SF 253 .S8 Copy 1 ^HE LECITHIN CONTENT OF BUTTER AND ITS POSSIBLE RELATIONSHIP TO THE FISHY FLAVOR A THESIS Presented to the Faculty of the Graduate School OF Cornell University for the Degree of DOCTOR OF PHILOSOPHY BY GEORGE CORNELL SUPPLEE [Reprint from the Cornell University Experiment Station Memoir 29, November, 1919] THE LECITHIN CONTENT OF BUTTER AND ITS POSSIBLE RELATIONSHIP TO THE FISHY FLAVOR A THESIS Presented to the Faculty of the Graduate School OF Cornell University for the Degree of DOCTOR OF PHILOSOPHY BY GEORGE CORNELL SUPPLEE [Reprint from the Cornell University Experiment Station Memoir 29, November, 19 19] O' ^>3 X Of -.'. CONTENTS PAGE Previous inv<>stisafi()iis 101 Lecithin cleconii)osition in butter as a possible eaus(^ of the fishy flavor lo4 Chemical constitution, properties, and distribution of lecithin. . . . 104 Theoretical discussion 106 Investigational work 108 Quahtative determination of lecitliin in butter 108 Amount of lecithin in butter 110 Trimethylamine salts of the fatty acids Ill Effect of working trimethylamine salts of the fatty acids into butter. ..........'.. ' 113 Quantitative estimation of trimethylamine^ in fishy butter 123 Development of fishy flavor in experimental butters 12(3 Variation in acid value of experimental butters 129 Trimethylamine and ammonia content of experimental butters. , 134 Bacteriological studies 136 Bacterial analysis 137 Inoculations for the purpose of developing the fishy flavor. . 137 Longevity of Bactermni ichthyosmius in butter 141 Further studies with Bacterium ichthyosmius 142 Trimethjdamine and ammonia production by Bacterium ich- thyosmius in milk and cream 144 Production of trimethylamine from lecithin and choline^ by bacterial action 145 Bacterial inoculations into l;)utterfat 146 Summary 148 Acknowledgments 149 Bibliography 150 97 THE LECITHIN CONTENT OF BUTTER AND ITS POSSIBLE RELATIONSHIP TO THE FISHY FLAVOR THE LECITHIN CONTENT OF BUTTER AND ITS POSSIBLE RELATIONSHIP TO THE FISHY FLAVORS George Cornell Supplee The commercial value of butter is based to a great extent on its quality, which in turn is determined" by its flavor. The commercial grading of butter on this basis, and the recognition of certain characteristic defects, have resulted in the establishment of certain terms more or less descriptive of the flavors found. Among the terms commonh^ applied to the flavors in butter are such words as rnetaUic, fishy, oily, raucif}, (alloivj/. Since the presence of any of these flavors carries with it a reduction in commercial value of the butter, considerable effort has bc^en made to detei'mine their causes and prevent their development. Unfortunately most of these efforts have not met with a high degree of success. This may be ascribed to several reasons, among which are the following: lack of positive idcuitificatioTi of the same flavor by different investigators; lack of ad(>quate chemical methods for the isolation and measurement of the small amount of sub- stance capable of producing the flavor; and lack of coopei'ation between the chemist, the l^acteriologist, and the experienced butter judge. Fishiness in buttcM-, with which this investigation is primarily con- cerned, is usually (h^scribed as a flavor resembling that of salmon or mackerel, altho the names of other varieties of fish are occasionally used to describe the flavor more explicitly. While the typical fishy flavor in butter is readily recognized by experts, it is oftcni accompanied by a more or less oity condition whi(;h tends to create differen(;es of opinion as to its exact nature. But if the opinion of butter judges of long experience is to be considered as trustworthy, it may be said that the true fishy flavor is entirely distinct from the oily flavor even tho the oily condition may precede or accompany it. PREVIOUS INVESTIGATIONS The earher investigations bearing on fishiness in dairy products have been largely confined to milk and butter. One instance is recordcKl, how- ever, in which this condition was observed and studied in evaporated milk. ' Also presented to the Faculty of the Graduate School of Cornell University, December, 191 S, as a major thesis in partial fulfiUnient of the reiiuirenients for the degree of doctor of philosophy. 101 102 George Cornell Supplee Harding, Rogers, and Smith (1900)- report the fishy flavor in a sample of milk brought to the New York State Agricultural Experiment Station at Geneva in 1900. The source was traced to a single animal in the herd, but investigation failed to locate any pathological condition or any irregularity in the feeding which might cause the trouble. Attempts to reproduce the flavor by inoculating milk with bacteria isolated from this cow's udder also fail(Ml. The same authors mention also an instance brought to their attention by W. E. Griffith, in which a peculiar flavor developed in June butter after storage at 18° to 22° F. This flavor was described by butter experts as fishy. Piffard (1901) discusses the fishy flavor in dairy products, and suggests the possible relationship between certain algae found in stagnant water, and fishiness in milk and butter. His theory is supported l)y the fact that the flavor is often produced in water by the growth of algae and diatoms, and he lielieves, therefore, that cows having access to such water may transmit the condition to milk. Refening to tlie flavor in butter, he considers the idea that salt may be resi)onsil)le and mentions the ability of salt to absorb flavors and odors of materials stored near it. Harrison (1902), discussing butter defects at about the same time, states that the characteristic off-flavors of butter — fishy butter being specifically mentioned — are caused l)y the growth of undesirable bacteria in the cream. O'Callaghan (1902) published certain observations on fishiness in Australian butter. He states that he has found this condition in butter only two hours old. From his investigations he concludes that O'idium lad is is the causal agent. Later (O'Callaghan, 1908) he elal)orated on his former views, concluding that Oidium lactis associated with the lactic- acid bacteria in cream will usually produce a fishy flavor in the butter. He mentions the presence of the defect in uiLsalted butter, and recom- mends the improvement of sanitary conditions in the creameries, and pasteurization, as a remed}'. His conclusions have not been confirmed by other investigators. Rogers (1909), after a rather exhaustive stutiy of tiie occurrence and cause of fishy butter, confirms many observations commonly noted in connection with this ti'ouble but is unable to confirm the ol)servations of O'Callagiian. He also s-emingly eliminates the theory that trimethyl- - Dates in parenthesis refer to Biblioyraphy, page 150. The Lecithin Content of Butter 103 amine is directly responsible for the flavor. After studying the effect of high-acid cream, overworking, and the consequent increased oxygen con- tent of the butter, and by conducting bacteriological investigations, he concludes that high-acid cream is essential to loring about the condition, altho he points out that not all l^utter made from such cream develops the fishy flavor. In this respect he states (page 20 of reference cited) that '^ fishy flavor may be produced with reasonable certainty by over- working the butter made from sour cream." From his viewpoint the probability that microorganisms are the cause falls into disfavor by the advancement of the opinion that " fishy flavor is caused by a slow, spon- taneous, chemical change to which acid is essential and which is favored by the presence of small amounts of oxygen " (page 20 of reference). More recently the same author (Rogei'S, 1914, a and b) points out that fishiness in butter may bo preceded by an oily or a metalhc flavor, and reiterates his views that the evidence is against the theory that the fishy condition is of a bacterial nature. He also states (1914b) that "fishy flavor is said to occur rarely or not at all in unsalted butter and it is possible that the salt furnishes certain conditions which are essential to the development of the fiavoi'." Reakes, Cuddie, anil Reid (1912) find no significant differences in the bacterial flora of fishy and of high-grade butter, and, in agreement with Rogers, state that " the development of fishy flavour in butter arises as