered categorical scales for each quality were developed. Later quality descriptors were added for semisolid foods, beverages, 54 ' 55 and skin-feel products.56 Prospective panelists are screened based on interest, availability, and attitude. They are further selected on the basis of ability to discriminate known textural differences in the product to be tested. They are introduced to the principles involved in the product to be tested. An evaluation of a product after the panel is trained involves independent evaluation by each panelist using one of a number of possible scales, then the generation of a panel verdict. The verdict may be obtained by discussion and group consensus similar to the method for obtaining a flavor profile or by statistical analysis of the data. Quantitative Descriptive Analysis was developed to overcome weaknesses in the descriptive test previously described. It was designed to be responsive to flavor, aroma, and texture simultaneously, to be applicable to a broad range of products, to be quantitative in evaluation of panelists' qualifications and in development of profiles, to use a small number of panelists, and to have flexible panel-generated terminology. Subjects are qualified before participation. They must be available and be users of the product class. They must demonstrate ability to perceive differences within the class of products and to articulate those differences. The terms used to describe qualities may be available from previous work. If so, the panel learns and experiences the definitions of all the qualities. If not, the terms describing the qual- Table 3,2 RELATIONSHIP BETWEEN TEXTURAL PARAMETERS AND POPULAR NOMENCLATURE Mechanical Characteristics Primary Parameters Hardness Cohesiveness Popular Terms Secondary Parameters Brittleness Chewiness Gumminess Viscosity Elasticity Adhesiveness Soft, firm, hard Crumbly, crunchy, brittle Tender, chewy, tough Short, mealy, pasty, gummy Thin, viscous Plastic, elastic Sticky, tacky, gooey Geometrical characteristics Class Examples Particle size and shape Particle shape and orientation Gritty, grainy, coarse, etc. Fibrous, cellular, crystalline, etc. Other Characteristics Primary Parameters Moisture content Fat content Popular Terms Secondary Parameters Oiliness Greasiness Dry, moist, wet, watery Oily Greasy Reproduced with permission from ref. 52. P l e a s e m a r k this line in a position that indicates how w e a k / f i r m you feel this yogurt body to b e . Extremely weak Extremely firm Figure 3.11 An example of a horizontal line scale used by descriptive panelists to indicate the strength of a particular flavor or aroma quality. The subjects marks the position of the line that describes the intensity of the quality. ities are selected and defined by the panelist as they train. Reference materials that are examples of the qualities are used to aid in definition of qualities. When evaluating actual product, if new qualities are found, the panel reconvenes to define and train on that quality. Scales used are horizontal lines of a consistent length with word descriptors at or near the ends (Fig. 3.11). Intensity always increases from left to right and the subject marks the line at a position that is appropriate for the intensity of the quality. Evaluation during training and on actual product is done individually Aftertaste Bitterness Aroma Malt Flavor Sweet Crunch (final) Sour Crunch (initial) Figure 3.12 Visual display of the sensory characteristics based on the results of a Quantitiative Descriptive Analysis (QDA) test. For each characteristic, the relative intensity increases as it goes further from the center. (Reproduced with permission from ref. 40.) and usually in isolated sensory booths to ensure independent analysis. Replicate samples are included so that ANOVA can be applied to evaluate the panelists' consistency as well as to statistically compare the intensity of qualities of the different samples. The panelists who are best able to replicate themselves on all the qualities and who agree best with the rest of the panel on each of the qualities are best qualified to evaluate product. Usually between 8 and 12 qualified subjects constitute a panel. The product QDA profile is a listing of the qualities and the means for each of those qualities. Significance of difference between samples in each quality is obtained by ANOVA.40 Multiple-range tests are applied to establish the significance of differences between multiple samples. Profiles of individual samples can be shown in a number of formats. A "spider web" format is shown in Figure 3.12. Each quality is depicted as a "spoke" of a wheel with its length being indicative of the intensity of the quality. With the ends of the "spokes" connected, a shape is formed that is distinct. A change of intensity in one attribute produces a readily distinguishable difference in shape. 