a result of a chemical change inducing a splitting-up of some of the con- stituents into compovmds possessing this peculiar character of smell and taste, the factors responsil>le for such change being apparently a degree of high acidity of the cream and overworking." Hunziker (1916) states that high pasteurization temperatures (185° F.) when used on sour cream tend to produce a very i)Oor quality of butter, which often has a disagreeable oily taste suggestive of fishiness. He points out that this is particularly true when cows are on green pasture and the butterfat contains a rather high percentages of olein, which may b.' oxidized with relative ease in the presence of high temperatures and high acid. Hammer (1917) reports that he found a can of evaporated milk which possessed a marked fishy flavor and odor and fi-om which he was able to isolate an organism heretofoi-(? utulescribed. H(^ gives to this organism the name Bacterium ichthyosiniu-'^, which was suggested by Di-. A. W. 104 George Cornell Supplee Dox. The flescription of the ori>;anisin seems to indicate that it is closely allied to the Proteus group. By inoculation experiments Hammer was able to reproduce the flavor in milk and cream under both aerobic and anaerobic conditions. He noted that the intensity of the odor was in- creased by the addition of alkali to the milk after the incubation period. He was unable, however, to produce fishiness in butter by direct inocula- tion or by inoculating the cream before churning. Bacteria counts at various intervals during the storage period showed an immediate decrease in numbers in salted butter, and an increase tluring the first few days in unsalted buttei- followed by a pronounced decreas(\ Washburn and Dahlberg (1918), while studying the influence of salt on storage butter, found that salted butter was more likely to turn fishy in stoi-age than was vmsalted butter, and furthermore that there appeared to be a tendency toward a progressive development of the flavor thru metallic to oily and finally to fishy. LECITHIN DECOMPOSITION IN BUTTER AS A POSSIBLE CAUSE OF THE FISHY FLAVOR CHEMICAL CONSTITUTION, PROPERTIES, AND DISTRIBUTION OF LECITHIN Lecithin, which stands in close relation to the fats, belongs to a more or less definite group of substances known as phosphatidt^s, or phos- phorized fats. These bodies appear to be a group of esters containing nitrogen, phosphoric acid, and fatty-acid radicals. Lecithin, wliich is the best known of the phosphatides, contains two fatty-acid radicals and the nitrogenous base choline, combined with glyccn-ophosphoric acid. According to the kintl of fatty acid present in the molecule, it is possible to have various types of lecithin, such as stearyl, palmityl, and oleyl. A number of investigators seem to agree that every true lecithin con- tains at least one oleic- acid radical. There seems to be uncertainty as to whether choline is th(> only base present in lecithin. MacLean (1909) was able to get only 42 per cent of the theoretical amount from lecithin isolated from heart muscle, and 65 per cent from k^cithin of egg yolk. Other investigators have found the same to hold true of lecithin from different sources. Lecithin has certain properties in common witli the fats, particularly with respect to its solvents. It diliers, however, by being less soluble The Lecithin Content of Butter 105 in ether and more soluble in alcohol. It is precipitated from alcoholic solution by acetone; in water it swells to a colloidal mass which on micro- scopic examination appears as oily drops and threads. It saponifies with alkahes and baryta water, yielding the corresponding soaps, salts of glycerophosphoric acid, and choline. Hammarsten and Hedin (1915) state that it is slowly decomposed by dilute acids and enzymes (lipase). Barger (1914) states that Bacterium prodigiosus produces trimethylamine from chohne and lecithin; he also cites references to show that lecithin is decomposed during putrefaction, yielding fatty acids, glycerophosphoric acid, choline, and ultimately trimethylamine. Hasebroek (1888) claims that methylamine, ammonia, methane, and carbon dioxide may be finally produced from choline during putrefaction. On being heated with strong caustic soda or potash, lecithin yields trimethylamine, which has a dis- tinct fishy odor, this being one of the characteristic qualitative tests for it. Leathes (1913), in citing the work of various investigators, seems to think that lecithin is rather unstable. He thinks this property is due to the unsaturated oleic-acid radical contained, and offers this as the reason why the substance gives Pettenkofer's reaction. The work of Long (1908), however, seems to indicate that lecithin is more stable than has been generally believed. Koch (1902-03) has shown that various salts will cause lecithin to precipitate as a gelatinous mass, and that acids, if sufficiently dissociated (0.005 M sulfuric), will accomphsh the same thing. Lecithin seems to be widely found in nature, being present in many plant cells and animal fluids. It is particularly abundant in the brain, the nerve tissue, and the yolk of egg. It is also reported as existing in blood corpuscles, blood plasma, lymph, milk, and bile. Since the methods used for the quantitative^ estimation of lecithin depend on the deter- mination of phosphorus in alcoholic or alcohol and ether extracts, it is doubtful whether the figures given are absolutely correct due to the fact that other phosphatides are extracted and also because the empirical fornuila used in the calculation may be inaccurate for the particular lecithin Involved. Altho there have been conflicting statements as to whether milk contains lecithin, there seems to be sufficient evidence that it does. The results obtained by Nerking and Haensel (1908) are sub- mitted in table 1: 106 George Cornell Supplee TABLE 1. The Lecithin Content of Various ^NIilks (From Nerking and Haensel) Kind of milk Human, 10 samples Cow's, 17 samples. . Ass's, 6 samples. . . Ewe's, 4 samples . . Goat's, 11 samples. Mare's, 8 samples . Percentage of lecithin Highest Lowest Average 080 024 0.050 110 036 003 o:]9 0.000 016 107 051 083 075 030 040 017 007 Oil Glikin (1909), studyino; the lecithin mid iron content of milk, reports 0.0515 per cent lecithin in whole milk, 0.05 per cent in cream, and 0.1329 per cent in Iminaii milk. Fetzer (1911), studying the lecithin content of milk imder pathological conditions, finds (hat it is lower in milk from cows suffering with mastitis than in milk from normal cows. He finds also that the lecithin content decreases as the fat decreases. The work of Bordas and De Haczkowski (1902) indicates that the amount of lecithin varies witli the lactation ])eriod. They find that it is at the maximum at the bi'gimiing, and gradually decreases during the remainder of the period. Their observations were from seven cows. THEOKETieAL OISCI'SSIOX Trimethylaniine as a decomiiosition product of lecithin was brought to the attention of the writei- as a possible cause of the fishy flavor in butter about thrcH' y(\us ;igo, when he was working with lecithin isolated from brain tissue. Altlio tlicre seem to be no pul)lished data concerning trimethylaniine in relation to this subject, and Hogers (1909) claims that it can be work(Ml into l)utter in large amounts without producing the fishy tiavor, it is neveithdi'ss believed by many that this substance is in some way resp()nsil>le. On boiling l(>citliin isolated from brain tissue and egg yolk with stiong caustic soda. tli(> writ(a' has been able to obtain a distinct ()il\- ;ind fishy odor which \vas assei'ted by many to be typical of the odor of fishy buttei'. The only possible sourc(> of such an odor in this case w;!S tlu* t riniet hylnniine derived from the lecithin. This i-esult, together with the fact that there s(>ems to I;)e good evidence that lecithin is [iresent in milk, led to the assumption that the substance The Lecithin Content of Butter 107 may exist also in butter and that by its decomposition it can slowly liberate trimethylamine with the consequent production of the fishy flavor and odor. Provided that lecithin can be shown to exist in butter, the above assump- tion is supported by several facts. It is well known that enzymes are capable of bringing about many chemical decompositions which result from the