3.3 Application of Sensory Analysis to Dairy Products The system for evaluating dairy products for defects was developed long before the generally applicable tools of affective, difference, and descriptive analysis. These Next Page newer generally applicable tools are as useful for dairy products as they are for other foods and are essential when sensory information needs to be quantified for research purposes. Any treatment of sensory analysis of dairy products without their mention would be incomplete. The remainder of this chapter, however, will focus on evaluation of dairy products for defects or judging of dairy products. This ability, although not designed for statistical analysis or research, is still very useful to dairy product manufacturers, enabling them to recognize defects, identify causes and take corrective action. 3.3.1 The Philosophy of Judging of Dairy Products Judging of dairy products is related to descriptive analysis. It is similar in that flavor (including aroma), texture, and appearance can all be evaluated. It is similar too in that the names of the qualities and their definitions are standardized. The quality terms and definitions have evolved over the years with USDA and industry "experts" involved and a committee of collegiate coaches, who serve as the American Dairy Science Committee on Dairy Product Evaluation, periodically modifying the terms and definitions. It is different from descriptive analysis in that normal ideal base qualities of the products are not identified and only the defects are noted. The judges score the products on flavor, texture, and appearance. Score ranges are established for each defect. Defects that are indicative of serious problems have lower score ranges than less serious defects. Higher scores in that range are given if the defect is slight and scores at the lower end of the range are given when defects are pronounced. In the event of multiple defects, the score is based on the defect that would result in the lowest score. In that way, scoring takes into account the magnitude and seriousness of the defects as determined by these "experts." No attempt has been made to tie the scores to consumer acceptance of the products. 3.4 Descriptive Sensory Defects of Dairy Products 3.4.1 Fluid Milk and Cream 3.4.1.1 Introduction Fluid milk is the material from which all other dairy products are made. Defects in milk will cany over into those products so it is important that these defects be recognized first. Coaches of collegiate judging teams spend a generous amount of time on fluid milk because the defects of milk are closely related to the resulting defects in products, and because "doctoring" milk to simulate the defects is relatively easy. 5 A wide variety of fluid milk and cream products are available. A listing of products is shown in Table 3.3. Complete evaluation of fluid milk can include examination and scoring of a sediment disk, evaluation of the package, storage temperature, and bacteria count.5 Table 3.4 shows flavor defects that can be found in milk and the range of scores that can be assigned. A score card that includes all these important defect descriptors is shown in Figure 3.13. It is based on a possible 25 Previous Page newer generally applicable tools are as useful for dairy products as they are for other foods and are essential when sensory information needs to be quantified for research purposes. Any treatment of sensory analysis of dairy products without their mention would be incomplete. The remainder of this chapter, however, will focus on evaluation of dairy products for defects or judging of dairy products. This ability, although not designed for statistical analysis or research, is still very useful to dairy product manufacturers, enabling them to recognize defects, identify causes and take corrective action. 3.3.1 The Philosophy of Judging of Dairy Products Judging of dairy products is related to descriptive analysis. It is similar in that flavor (including aroma), texture, and appearance can all be evaluated. It is similar too in that the names of the qualities and their definitions are standardized. The quality terms and definitions have evolved over the years with USDA and industry "experts" involved and a committee of collegiate coaches, who serve as the American Dairy Science Committee on Dairy Product Evaluation, periodically modifying the terms and definitions. It is different from descriptive analysis in that normal ideal base qualities of the products are not identified and only the defects are noted. The judges score the products on flavor, texture, and appearance. Score ranges are established for each defect. Defects that are indicative of serious problems have lower score ranges than less serious defects. Higher scores in that range are given if the defect is slight and scores at the lower end of the range are given when defects are pronounced. In the event of multiple defects, the score is based on the defect that would result in the lowest score. In that way, scoring takes into account the magnitude and seriousness of the defects as determined by these "experts." No attempt has been made to tie the scores to consumer acceptance of the products. 3.4 Descriptive Sensory Defects of Dairy Products 3.4.1 Fluid Milk and Cream 3.4.1.1 Introduction Fluid milk is the material from which all other dairy products are made. Defects in milk will cany over into those products so it is important that these defects be recognized first. Coaches of collegiate judging teams spend a generous amount of time on fluid milk because the defects of milk are closely related to the resulting defects in products, and because "doctoring" milk to simulate the defects is relatively easy. 5 A wide variety of fluid milk and cream products are available. A listing of products is shown in Table 3.3. Complete evaluation of fluid milk can include examination and scoring of a sediment disk, evaluation of the package, storage temperature, and bacteria count.5 Table 3.4 shows flavor defects that can be found in milk and the range of scores that can be assigned. A score card that includes all these important defect descriptors is shown in Figure 3.13. It is based on a possible 25 Table 3.3 A USTING OF FRESH MILK AND CREAM PRODUCTS WITH FAT CONTENT IN PARENTHESES57 Half and half (10.5-18%) Light cream (18-30%) Light whipping cream (30-36%) Heavy cream (s*36%) Whole milk (2*3.25%) Skim milk (<0.5%) l%Milk(l%) 2% Milk (2%) Table 3.4 THE ADSA SCORING GUIDE FOR OFF-FLAVORS ON MILK AND CREAM Intensity of Defect Flavor Criticisms3 Acid Bitter Cooked Feed Fermented/fruity Flat Foreign Garlic/onion Lacks freshness Light induced (oxidized) Malty Metallic (oxidized) Rancid Salty Unclean Slight Definite Pronounced 3 5 8 6 5 9 5 5 8 6 5 5 4 8 3 1 3 8 4 3 8 3 3 7 4 3 3 1 6 1 0b 1 6 1 1 7 1 1 6 1 1 1 0 4 0 Source: American Dairy Science Association, 1990. a "No criticisms'' is assigned a score of 10. Normal range is 1 -10 for salable product. b An assigned score of 0 (zero) is indicative of unsalable product. points with 10 possible on flavor, three on sediment, five on package, five on bacteria count, and two on temperature. The electronic score card now used in collegiate competition in which only flavor is judged is shown in Figure 3.14. The flavor of milk is usually judged after sediment, closure, and container are judged. This treatment will cover only flavor. For information on how the other factors are judged see Bodyfelt.5 To best judge flavor, the milk or cream should be tempered to 12.8 to 18°C. The judge should swirl the bottle and then smell the milk or cream. Swirling serves to mix the sample and to spread a fine film on the inside of the container which gives maximum opportunity for volatiles to fill the headspace. A small amount of sample should be poured into a clean odorless container. Glass is preferred but plastic or paper is acceptable. The judge should then take a sample into his or her mouth, and move it around in the mouth making sure to coat all the surfaces of the SCORE CARD FOR MILK QUALITY Product: Date: 1 Flavor 10 No criticism 10 Unsalable 0 Normal range 1-10 Sediment 3 Package 5 No criticism 5 Unsalable 0 Normal range 1-5 Bacteria 2 3 SAMPLE NO. 4 6 5 7 8 Criticism Score Acid Astringent Barny Bitter Cooked Cowy Feed Fermented/fruity Flat Foreign Garlic/onion Lacks freshness Malty Oxidized light induced Oxidized metal induced Rancid Salty Unclean Score Score Container bulging/distorted Dented/defective Dirty inside Dirty outside Leaky Not full Closure defective Coating flaky/cracked Heat seal defective Illegiblejjrinting Labeling/code incorrect Lip chipped Cover not waterproof Unprotected 5 Score Standard plate count Coliform count Keeping quality Temperature 2 Temperature (0F or 0C) Total score of each sample Desired % Fat content (%) Desired % Solids not fat (%) Under/over filled Titratable acidity Functional and other tests performe:d on samples Score Score Signatures of evaluators Figure 3.13 A modified and expanded version of the ADSA milk score card. (Reproduced from ref. 5, with permission of the ADSA, Champaign, IL.) MARKING INSTRUCTIONS M I PROPER MARKS PROPER MARK ERASE CHANGES CLEANLY AND COMPLETELY OO NOT MAKE ANY STRAY MARKS MILK NCS Tm i e-Opcti* MP30-73629-321 A2400 SAMPLE NUMBER WTTER NO CRT IC IS I M FE£O 10 FLAT NORMAL RANGE 1-10 GARUC/0NI0N MALTY waomo - Mm*, Mouotft SAtTY BODY AND TEXTURE NO CRITICISM 5 NORMAL RANGE 1-5 APPEARANCE AND COLOR NO CRITICISM 5 NORMAL RANGE 1-5 Figure 3.14 Collegiate contest milk score card. (Reproduced from ref. 5, with permission of the ADSA, Champaign, IL.) mouth from the front to deep in the back down to the throat, noting any off-flavors. While the sample is in the mouth, airis moved up through the nose to enhance odor detection. The sample should then be expectorated and a few moments allowed to observe aftertaste. Aftertaste and aroma sensation are enhanced by exhaling slowly through the nose. Swallowing sample is not advised. According to Bodyfelt,5 the flavor of whole milk should be pleasant and sweet and with neither a foretaste nor an aftertaste other than that imparted by the natural richness. A listing of flavor criticisms with a scoring guide is shown in Table 3.4. A list of these defects, and their verbal descriptions, causes, and methods of preparing training samples follows. 3.4.1.2 Flavor Defects Acid or Sour Milk Description. Acid or sour is detected by both the senses of taste and smell. The tip of the tongue is sensitive to the "peeling" or "tingling" sensation. A general feeling of "cleanliness" and enhanced ability to taste is part of the sensation. Other flavors such as diacetyl may accompany acid as byproducts of fermentation.5 Cause. Acid or sour milk is a result of bacterial action on lactose converting it to lactic acid. It can be produced by culture organisms such as Lactococcus lactis ssp. lactis, or Lactococcus lactis ssp. cremoris or by any other lactic acid fermenting organism that purposely or accidentally is present in milk and is allowed to grow. Training Sample Preparation. Small amounts of lactic acid can be dissolved in milk until the desired intensity of acid is obtained. Addition of 25 ± 5 to 10 ml of fresh cultured buttermilk can be added to 575 ml of fresh milk. It should be prepared 1 or 2 days before tasting and held refrigerated until use.5 Usually a diacetyl flavor accompanies the acid flavor. Astringent Description. This sensory defect is actually a tactile sensation. Other descriptive words used are mouth coating, dry, puckery, chalky, and powdery. It is classified here with flavor because it is sensed when the product is taken into the mouth. It is not a common defect in beverage milk. After expectoration, the lining of the mouth may feel shriveled or puckered. Cause. Not all the causes are known but it is usually associated with high heat treatment of milk that has caused some aggregation of milk proteins. A specific particle size of milk proteins or other milk constituents is thought to be responsible for the sensation. Training. Green persimmon or alum are extreme examples of astringency. They may be used to demonstrate the sensation. Barny Description. The flavors ' 'cowy," ' 'barny," and * 'unclean'' seem to be quite alike but differ in intensity and cause. The descriptive term "barny" is quite accurate, referring to the typical smell of a poorly maintained bam atmosphere. It is noticed immediately after the milk is expectorated.5 Cause. The smells of the barn are thought to be transmitted to the milk through the cow's respiratory system when cows are stabled and milked in a foul smelling barn environment. Training. Trainees could be taken to some milking operations and the atmospheric aroma noted. Milk could be collected from cows that are kept in this type of closed environment, lab pasteurized, and used soon after as training samples. Bitter Description. Bitter is a taste sensation with no associated aroma. It is detected at the base of the tongue. The reaction time is fairly slow so it is most strongly sensed after the milk is expectorated. The intensity builds and it is hard to rinse away and refresh the tongue. It seems to be a component of "rancid" and "soapy" flavors.5 Cause. It is generally acknowledged that some protein fragments taste bitter. These fragments can be produced by enzymatic breakdown of milk proteins. Enzyme sources in milk are likely psychrotrophic microorganisms that have grown in the cool milk. Milk that is stored at temperatures at or slightly above 4°C for several days will become bitter if these contaminating organisms are present. Under those conditions they will grow to large populations and release proteases. Certain weeds consumed by the cow will also impart bitterness to the milk. Conditions that produce rancidity may be to blame for bitterness that is a component of rancidity. Preparation of Training Samples. Traces of quinine dihydrochloride or quinine sulfate added to milk will give a clean bitter flavor. A 1 % stock milk or water solution can be made and added at the rate of 1 to 2 ml per 600 ml of milk.5 Cooked