action of acids and alkalies on organic substances. It also appears to be a fact that butter may contain enzymes derived from the udder, and from the action of bacteria in the milk or the cream before it is made into butter, and furthermore that the activity of such enzymes is not entirely stopped at the temperature at which butter is stored. Hammar- sten and Hedin (1915) state that lecithin is decomposed by dilute acids and enzymes. The citations of Hasebroek (1888), Barger (1914), and others show that lecithin is decomposed by bacteria yielding choline, which finally yields trimethylamine. It is also a well-known fact that this substance in very small amounts possesses a distinct fishy odor, but in concentrated solution it has a strong ainmoniacal odor. Speaking of the former property, Davis (1912), quoting Tayloi', states that the " odour [referring to the peculiar fishy odor suggestive of herring brine] is grad- ually developed by adding lime to a solution of the base, but requires some time to reach its maximum intensity." In addition to the forc^going facts the writer has observed certain fea- tures that may have some beaiing on this problem. In inspecting butter u.sed in the Navy, it has been noticed that certain samples of cream evolved a peculiar fishy odor on the addition of alkali used for titration. This phenomenon was first brought to the attention of the writer by A. M. Besemer, and has since been confirmed by a number of men, some of whom have wide reputation as butter judges. Since trimethylamine is a base which is liberated from its acid combinations by alkalies, it is quite possible that the odor mentioned above was due to this substance's having been liberated from its acid combination in the cream. If such were the case, it is conceivable that butter made from such cream might, during storage, give up its trimethylamine thru the action of enzymes. In this connection it has been noted that certain samples of old butter, which wei-e not scored as fishy, when brought into contact with a warm solution of soap powder would give off a strong herring-like odor. This phenomenon might also be explained as in the case of the cream. In addition to these features it has been noted that certain samples of fishy 108 George Cornell Supplee biittcn- may lose tluMr charaetoi-istic flavor aftcn- a period of time. This has also been observed by other investigatois. It is possible that this characteristic may be explained by the fact that, since trimethylamine is extremely volatile, it may pass off, or that the instability of the acid combination clianj^cs so that tlie conditions are not right for its manifes- tation. The writer has noted a very strong fishy odor in partially decom- pos(Ml egg yolk held at refrigerator temperature, which had entirely dis- appeared two wcH'ks later. On the basis of the foregoing facts and observations and the evident lack of contradiction of most of them with what is known about fishy butter, the following experimental woi'k was plamicd with the object of determining the possible i-elationship of trimethylamine to this flavor. In calling attention to the lecithin, it may be stated that the writer is cognizant of the fact that trimethylamine may Iw produced from other sui)stances. This material has been chosen as the object of study pri- marily Ix'cause there is exact knowledge concerning its cleavage and some of the agencies bringing this condition about. TXVESTIGATIONAL WORK QUALITATIVE DETERMINATION OF LECITHIN IN BUTTER The first experimental work undertaken was to demonstrate the pres- ence of lecithin in butter, since there ap[)ear(>d to be no rei)orts on this point in the hterature. One hundred gi'ains of melted butter was thoroly mixed with sufficient anhydrous calcium sulfate (alwut one kilogram) so that tlu^ mixtui'e retaininl its dry powdered form to sucli an extent that it could be icadily sifted l3etwe(ni the thumb and tlie finger. The mixture was transfencd to a sp(>ciallv constructed pcM'colator and extracted for 48 hours with D.Vpei-cent alcoh.ol at (50° (\ The alcoholic extract w.as evaporat(>d down and the residue was treated with a small amount of ether, wiiich rook uj) the fat, the fatty acids, ;uid part of the lecithin. The ])ait iiisohible in eth(M- was again taken up with warm alcohol, and what may i)e termed the lecithin Jradion was precipitatcnl by thoroly cooling \\\v alcoholic sohition. The substance thus obtained precipitated iTi the foiin of small, wart-like masses, which clung tenaciously to the sides of the })eakei'. On this material, which presumably contained a high proporiioii of lecithin, various observations were madf^ and qualita- tive tests applied. The following charactei'istics were noted: The Lecithin Content of Butter 109 On drying at oi-dinary temperature and pressun^ tlie material appeared as a semi-amorphous and oily substance of a pale, dirty yellow color. It was entirely soluble in alcoliol but was partially thi'own out of solu- tion by the addition of an excess of ether. The precipitate formed in this manner was finely granular and was white in color. In water it formed a semi-colloidal solution which on microscopic examination ap- peared as minute oily droi)s. When the watery suspension was heated, the particles would cohere to form a sticky mass which changed to a distinct brown color. Both the dry substance and the watery suspension, when heated with strong caustic soda, gave off a marked fishy odor resembhng sometimes dried herring and sometimes salmon oil. This observation was in the great majority of cases confirmed by a number of colleagues. The fishy odor obtained from the material in this manner seemed to furnish positive evidence that lecithin was present. To further strengthen this behef, Pettenkofer's test with sugar and sulfuric acid was applied to the dry material with positive results. The above observations were confirmed with lecithin extracted from fresh butter, salted and unsalted, and from other miscellaneous samples of normal butter. Altho the evidence that lecithin exists in butter in del (K't able quanti- ties seemed conclusive, it was decided to determine, if }")ossible, the pres- ence of choline, which, as already pointed out, is one of the components of the lecithin molecule. This was accomplished by boiling the residue of the first alcohohc extract witli baryta water, which removed all fat, fatty acids, and fatty-acid radicals of the lecithin in the form of barimn soaps. After the barium soaps were filtered off, the excess barium was removed with carbon dioxide, the barium carbonate filtered off, and the filtrate containing choline^ and barium glycerophosphate evaporated to a sirupy consistency. This I'esidue was then treated witli absolute alcohol, in which choline is solubl(> but baiium glyeerophosj)hate is insoluble. On evaporation of the al)solute alcohol a small amount of sirupy sub- stance remained. To tliis material qualitative tests for choline were applied. The most characteristic of such tests is the periodide test described by Stanek (1905), which is made by adding a small amount of strong iodine solution (153 grams of iodine and 100 grams of potassium iodid(> in 200 grams of water) to an aqueous solution of choline. A posi- tive test is indicated by the formation of a brown precipitate of choline periodide, which on microscopic examination in the presence of the reagent appears as dark brown refractive and notched prisms or rhomboidal no George Cornell Supplee leaflets. On evaporation of the reap;ent the crystals lose their shape and appear to liquefy, forming l^rown, oily droi)lets which again assume their crystalline structure on the addition of more reagent. On the application of this test to the choline ol)tained from butter lecithin. ^^ was foimd that the results conformed in all respects to the descriptions of this periodide. The accompanying platc^ of photomicrogi'aphs (Plate VI) shows the characteristic crystals and oily droplets of the periodide foi-med by the cholin(> from flutter. In addition to this test it was shown that the small amount of eiioline obtained would give off a slight but distinct fisliy odor on being heated with solid caustic soda. AMOUNT OF lecithin IN BUTTKR Since the qualitative tests sc(Mned to leave no room for doubt as to the pr(\sence of lecithin in butter, the next step was the quantitative estimation of this sul)stance. In view of the evident (Hfhculty in securing an absolutely pure lecithin free from oth(>r phosphatides, the estimations were basc^d on the phosphorus content of extracts and the amount of lecithin calculated acconUng to thc^ formula of the distearyl type. The results of such determinations on various types of butter made from different lots of cream are shown in table 2: TABLIC 2. Lecithin Content in Vaiiious Butters* Sample Type of creani from which butter was made Age of butter (days) P^O, (per cent) Lecithin, distearyl type (per cent) 1 Raw sweet Pasteurized sweet Raw ripened Pasteurized ripened Raw sweet Pasteurized sweet Raw ripened Pasteurized ripened Raw sweet G (i (> G 48 48 4S 48 72 72 72 72 0127 0127 0122 007.5 00i)2 0120 008(1 {){)K', 0111 OOS'.t 00S3 0.009.5 0.0723 2 0723 3 0G93 4 0133 5 ()r.22 6 0G82 7 (M8S 8 0171 9 Otiol 10 Pasteurized sweet Raw ripened OoO.") 11 0471 12 Pasteurized ripened OMO * Thcsp figures were furnished by J. T. Cusick, fbeiiiist for the State Department of Agriculture, located at Cornell University. Memoir 29 Plate VI °>^ \> ^h ^ \ I '. ,./ * \\ / .