Description. Four kinds of heat-induced flavors have been recognized: sulfurous, rich, caramelized, and scorched. All are easily identified.58 They are detected immediately as the sample is placed in the mouth and are usually considered to be pleasant. The sulfurous and rich descriptors are common in milk. The detection of a cooked egg white smell is characteristic of this defect. Cause. The mild sulfurous flavor develops when milk reaches 76°C to 78°C.59 This is slightly above HTST pasteurization temperatures. Its development is associated with the breaking of disulfide bonds and the development of conditions that discourage oxidation. The more severe flavors of scorched and caramelized develop at higher temperatures and by a different mechanism and are not normal in beverage milk. The heated flavor is what remains after cooked milk is stored cold for a period of time. Caramelized flavor frequently intensifies and becomes more objectionable with age.5 Preparation of Training Samples. Fresh pasteurized/homogenized milk is heated to 800C and held for 1 min and then cooled.5 This can be done on a plate pasteurizer, in a water bath, or in a pan on a stove top with continual stirring. Cowy Description. Usually a "cowy" flavor suggests a cows-breath-like odor and a chemical aftertaste. It seems to be associated with the presence of acetone bodies in milk.5 Cause. Cows that have acetonemia or ketosis will give milk with this off-flavor defect. Feed Description. A "feed" flavor is aromatic and sometimes pleasant. After the milk is expectorated a mild aftertaste of "cleanliness" can be present that disappears rather quickly, leaving the mouth free of off flavors. Cowy, barny, and unclean flavors by contrast persist with an accompanying unpleasant or "dirty" aftertaste. Feed flavor varies with the type of feed consumed. The odor is characteristic of the feed.5 Cause. High-volume roughage feeds consumed within 3 h of milking impart flavors and aromas to the milk.5 Silage, some hays, and brewery waste are particularly notable for this. A change of feed from dry hay to fresh green pasture often initiates a strong feed flavor in the milk. If 3 h is allowed to pass between consumption and milking, almost all feed flavors are absent from the milk.5 Preparation of Training Samples. An alfalfa flavor can be simulated by adding and placing 2 to 3 g of alfalfa hay in 100 ml of fresh pasteurized and homogenized milk and holding for 20 min. The milk is then strained through a cheesecloth or paper towel and used as a stock solution. To 575 ml of fresh pasteurized and homogenized milk, add 20 to 35 ml of this stock milk solution. Grass or corn silage can be used to prepare feed flavored milks in the same manner.5 Fermented/Fruity Description. This defect is detected by its odor which resembles the odor of sauerkraut, vinegar, pineapple, or apple. There will also be an unpleasant flavor that will linger long after the sample has been expectorated. Cause. This flavor is often found in bulk raw milk after lengthy storage. Certain microorganisms such as Pseudomonas fragi and other Pseudomonas species are among those that produce aromatic fermentation products.60 Preparation of Training Samples. Bodyfelt suggests the preparation of a stock solution of 1% ethyl hexanoate. About 1.0 to 1.25 ml of this solution is added to 600 ml of fresh pasteurized and homogenized milk.5 Flat Description. Flat milk gives a watery sensation or a lack of flavor richness. No aroma is associated with flat flavor but the lack of sweet and salty notes becomes apparent immediately as the milk enters the mouth and the subtle thinner mouth feel may also be notable.5 Cause. Flat flavor is generally caused by dilution with water. It can happen at the farm or in the plant by allowing too much rinse water to pass into the milk before it is diverted. Purposeful dilution with water is also possible. Preparation of Training Samples. To prepare slightly flat samples add 75 to 100 ml of good quality tap water to 500 ml of fresh pasteurized and homogenized milk. For definite flat use 110 to 120 ml of water to 485 ml of milk.5 Foreign Description. The term *'foreign" is used to describe a number of flavors that are imparted by addition of detergents, disinfectants, and sanitizers to milk. The flavor is characteristic of the chemical that has been added. The flavors are atypical of milk and do not develop in milk. In some cases the chemical may be detected by smell but in others it may not be detected until it is tasted. Cause. Adding milk to a vat or running milk through piping that has been washed or sanitized but not rinsed can cause a foreign flavor especially if allowed to comingle with a considerable amount of liquid containing the chemical. Other possible causes include treating the udder with ointments or medication, contamination with insecticides, and drenching the cow with chemical treatments. Preparation of Training Samples. Bodyfelt et al. suggests that a foreign flavor may be created by adding 3 to 4 ml of twofold vanilla extract to 600 ml of milk and that a foreign flavor caused by sanitizer can be produced by adding 1.0 ml of a 5% sodium hyperchloride solution to 600 ml of good quality milk.5 Samples can be made by adding traces of other nontoxic chemical cleaners and sanitizers to milk at low concentrations. Garlic/Onion (Weedy) Description. These flavors are identified by their characteristic pungent flavor and aroma and persistent after taste. Cause. Milk is tainted with these flavors during the warm months when cows are feeding in pastures that are infested with onion, garlic, or other weeds that impart these flavors to the milk. They are especially strong when the cows consume these plants shortly before they are milked. Preparation of Training Samples. To produce a definite garlic/onion intensity, add 0.15 g of garlic or onion salt or two drops of extract to 600 ml of good quality pasteurized and homogenized milk. Vary the amounts to get the desired flavor strength. Lacks Freshness Description. This flavor lacks descriptive characteristics. It suggests the loss of fine taste qualities typically noted in good milk. It is not as pleasantly sweet and refreshing or as free of an aftertaste as is typically desired in milk. Frequently lowfat milks when compared with whole milk will exhibit this characteristic. Cause. The ' 'lacks freshness" characteristic is often considered to be early stages of the development of oxidized or rancid flavor or it could be the beginning of degradation by psychrotrophic bacteria. Preparation of Training Samples. This characteristic is often present in milk that is approaching its pull date about a week and a half to two weeks after processing. It can also be simulated by addition of 10 to 15 g nonfat dry milk powder to 600 ml of pasteurized and homogenized milk.5 Malty Description. As is suggested by the descriptive term, this flavor is suggestive of malt. Malt, which is grain (barley) softened by steeping and allowed to germinate, has this characteristic flavor. This flavor can be detected by smelling or tasting the milk and is often accompanied by or is the forerunner of an acid taste.5 Cause. This flavor in milk is usually caused by the growth of Streptococcus lactis ssp. lactis var. maltigenes bacteria. They grow well when the temperature is allowed to rise above 18.2°C for 2 to 3 h.60 Preparation of Training Samples. This flavor can be easily transferred from malted cereals to milk. A stock solution is made by soaking 15 g of Grape Nuts in 100 g of milk for 30 min. The milk is filtered through cheesecloth or a napkin. Thirteen milliliters of the stock solution is added to 590 ml of pasteurized and homogenized milk to give a malty flavored milk of definite intensity. Oxidized (Metal-Induced) Description. This flavor is a result of lipid oxidation that is induced by catalytic action of certain metals. Other synonymous terms are metallic, oily, cappy, cardboardy, stale, tallowy, painty, and fishy. It is characterized by an immediate taste reaction on placing the sample in the mouth and a moderate aftertaste. A puckery mouth feel characterizes high-intensity oxidized flavors. It is similar to the flavor of metal foil, a rusty nail, or an old penny.5 Cause. The presence of this flavor usually means that some corrodible metal has come in contact with the milk. It usually can be traced to a fitting or some piping that is made of "white" metal. For years, dairy plants and equipment have been made entirely of stainless steel to avoid the development of this defect. Oxidation of the phospholipids that were originally in the fat globule membrane is blamed for the majority of the flavor. Two oxidative products, 2-octenal and 2-nonenal, have this characteristic flavor at <1 ppm.61 Preparation of Training Samples. The flavor can be generated by soaking clean pennies in milk until the flavor intensity reaches the desired level. Another method is to prepare a 1% stock solution of CuSO4 and add the following amounts to 600 ml of milk: 0.75 ml for slight, 1.2 ml for definite, and 1.8 ml for pronounced. These samples are held refrigerated for 1 to 2 days before use.5 Oxidized (light-Induced) Description. Synonymous descriptive terms that have been used for this flavor are burnt, burnt protein, burnt feathers, cabbagey, and medicinal. Some synonymous terms designating cause are light-activated and sunlight flavor. Cause. Two reactions are involved in the development of this flavor which develops when milk