^ -^^ ^ CHOLINE PERIODIDE CiiYSTALS OBTAINED FROM CHOLINE OF BUTTER LECITHIN Photouiicro-^raplis, X 3S0 The Lecithin Content of Butter 111 A study of table 2 shows a fairly constant lecithin content in butter from various lots of cream and in different types of butter made from the same lot of cream. There is one feature, h()\v(>\-ei-, wliicli is worthy of note, and this is that in most instances there is a tendency toward a lower lecithin content in the ripen(>d-cream butter than in that made from unripened cream. This may be sio;nificant in the litiht of the statement by Hammarsten and Hedin (11)1")), that lecithin is decomposed by dilute acids and enzymes. This fact applied to these nsults mioht indicate that the acidity of the cream slowly decomposed the lecithin, and its decomposition products, particularly the tilycerophosphoric acid, were washed out with the buttermilk. If such wei-e the case it would be very easy to acc-ount for the lower phosphoi-us content in sour-cream butter. TRIMKTHVLAMIXK SALTS OF THE FATTY ACIDS It was decided that l^efore an attempt was made to correlate trimethyl- amine with the hshy flavor of butter, this substance should be prepared in a pure state and those characteristics determined which mifi;ht have a bearing on this joarticular jiroblem. Trimethylamine was made by heating 50 grams of ammonium chloi'ide and 440 grams of a 40-per-cent solution of formaldehyde in the autochive at 122° ('. for thirty minutes. Any excess formaldehyde was then exiK>ll(>d and th(> trimc^thylamine Uberated from its hydrochloride by distilling from an alkaline solution. A 10-per-cent solution was easily obtained at the oixlinary temperature and pressure. The trimethylamine thus procured was combined with lactic, butyiic. oleic, and stearic acids, and also with the mixed soluble and insoluble fatty acids obtaincMJ from butter according to the procedure outlined by Browne (1899). While \hv piopinties of these acid addition products seemed to offer an interesting fi<^ld for study, only such of their characteristics luv recorded here as might have a direct relationship to the fishy flavor in butter, namc^ly, their stability, volatility, and behavior in the pres(>nce of sodium chloride. The lactic-acid combination with (limethylamine proved to be a rela- tively stable oily licjuid })()ssessing no characteiistic taste other than that shown by many conunon salts. The odor, c>s()(H'ially after the liquid had been standing in a stoppei-ed bottle, seemed to be slightly i\shy. Evidence on this point is not conclusive because it is possible^ that this was due to excess trimethylamine added at the time of neutralization and 112 George Cornell Supplee not shown by the indicator used. Tlc^atinji; socnu^d to intensify the odor to some extent, wliich would incUcute instabihty at high temperatures. The butyiic-acid (■oml)ination witli trimethylamine was a substance extremely volatile^ at ortUnar}' temperatures. Tlu^ odor greatly resembled fish oil; the taste resembling this product was manifest onh^ when very small amounts were us(>d, and then not regularly. Oleic acid and Irinu'thylamine formed a soft soai) whi(;h was very unstable as evidenced by the lilx'i'ation of the trimethylamine; the greater the dilution, however, th(^ less this condition was manifested. This soap could not \)c obtained entirely free from water, even at the ordinary temperatures, because of the sinuiltaneous gi\-ing-off of ti'imethylamine and watcn'. Tliis liberation was sut'li that nothing remained but the free acid. Fui'thermoi-e, on the addition of sodium chloi'ide to its water solution, the sodium ion i'eadil\" r(>i)iaced the trimethylamine radical, with the consequent jirecipitation of the sodium soap and the formation of trimethjdamine hydrochloride. The trimethylamine stt>arate showed the same charactei'istics as the oleic soap, but to an e\'en greater extent. 'I'rimetliylaniine was con- stantly given off in large (luantities, and the onl>' way in which it could be handled as a soap was in a mixture of alcohol and wafer in a tightly stoppered bottl(\ The mixed solubl(> and ins()lul)le fatty acids coml)ined with trimethyl- amine showed the same genei'al characteristics as the l)utyric and stearic combinations, respectively. The instability of the combinations of fatty acid and trimethylamine can undoul)tedly be accounlc^d foi' l)y the fact that they are addition products in which the tri\alent nitrogen of th(> latter substance cb.anges to the pentavalent condition in the presence of an acid. The fact that these are weak acids with relatively large molecules is probably also significant. It was observed that the hydrochloride was more stable than the above salts, and that the sulfate was even more stable than the liydrochloride. Tliis instabihty M the fatty-acid combinations and tlieii- iT'action in the presence of sodivun chloride may have an imi)or- tant bearing on the relation of trimethylamine to fisliiness in butter, and may ho of particular significance in explaining why the flavor is usually foimd in salted butter. As 3'et, however, the relationsliip is not clear. The Lecithin Content of Butter 113 effect of workinci trimethylamine salts of the fatty acids into butter Tho character of the ii-inu'lliylaminc salts of tli(> fatty acids in pure state seemed to justify t!ie following series of experiments, in which these salts are incorporated into various typ(^s of ])utier for the purpose of determining tiic possibility of their producing the fishy flavor in the presence of butterfat. In vi(>w of the d(>sirability of incorporating the trimethylamine in logical amounts, tlu> following plan was adoi)t(Hl: The largest quantity of lecithin reported in cow's milk by Nei'king and Haensel (1908) was used as th(^ rxisis of calculation. These authors report 0.1 IG per cent as th(; largest amount found in seventeen samples. For the calculations of this experiment, this was assumed to be lecithin of the dist(>aryl type, and it was further assuuKHl to be pure lecithin with the empirical fornmla assigiunl to the type named. Granting these assumptions, this amount would yield on complete decomposition the equivak'nt of 8o i):irts pvr million of trimethylamine. This substance alone or in acid combination was therefore added to cream, wash water, or butt(>r on this basis. It is very evident that because^ of the volatihty of some of tlu> materials and because of mechanical loss, none of the samples of butter wh(m completed would contain 85 parts per million of trimethylamine. The method of arriving at the (juantity to be added seemed to afford a uniform basis and to a])pi-oximate in a logical manner the amount of this substance^ that might be i)r()duced in butter. When the fatty acids wei-e usetl alone; they were added in {]uantiti(>s equivalent to the amounts added in the corresponding trimc^thylamine salts. The addition of the acids was merely for the purpose of checking against the trimethylamine. In tabl(>s 3 to 9 inclusive are shown the comments of various judges on diffei-ent types of butter containing trimethylamine addc^l as already indicated and iMcor])orated by various means. Because of the great importance of the personal factor in judging butter, an effort was made in all cases to get a nunibca- of men familiar with the various flavor defects of the product. In all cas(>s tlie samples were so labeled that the judges had no knowledge of th(>ir contents. They were instructed to comment on the flavor and to work independently of one another, and it is believed that this injunction was can-i(Hl out. In presenting the results in tabular form the autho)- has intentionally omitted comments having no direct bearing on the fishy flavor. 114 George Cornell Supplee TABLE 3. Effect on the Flavor of Buttek, of Adding Tkimethylamine and Fatty Acids to Raw Sweet Cream with 0.23 per Cent Acid at the Bate of 85 Parts per Million of the Former {S indicates salted butter) Sample Material added Coinnients by judges No. 1 No. 2 No. 3 No. 4 No. 5 No. 6 AS Nothing . Oily Oilv Oily A Nothiiis Oily I'ishy 1 AS Trinietlivlaiiiinc . . Oilv I-ishy i'ishy Fishy 1 A Trimethylamine Triincthvlaniinc lactate 2 AS 2 A Triinetliylatiiiue lactate Triinethylamine liutyratc Tritncthylaiiiiiie butyrate Triiiicth>larninc oleate Fisiiy !"isliv i'ishy I'ishy ( )ily l-'ishv Oily I'i.shy iMshy I'ishy 3 AS 3 A 4 AS I'isliy I'ishy 4 A 5 AS Trinietliylamirie stcarate TriincMiylainine stearate TriiiiethylainiiK' aiul sohihle fatly acids i'ishv fisiiy 5 A 6 AS I'ishy Oily I'isliy I'lsliy Fishy 6 A Triinctliylaniiiic and solublf fatty iicids Fish\- l- 8 A Lactic acid Butyric acid Butyric acid FLshy I'ishv I'ishy Oily 9 AS 9 A Oily Fishy 10 AS ( )i(Mc acid I'isliy I'ishy 10 A 11 AS Solul)ie fattv acids 11 A Soluljle fatty acids 12 AS 12 A Insolultlc fattv acids The Lecithin Content of Butter 115 TABLE 4. Effect on the Flavor of Butter, of Adding Trimethylamine and Fatty Acids to Raw Sweet Cream with 0.23 per Cent Acid at the Rate of 85 Parts per Million and Then Washing the Butter in Water Containing the Same Concen- tration OF the Various Substances (S indicates salted butter) Sample Material added Comments by judges No. 1 No. 2 No. 3 No. 4 No. 5 No 6 BS Nothing Oily Oily Oily Oily Fishy Oily B Nothing Oily Fishy 1 BS Trimethylamine Fishy Fishy Fishy Fishy Fishy Fishy Fishy 1 B Trimethylamine 2BS Triraethvlamine lactate Oily Oily Oily Fishy 2B Trimethylamine lactate Trimethylamine butyrate Trimethylamine butyrate Trimethylamine oleate Fishy I'ishy Fishy I'Mshv Fishy l-'ishy Fishy 3BS Fishy 3B Fishy Fishy Fishy 4BS 4B Trimethylamine oleate Fishy Fishy Fishy 5BS 5B Trimothyiamine stearate Trimethylamine stearate Trimethylamine and soluble fatty acids Oily Fishy Fishy Fishy Fishy Oily GBS Fishy Fi.shy Fishy Fishy Fishy Oily Fishy Fishy Fishy Fishy GB Trimethylamine and soluble fatty acids 7BS Trimethylamine and insoluble Oily Fishy Fishy Fishy 7B Trimethylamine and insoluble fattv acids 8BS Lactic acid Lactic acid Butyric acid .... 8 B 9 BS Oily Oily I'Mshy 9 B I'ishy Oily Fishy 10 BS Oleic acid 10 B Oleic acid 11 BS 11 B Soluble fatty acids 12 BS Insoluble fattj' acids 12 B 116 George Cornell Supplee TABLE 5. Effect ON the Flavor of Butteh, of Adding Trimethylamine and Fatty Acids to Raw Sweet Cream with 0.23 per Cent Acid at the ]{ate of 85 Parts per Million and Then Working the Substances Directly into the Butter at the Same Rate (S indicates salted butter) Sample CS c 1 CS 1 c 2CS 2C 8CS 3C 4CS 4C r)Cs 5 C ()CS «)C 7CS 7C 8CS 8C «» C^ 9C IOCS 10 C 11 CS 11 c 12 CS 12 C Material added Nothing Nothint! Triinethylainiiio. Trimethvlainiue. Triinethylatiiine lactat(^ . . Triinethylaniiue lactate . . Triinethylainine hufyraie Triiiiethvlainitie Inityrate Triinethylainine oleate Trinu'thylaminc oleate Triinethylainine stearate Triinethylainine stearate Triinethvlamine and sohible fatty acids." Triinethylainine and soluble fatty acids Triinethylaniiue and insoluble fatty acids Triinethylainine and insoluble fatt.v acids Lactic acid . Lactic acid . Butyric acid . Butvric acid . Oleic acid . Oleic acid . Soluble fatty acids Soluble fattv acids Lisolnbie fat.ty aeids Iusolu!)le fattv ticids Comments by judges Xo. 1 I'ishv I'ishv I'ishy Fishv Kishv Kishv l-'ishv Fishv ■'ishy ■ishv Fish Oilv Fishy Fisiiv Oilv Oily No. 2 Oily Fi.shv l-'ishy I'ishv Fishy Fishv l-'i.shv Fishv l''ishv Xo. 3 l"i.shv Fishv iMshy I'ishv I'ishy I'ishy I'ishv Fishy Fishy Mshv Fisliv I'lshy Fishv Xo. 4 No. 5 Oily l-'ishy Fishy iMshy I'ishv Fishv Fishy Fishy Fishy Fishy F'ishv Fishy Fishv I'^ishv Oily Oily F'ishy Fishy Fishy Fishy I'ishv' Fishy iMshy Fishy Fishy Fishy Oily No. Fishy Fishy Fishy Fishy Fishv Fishy Fishy Fishy Fishy Fishy The Lecithin Content of Butter 117 TABLE 6. Effect on the Flavor of Butter Made from Pasteurized Ripened Cream with 0.32 per Cent Acid, Working Trimethylamine and Fatty Acids Directly into the Butter at the Rate of So Parts per Million (S indicates salted butter) Sample Material added Comments by judg 3S Xo. 1 No. 2 No. 3 Xo. 4 Xo. 5 No. 6 DS Nothing D Nothing; 1 DS Trimethvlaniine Fishy Fishy Fishy I'ishy r'ishy Fishy Fishy 1 D Trimethylamine 2DS Trimethylamine lactate 2D Trimethylamine lactate 3DS 3 D Trimethylamine butyrate Trimethylamine butvrate Oily I'ishv Fishy Oily Oily Fishy Fishy Fishy 4DS Trimethylamine olcate 4D Trimethvlaniine oleate 5 DS Trimethylamirie stearate I'isliv Oily Fishy I'"ishy 5 D Trimethylamine stearate Trimethylamine and soluble fatty acids . Oily 6 DS Fishy 6D Trimethylamine and soluble fatty acids 7 DS Trimethylamine and insoluble Fishy 7 D Trimethylamine and insoluble fatty acids l''ish\- 8 DS Oily 8 D 9 DS Butvric acid Fishy Fishy Fishy 9 D ButjTic acid 10 DS Oleic acid Fishy Fishy 10 D 11 DS 11 D 12 DS Insoluble fatty acids Oilv Oily Oily Oily 12 D 1 118 George Cornell Supplee TABLE 7. Effect on the Flavor of Butter Made from Pasteurized Sweet Cream WITH 0.16 per Cent Acid, Working Trimethylamine and Fatty Acids Directly INTO THE Butter at the Rate of 85 Parts per Million {S indicates salted butter) Sample Material added Comments by judges No. 1 No. 2 No. 3 No. 4 No. 5 No. 6 ES Nothing E Nothing . 1 ES 1 E Trimethylamine Trimethylamine Trimethylamine lactate Trimethylamine lactate Fishy Fishy Fishy Fishv FishV I-'ishv Fishy Fishy I'ishy I'ishv I'ishy Fishv Oily" Fishy Fishy I'ishy Fishy I'lshv Fishy Fishy I'^ishy Fishy I-'ishy Fishy Fishv 2ES 2 E Fishy I'ishy 3ES 3 E Trimethylamine butyrato Trimethvlamine butyrate Fishy I''ishy Fishy Fishy Fishy Fishy 4ES Trimethylamine oleate Fishy 4 E Fishy Fishy I'ishy Fishy Fishy Fishy Fishy Fishy Fishy Fishy Fishy Fishv 5ES Trimethylamine stearate . . . 5 E Trimethylamine stearate 6ES 6E Trimethylamine and soluble fatty acids Trimethylamine and soluble fatty acids Fishy P'ishy 7ES Trimethylamine and insoluble fatty acids Trimethylamine and insoluble f attv acids ... Fishy l''ishy Fishy 7E 8 ES Lactic acid 8E Lactic acid . . 9 ES Butyric acid Fishy Fishy Fishv Fishy Fishy 9E Butvric acid Fishy Fishy F^ishy Fishy Fishy 10 ES Oleic acid 10 E Oleic acid 11 ES Soluble fattv aciils ... .... 11 E Fishy 12 ES Insoluble fatty acids Oily Oily Oily 12 E Insoluble fatty acids . . The Lecithin Content of Butter 119 TABLE 8. Effect on the Flavor of Butter Made from Pasteurized Ripened Cream WITH 0.38 PER Cent Acid, Working Trimethyl amine and Fatty Acids Directly INTO the Butter at the Rate of 40 Parts per Million (S indicates salted butter) Sample Material added Comments by judges No. 1 No. 2 No. 3 No. 4 FS Nothing F Nothing 1 FS Trimethvlamine Oily Oily Fishy Oily Fishy Fishy Fishy Fishy Fishv Oily' Fishy 1 F Triniethvlaruine 2FS Trimethvlamine lactate 2F Trimethjlaminc lactate 3FS Trimethvlamine butjTate Fishy Fishy Fishy Fishy Fishy 3F Trimethvlamine butjTate 4FS Trimethylamine oloate 4F Trimethylamine oleate Oily 5FS Trimethylamine stearate 5F Trimethvlamine stearate 6FS Trimethvlamine and soluble fatty acids .... Fishy 6F Trimethylamine and soluble fatty acids Trimethylamine and insoluble fatty acids. . . Oily Fishy 7FS 7F Trimethylamine and insoluble fattv acids. . . Fishy 8FS 8F Lactic acid 9FS Butyric acid 9F Butyric acid 10 FS Oleic acid 10 F Oleic acid 120 George Cornell Supplee TABLE 9. Effect on the I'Yavor ok Btttkr NLvue from Pasteurized Ripenkd Cream with 0.2S per Cent Acid, of Wohkinc Tkimethylamine and Fatty Acids Directly into the Butter at the Rate of 40 Parts per Million (S indicates salteil butter) Sample GS G IGS IG 2GS 2G 3GS 3G 4GS 4G 5GS 5G 6GS 6G 7GS 7G 8 GS 8G 9GS 9G 10 GS 10 G Material added Notliins- ^'otllin{^. Triniethylaiiiine. TrimethyUyuiiie . Trimethylaniiiu' lactate , Triinethvlaiiiiue lactate . Trimetb ylainine butyrate . Triinethvlaiiiiue l)utyrate . Triiiif'thylainiMe oleate Triinethvlamine oleate Triiricthylainine stearate , Trimethvlaiiiine stearate . Triinetliylainino and soluble fatty .acids . Trinietiivlainine and soUiljle fattv acids. Triinetl!