is exposed to sunlight or fluorescent lights. One is produced by lipid oxidation as described for metallic oxidized flavor, and the other by amino acid degradation involving riboflavin. It is proposed that methionine is degraded to 3-methylthiopropanal (methional) by a Strecker degradationlike reaction yielding ammonia and carbon dioxide.36'62 Methional has an odor similar to that of lightexposed milk. Without riboflavin methional does not develop.36 Preparation of Training Samples. Milk with the light-induced oxidized flavor can be prepared by exposing milk in clear or translucent containers to bright direct sunlight for 8 to 15 min. The shorter times will produce slight levels of the defect and the longer time will give definite and pronounced levels.5 Similarly the flavor can be produced by exposing milk to bright fluorescent light for 2 to 8 h. Overnight exposure next to a 40-watt fluorescent light will produce pronounced flavor. Less intense samples can be prepared by diluting strongly flavored samples. Rancid Description. There are several characteristics of rancid off-flavor. There is a characteristic odor derived from volatile fatty acids that have been hydrolyzed from the fat. Immediately after putting the sample in the mouth, the objectionable flavor may not be apparent but as the sample reaches the back of the mouth, soapy, bitter, and possibly unclean flavors are perceived. The soapy and bitter notes reside long after the sample is expectorated. A high percentage of prospective judges do not detect or have a high threshold for the soapy and bitter notes.5 Cause. Rancid flavor is usually caused by disrupting the milk fat globule while active lipase is present. The lipase enzyme, which catalyzes the deesterification of the fatty acids from the glycerol, is able to get to its substrate when the fat globule membrane is disturbed. This happens when raw milk is held static in a running centrifugal pump, when raw milk is homogenized before it is pasteurized, or when raw milk is inadvertently mixed with homogenized milk. It may also occur when microorganisms, particularly psychrotrophs, produce and release Upases into homogenized milk.5 Preparation of Training Samples. Rancid milk can be prepared by adding equal quantities of raw milk to freshly pasteurized and homogenized milk and holding several hours cold while the flavor develops. Bodyfelt suggests mixing 100 ml of raw milk with 100 ml of pasteurized and homogenized milk in a Waring blender or a similar mixer for 2 min, then making it up to 600 ml with pasteurized and homogenized milk. He suggests making it up 2 to 3 days ahead and holding cold while the flavor develops. In both cases, it is important to heat the milk to 700C for 5 to 10 min and cool after the flavor has developed.5 Salty Description. The descriptive term "salty" is commonly known and a good term to describe this flavor. It is perceived quickly on placing the sample in the mouth. No aroma or odor accompanies the salty flavor. It lends a cleansing feeling to the mouth.5 The author perceives the salty sensation as "warm" and lacking refreshing character. Cause. Cows in the advanced stages of lactation and cows that have clinical stages of mastitis often have high salt content in their milk and a salty flavor. Comingled milk seldom has an abnormal salt level nor a salty taste. Preparation of Training Samples. Add a pinch of sodium chloride at a time to pasteurized and homogenized milk while stirring to dissolve the salt until it is at the desired strength. Unclean Description. Milk with an "unclean" flavor is readily noted when the sample enters the mouth. The flavor and odor are offensive, suggesting extreme staleness, mustiness, putrid, "dirty sock," or spoiled. The flavor fails to clean up after the milk is expectorated. Cause. This flavor develops in milk when psychotropic bacteria are allowed to grow to high numbers in milk and particularly when held at temperatures above 7.2°C. The presence of psychrotrophs is usually due to poor on-farm sanitation. High numbers are generally due to poor bulk tank cooling. Preparation of Training Samples. To find ' 'unclean'' flavored milk, examine several samples of milk that are beyond their pull date. If the flavor is not found, incubate them for 4 to 12 h at room temperature and reexamine them. When an exemplary sample is found, it may be maintained in the refrigerator and used as an inoculum for production of future training samples.5 3.4.2 Cottage Cheese 3.4.2./ Introduction Cottage cheese is a curd that is formed by the acid coagulation of pasteurized skim milk. The acid may be formed by lactic acid bacteria that are added to the milk which consume lactose and convert it to lactic acid.63 In one successful method, part