>laniin(! suid insoluble! fatty acids Triinetlivlaniine and insoluiile fattv acids Lactic acid Lactic acid Butyric acid , Butvric acid . Oleic acid Oleic acid . Comments by judges No. 1 I'ishy Fishy l''ishy I'^ishy Oily No. 2 I'ishv iMsliy llshv l-'ishv No. 3 I'^ishy Fishy Fishy Fishy Oily J'ishy Fishy No. 4 Fishy Fishy All of the samples of l)utter I'opi'osoiitod in. tables 3 to 9 inclusive were scored from three to five days aftcn- makiiic;. They were then placed in storage for di{^"er(>nt lenjiths of time and rescoi-ed l)y two or more judges. The results of tliis examination are shown in table 10. In this table are listcMl only those samples showing a fishy flavor by unani- mous opinion of the persons judging them. The Lecithin Content of Butter 121 TABLE 10. Presence of Fishy Flavor after Stor.^ge, in Butter to Which Trimethylamine Had Been Added at the Time of Making Sample Material added Age (days) Comments by judges At time of making After storage 3 CS Trimethylamine butvrate 40 267 20(5 2m 243 243 213 Fishy Fishv, oily Fishy Fishy Fishy, oily Fishy Fishy Fishy Fishy Fishy Fishy Fishy, tallowy Fishy Fishy 3 DS 3 E8 6ES 1 FS Trimethylamine butyrate Trimethylamine butyrate Trimethylamine and soluble fattj' acids. Trimethylamine 3 FS 3 GS Trimethylamine butyrate Trimethylamine butyrate The results obtained from these experiments bring out some very interesting faets. ^^'llil(^ there are several conflieting opinions as to the presence of the; fishy flavor in any particular sample, it is neverthe- less evid(>nt that the greatc^st number of j)ositive comments is found in the sami)les containing trinu^thylamine in one form or another. It will also be noticed that usually the greatest uniformity of such comments is found in the samples containing trimethylamine in unstable form. This is particularly true as to the sam])les to which trimethylamine was added alone, in combination with butyric acid, oi- in combination with the mi.xed soluble fatty acids of butter. These results are in harmony with the volatility, the taste, and \hv odor of tlie compounds in pure state. The lack, in a few instances, of a majority opinion with regard to the samples containing trimethylamine oieate, trimethylamine stearate, and trimethjdamine with tlu> mixed insoluble acids, might b(^ explained on the basis that, since these substances were so extremely unstable, the ti'imethjdamine had nearly all volatilized before the time of scoring. Th(^ gi'(>ater number of positive connnents from the salted- butter is also worthy of note, and, from what is gi>nerally known regarding the occur- rence of the fishy flavor in .•^uch butter, it might tend to strengthen the trimethylamine theory (jf this flavor. Another feature found in this series of experiments is that a greater number of fishy-flavored samples were found where the acidity of the cr(\-un was the lowest. This condi- tion is in harmony with the chemistry involved, for the reason that butter made from low-acid crcnun contains less lactic acid when fresh than is 122 George Cornell Supplee found in butter made from lii<>;li-a('id cream. It is thei'(>fore conceivable that trimethylamine given off by the unstalile compounds which were added could not be taken up by the excess lactic acid in the butter to form the more stable trimethylamin(> lactate. The findinfj; of numerous fishy-flavored sami)les where trimethylamine lactate had been added to low-acid butter mij>;ht be explained on the basis that certain conditions were present, possibly particular enzymes, which were capabk^ of bringing; about the moi-e ra])id dissociation of the trimethylamine lactate, and that due to the absence of sufficient free lactic acid to hold tiie trimethyl- amine it passed into the free state. This explanation is bornc^ out by the fact that in the buttei' from high-acid cream there werc> a small(>r number of sampk'S showing fishiness where the kictat(' was added. It might be stated further that one would naturally expect to find a gicatei- variety of enzymes capable of bringing about the above-described (k'composition in unripened than in i-ipened ci'eam. The evi(k'nc(» obtained from the various sam])les of buttei' to whicii trimethylamine butyrate was nddcd, indicates a striking relationship between this substance and the fish\' flavor. This seems to be true reganUess of the type of butter, and to a ceilain extent regai'dless of the presence of sodium chloride. The (>xtremely \olatile natui'e of this substance and its ciiai'actei'istic odor in pui'e statt^ easily account for the results obtained. The fact tliat Ihi'.i'e were more positive conunents on the butters cojitaining l)utyric acid alone than there were on butters containing the other acids alone, indicates tiiat this substance may be a contributing factoi- in the development of tiie fishy flavor under natural conditions. This feature indicates also that tiie fishy flavoi' may be dnv to a definite balance between a decomposition yielding trimethylamine and one yielding butyric acid, with the conse(iuent foi'mation of trinu^thyl- amine butyrate. This associative action would be entirely ])()ssible in storage butter, judging from what is known icgarding these f(>rmentations. In the foregoing discussion \\\c writer has called attention to certain theoretical possibiliti(>s which miglit correlate the findings with what is generally known regarding the various types of normal and of fishy- flavored ])u1ter. It may be said, however, that the evidence! points strongl}' toward trimethylamine as a conti'ibuting factor in fishy-flavored butter. The amount of this sul)stanc(^ responsible foi' the flavor described as fishy in these ex[)eriments is very small. It would be a hazardous The Lecithix Content of Butter 123 guess to assign a definite quantity, other than to say that in all cases there was less than 85 parts per milhon. QUANTITATI\T] ESTIMATION OF TRIMETHYLAMINE IN FISHY BUTTER The results obtained by working trimethylaniine into normal butter warranted an attempt to isolate this substance from samples of fishy- flavored butter found on the market. One of the first difficulties met with in this conn(H'tion was the lack of a method which would accurately measure the small amounts of trimethylaniine that would be found. The method that was finally woiked out consisted of a combination and modification of the methods of P'olin and Macallum (1912) for ammonia and of Budai [Bauerl (1913) for trimethylamine. The adap- tation of these methods for this pu!i)ose was as follows: The material in which trimethylamine and ammonia were to be deter- mined was concentrated to a volume^ not exceeding 15 cubic centimeters. This material was then i)laced in the pi'oper tube of the Folin apparatus, 10 grams of anhydrous potassium carbonate was added, and the mixture was covered with a thin layer of kerosene to prevent foaming. This mixture was aspirated for fiv(^ hours. The ammonia and trimethylamine set free by the potassium caibonate was collected in N/IO hydrochloric acid. The excess acid was titrated with exactly N/100 alkali, methyl red being used as the indicator. The results of this titration gave the total amount of the mixture of ammonia and trimethylamine. To this titrated mixture 10 cubic centimeters of a neutral 40-per-cent formalde- hyde solution was added. The annnonium chloride present reacted with the formaldehyde to foi-m h3'di-ochloric acid and hexa-methylcne-tetramine according to the equation 6 HCOH and 4 NH.CL = (CH2)r,N,, 4 HCL, and 6 H.O. The hexa-methylenc-tetramine being neutral, the hydrochloric acid liberated from the ammonium chloride was titrated and the ammonia was calculated from this titration. Since the trimethylamine hydro- chloride present was not affect(>d by the formaldehyde, the trimethyl- amine was calculated by differences. Since this is essentially a micro method, the technique involved is of the utmost importance. In all cases the volume of the solution to l)e titrated was kept as nearly con- stant as possible, and the same amount of indicator was used for each titration. A check on the standard acid and alkaU was made with each 124 George Cornell v^upplee determination. ;uk1 llie end-points of all neutralization processes were compared colorinietrically with the standard neutral color. In table il are shown the i-esuUs ()l)tained by this meth(id from mixtures of known amounts of trimelliylamine liydrochloride and annnonium chloride: TABLE 11. Ekficiency of the Modified Micro Method for Estimating Thimethylamine and Ammonia Samplo Ai'tual ;uu()unt of (CH:ji:iX and XHs as liydrocliloride.s ([uiliip:r;uiis' Amount recovered (milligrams) Percentage ()(> ()(j 2 30 1 84 92 92 40 40 270 27C) 27t> XH, (CIIsl.X XHa (CH3)3X NH3 1 (U) (•)(■) 2 30 1 81 92 92 4C) 40 27!') 27() 270 27(i 1 81 92 40 (il! 092 01 51 2 25 1 79 885 911 0.442 0.47(> 295 295 295 () 03, 03 2 278 1 820 035 935 47() 47() .323 297 297 289 1 820 918 1.59 0.040 105 99 70 97 75 97.83 97 28 90.19 99 35 90 09 103 48 100.88 100.88 100.88 99.55 2 99. 55 3 99.04 4 98 91 101.03 101.03 7 103 48 8 103 48 9 117 03 10 11 107.01 107.01 12 13 104 71 98 91 14 15 99 78 99.78 10 17 100.31 114.13 The results shown in talkie II having justified the reliability of the micro method (oi- nieasuriiiL;; sin;dl amounts of trimethylamine, a number of fishy-flavoird samples of butter were subjcH'ted to analysis. The })Utter was thoroly washed in a scpai'atory funnel five times with equal volumes of water aciditicMl witli hydi-ociiloric acid at the rate of 25 cubic centinK>t(M-s of noi-mal acid to the htcv. The wash water was then evapo- rated to a small vohime as (juickly as possible, and the trimethylaminc and ammonia were determined as outlined al)Ove. In table 12 are shown th(^ trimethylaminc and annnonia results obtained from fishy butter appearing in comnKM'ce and procured from widely different locahties. The annnonia results are shown as a matter of The Lecithin Content of Butter 125 interest but they probably have no direct bearino; on this particular problem. Since the expcM'iuients with artificially produced fishy flavor seertied to indicate the importance of acidity, the acid values of the samples are also included. The acidity is expressed as cubic centimeters of N/10 sodium hydroxide used to neutralize 20 grams of butter in boiling alcohol. Trimethylamine and anniionia are expn^ssed in parts per million. TABl.l'' 12. 'rpaMi.TTTTi. \MTX!o AND Ammon'i.\ Content .^nt) .'\cid Content of ' -Samples of Fishy-flavored Butter Sample TriinethylaniinR (parts per million) Ammonia (parts per million) Acid value 1 30 1 35 4 2S S 27 3 14 None None 20 No analysis No analysis No analysis 11.2 14.4 15 2 26.1 18.3 119 20 () 55.0 No analysis No analysis No analysis 3.8 2 5.7 3. . 5 5 4 5 6 . 5 4 6.8 3 5 7 2 7 8 3.8 9 3.8 10 11 3 11 The data submit li'd iii table 12 arc of considerable interest in view of the history of some of the saini)les. On arriving al the laboratory, all of the samiiles, with the cx('e])ti()n of samples o and 11, were scoi'cd as fishy by several judges. These two exceptions wvw samples of butter sent from a distance and were i)r(>sumably scored as fishy when shipp(>d but could not be so judged when received. It will be noted that in both cases there was a higher acid value than in any of th(> other samples, and also that th(> ti'imethylaniin(> c()nt(>nt of sample 5 is low. With these exc(^|)tions the acid value appears to 1)(> relatively constant, as does the trimethylamine content with the excei^tion of samples () and 7, in which no trimethylamine whatever was found. Tlu^ majority of these results would seem to point to a definite trinuMhylamiiie-acid relationship, as referred to elsewhen^ in this pai)er. It is to hr regr(>tted that in three instances the sami)l(^ of butter submitted was too small to warrant analysis. The available^ data, however, point to trimethylamine as one of the causal agents in fisjiy-flavored butter. 126 George Cornell Supplee DEVELOPMENT OF FLSIIY FLA\"OR IN EXPERIMENTAL BUTTERS In oixlcr that the (l(n'('loi:)ni(Mit of th(^ fishy flavor might bo more care- fully s1u(li(Ml. throe s(>rios of oxporiuiontal l)uttors wore made with the olijoct of determining th(> influence of pasteurization, of acidity developed during lipening of the cream, of adding lactic acid to the cream, of inocu- lating butter with lactic-acid bacteria, and of salt. The procedure fol- IowchI in each of tlu^se series consistiMl of making nine different types of butter, salted and unsaltcd, fi-om tlu> same oi-iginal lot of cr(\'un. The different series were made at intervals of from three to six weeks. The description of (\icli of tlu^ diCfercMit types of butter in each of the series, and the designation of the sam|)les, are shown in table i;>: TABLI'i I'A. 1)ksti()N of Tvi'ks ov Bittku IMadk to Sti uy thk Dkvelopme.nt (iK Fishy Fi.a\ou Xame o ' sample Treatment, of cream or butter Salted Pnsalted A series B series C series ('SP>S CSITS A s(^ries B series C scries l!a\v sweet cream I'asteurized sweet cream . ASiJS ASILS BSI!S BSHS ASR ASH BSR BSII CSR CSII tlaw cream ripened with starter rasteurized cream rijiened with starter . . . A tins APRS AiiP.XS BRIiS BPPS iUMJXS (•i!HS Cl'iJS c'tnsxs AlMl ABH A[u:x bi;r BPR BURN CRR CPR Raw cream ripened natu- rallv CRP.N Raw sweet cream with Bnc- icrium lac'is (icidi worked int.o l)utter ASRBS BSRBS CSI{P,S ASI!B iiSHB CSRB Pasteurized creatn witli Unci lit mil liirlis acnii worked into butter ASIIBS BSIIBS rsiiBs ASHB I'.SIIB CS 11 B l?aw sweet cream acidifii'i! with lactic acid ASK'LS BSRLS CSRLS AS1!B ]',SHL CSP.L Pasteurized cream aciditievi with lactic acid ASIILS i5sm,s CSifLS AS;iL liSlIL CSIIL The samples indicated in tabk^ 13 were placed in storage at a tempera- tm'C of 0° F. or lower, and were scored by tliree or foiu' judgc^s at various intervals. The I'csults of these scorings are gi\"eti ifi table 14. Non- characteristic flavors are purposely omitted from tiiis table. The Lecithin Context of Butter 127 TABLE 14. Comments of Judges on the Different Types of Experimental Butters after Various Lengths of Time in Storage at 0° F. Sample ASR . . . ASRS . . ASRS . . ASRS . . BSR BSRS . . BSRS . . CSF>S.. CSRS . . BSHS.. CSH . . . csirs.. ARR... ARRS.. ARRS.. ARRS.. ARRS.. BlUiS.. CIUIS.. CRRS.. BPR... BPR. .. API^S.. APRS.. BPP.S. . BPHS.. BPRS. . CPRS.. ARRNS BRRN . BRRXS BRRXS BRRNS CRRXS ASRB.. ASRBS. ASRBS. BSim.. CSliBS. CSHBS. CSRBS. Acid in cream (per cent) 27 0.27 0.27 27 IS IS O.IS 0.1(52 0.102 0.10 0.111 0.111 O.C.S 0.()S 0.()S O.OS (IS o.r)()7 o.r)(i7 o.r)07 :)2 .J2 O.C)() 0.()() 52 o.r)2 o.r)2 0.5(12 0.()75 57() 0.57() 57!') 57(1 O.dlK) 27 27 27 IS 1(12 0.1(12 0.102 Days in storage 45 45 180 285 45 45 285 45 285 l,*) i:!() 45 45 45 90 130 285 45 (10 285 45 l.'iO 130 285 ■15 130 285 <)0 130 45 45 00 285 00 45 45 130 45 45 90 285 No. 1 Fishy I'ishy Fishv Oily Fishv Metallic ^ietaliic MetaWic MetaUic Metallic' :\letallic lishy ^ietallic' Metaliic Oilv Oily, fishy Fishv Fishv Fishy Fishy Fishv Fishy Oilv, fishy Comments by judges No. Fishy Fishv Oilv Metallic Aletaliic ( )ily Metallic Meoallic Metallic Oilv Fishy MetalUc Metallic Oilv I-4shv I'i.shv Fishy Fishy Oily * No. 3 I'ishv Oily Oily 14shy Metallic Metallic Oilv, fishv Mx'tallic Fishy ^ietaliic MetalHc Oily M(!tallic Oilv Fishv Oily Oily, fishy I'4shv Oily,' fishy Oilv Oily Metallic No. 4 Fishy Fishv Fishy Fishv Oily 128 George Cornell Supplee TABLE 14 icondudi-d) Sample Acid in creiini (per cent 1 Days in storage Comments by judges No. 1 No. 2 No. 3 No. 4 ASHBS 189 0.1 S9 141 141 (i^S (i:?s (l.'iS (•>;^s (Km (Km ()()().■) 472 472 (;'.)() (i'.K) -^\)\ ' ;■■;•. 11 -)04 n{)[ 501 130 28.-) «)() 4.-) 00 i:50 2s:) 4.-, 00 2sr, 285 130 285 15 90 45 45 285 Fisliy M eta' lie' Metallic I'ishy Oily ASHBS Fishy Metallic Metallic CSHBS Metallic CSHBS ASRLS ASRLS Fishy Fishy Fisliy ASRLS I'^ishv I'^ishv I'ishv Fishv Fishv Metallic Metallic Metallic I-ishv Oilv I'isliv ASRLS BSRLS iMshy r'ishy B.SRLS BSRLS . Fishy Metallic CSRLS CSRLS Fishy Metallic Fishy Oily ASIILS ASHLR BSIILS I'ishy BSHLS CSHL I'ishy Fishy CSIILS CSIILS I'-isliv Oily In ('onsidcrliii;- the r('.'^ul!.-< from the (liffcnMit tyix's of c>x])('riin('ntal butler, it is evident that there is consideral)!'' diversity of opinion among the judges as to the ])resenee or i\\c absence of th(^ fisliy flavor in certain sanij)les. It is also (>vident I hat there is some i-elationsliip between the metallic, oily, and fishy flavoi-s, jjarticnlarly wIkmi these flavors are not sufficiently pronounced to l")e distinctive as was the case in those samples. This would secnn to indicate^ that there are jMOssibly certain fundamental conditions which are common to lh(> development of (^ich of these flavors. Even tho there is ditf(>rence of opinion as to the presence of the char- acteristic flavors, certain conclusions may be diawn from these experi- ments. Probably one of the most significant is the presence of the fishy or the metallic flavor in thc^ salted iMitters. Of a total of 105 character- istic comments, 93 are found in the samples containing salt. Another conclusion which may be drawn from tlie relative agreeuKnit of the judges, is that tlie fishy fiavor apix'ars oftener in the butter made from The Lectthix Content of Butter 129 hisli-acid croam tluui in thai from low -acid cream, there being little difT(M-ence whether the acid was developed l)v the use of starter, by ripen- inmical change in which su('h agencies do not play a i)ai-t. While it is evident that acid plays an important role in the developnuMit of the fishy flavor, it is equally clear that there are ()th(>r important contributing factors. Just what these factoi'S are, is unknown. Tlu^ variable results obtained from the same tyjx> of I)utter in the different s(M-ies would indicate* that the original cream or milk possessenl the unknown factors which in the presence of lactic acid deteiinincMl the dev(>lopment of the flavor. Fi'om the fact that pasteurization tends to re(luce the occurn^ice of tiie fishy flavor, it is quite probal)le that these; agencies arc bacterial enzym(>s which are only partially inactivated by heat ; oi- it may even be possible that certain microorganisms wliich nic incoi-poi'ated in the butt(M' fi'om the cream, either in a living or in a dead condition, could on autolysis liberate the enzymes capal)le of supi)!ying the detcM'inining factor. It may also be added that pasteurization may kill certa.in (Mizymes and not others, the particular ones that are inifiortant l)eing among those killed. These contentions are fui'thei' supported by the fact that in the butters made from raw sweet cream tluM-e is a sugge^stion of fishiness after the first storage period which is not found after the longer periods, the dis- appearance or lack of furthei- dc\'(dopmeiit of the flavor l)eing due to the absence of tlu* pr()[)(M- acid condition. It is clear that large numbers of Bdcteriuni lacfis ac/'fJi added dii'CH'tly to l)Utter without their usual accom- panying by-products are not the cause of any characteristic change in flavor. VAUIATIOX IX ACID VALUE OF EXI'ERIMEXTAL BUTTERS The importanc(> of acidity in the manifestation of the fishy flavor by trimethylamine, and the i-idativ(dy constant acid value of the miscel- laneous samples of fishy buttei- found on the market, emphasized the importance of studying this factor in the experimental butters described 130 George Cornell Supplee above. The variation in acid value of the different types of butter in each of the three series is shown in tables 15, IG, and 17. Results are expressed as cubic centini(4ers of N 10 alkali necessary to neutralize 20 grams of l)utler in boilin"; neutral alcohol. TABLE 15. Variation in Acid Value of Experimental Butters of A Series AFTER Various Storage Periods Sample Acid value after various storage periods 4.3 days 86 days 128 days 310 days ASR 8.8 7.4 5.8 5.7 8 2 8.7 7.7 8.1 8.9 8,8 9 7 7.4 5 7 5,8 8 9 8 9 7 7 4 10 5 8 8.6 5 9 8.5 8 4 7.9 7.S 8 () 8 4 10 4 7 , 7 8 6 5 9 8,8 8.9 7 1 6 5 10 8 8,2 10 5,7 9 ;^, 8,6 8.2 8 9 2 8.7 11.2 7,8 9.5 5.8 8.8 9.0 7.2 7.1 13 ASKS 10 ASH 12 1 ASUS . 7 2 ARl! . . 10 3 AKRS 10 3 APR 9 8 APRS 9 6 ARRX 10 5 ARRNS 10 5 ASRB 15 ASRBS 9.8 ASHB ASHBS 14 3 9 7 ASRL 10.3 ASRLS 10 ASHL 8.0 ASHLS 8.5 The (lata presented in tabl(>s 15, IG, and 17 show many interesting featuivs, some of whicli are wortiiy of discussion in connection with this pi'cbl(>m. It may be stated in the Ix'oinning that the variations in acid value of the different types of butter i)oint to biological agencies as the cause of those variations. The lower acid value obtained in nearly all instances from salted l)utter indicates a pi'cscrNative action by the salt, a function wliich is well known. The greatest increase in acidity is The Lecithin Content of Butter 131 shown in the butter made from raw sweet cream. It is interesting to note that very few of the samples were scored as fishy. When such a condition was suggested, it is to be noted that it occurred after the first storage period, when the acid vahie was lowest. TABLE IG. Variation in Acid Value of Experimental Butters of B Series AFTER Various Storage Periods Sample Acid value after v-arious stora.!?e periods 20 days 88 days 126 days 286 days BSR BSRS 13 9 9.7 7 3 G 3 10 1 10 1 10 1 9.8 11.3 11 2 13 6 10. G 8 1 G.8 11.5 11 5 9 4 9.5 14. G 9 5 9 G 5 10 2 10.1 10 1 10 1 11 7 11.7 13 5 10 8 8 4 G.G 11 5 11.1 9.4 9.G 15 2 10 10 6 G.S 10 8 10 4 10 9 G 12 G 11 8 14 3 10 8 9 6.8 11.6 11.6 9.5 9 6 17.5 10 6 BSH 13 4 BSHS 7 6 BRR ERRS 11.9 11 3 BPR 11 2 BPRS 11 BRRN BRRNB 14 3 13 8 BSRB 15 4 BSRBS 12 BSHB 9 2 BSHBS 8 BSRL 12 9 BSRLS 12 6 BSHL 11 BSHLS 10 1 With reference to the samples from pasteurized cream as compared with those fiom raw sweet cream, it will l)c^ noticed that pasteurization has tended to cause a lowering of the acid vahu^ l)ut has not entirely prevented its gradual increase. This would be in accord with possi- bilities already stated regarding bacltMial etizynies. The data show also a retarding action exerted by the acid originally in the cream. This is evident in the butter made from both raw and pasteurized cream ripened 132 (^tEokgk Cornell Supplee with startcM', from i'jiw ci-cain ri})('iu'(l iialnriilly, and from both raw and pastciii'izcd croam (o which lactic acid has hc(ai added. In coxaparing the I'csuhs from these samples it v.iil be obsei'ved thai cream ripened naturally shows the greatt^si increase in acid value, raw (a'eam ripened TABLE 17. N'akiation iv Acu) Value of E.kperimental Hutteus of T Series AFTER \'ar!')i;.s St >'i\';-j Peuiods Siirnpk! CSR . . . CSRH . . CSH . . . CSHS.. CRR. . . CRRS. . CPR .. CPRS. . CRRN . CRRNS CSRB . CHRB8. C8HB . CSHBS. CSP.L. CSRLS. CSHL . CSIILS. Acid viilue after A'ariou.s storage periods G Ja\"s 4.4 4.4 7.() 7.5 G.t) 6.5 S.5 S.2 7.S 6.1) 4,4 4.4 s:.i S.4 5.8 4S dav.- 00 days 10.4 7.9 11.0 8.0 7.G 4.7 11 4.4 S . 2 10 4 8.4 7.5 7.0 8.9 7.0 9.5 9.0 10.0 9.1 10.2 7.4 11.8 7.9 0.5 4.4 7.7 4.4 8.1 8.5 8.0 8.4 GO 5 . 7 5.9 5.G 272 davs 13 5 9 7 12 5 2 12 8 10 10 8 2 12 9 11 3 13 5 9 7 1 9.8 9 5 G.7 6.3 with starter a little less in»Te;ise, pasteurized cream ripened with starter a still less increase, an.d cream to which lactic acid has been added the least incr(>ase of any in the grouj). The low acid \'alue caused by the addition of lactic acid mia;hlt i)Ossib]y be ex])lained in one of two ways: either the addition of the acid in ])ure form has tended to inactivate the enz^^mes, oi- the lactic aciTl rc^aintn! in the butter lias been changed to butyric acid during stoi-age — which is entirely possibl(> by enzymatic The IjEcithin Content of Bittter 133 action. If such a chan^c^ as tli(> latter did take jilaee, a lowonn"- of the acid valiu^ would l)o manif(>sted IxH-ausc of the formation of a weaker acid which pro))al)ly has resulted from the splittino; and condensation of two parts of the stronger' lactic acid. The possibilit}'- of butyric acid being formed in this way might })e supportcnl l)y the fact that the other samples from high-acid cream showc^l a lower acid value than those from sweet cream but had a high(>r ^'alue than those to which pure lactic acid was added. Even tho this change^ did take place, however, it is improbable that it coukl (nitin^y accomit foi- the low value indicated. It is more probable that the lactic acid acts as an inhibiting agent. A study of the tabl(\s will show that the incrc^ase in acid value of the sour- cream butters seems to be gi-eater when the amount of i)ure lactic acid is lowest in the cream. In using the phi-ase " ])ure lactic acid," rc^fei-ence is made to tliat which was added and also to that developed by bacteria, it being logical to assunK> that tli(> gr(>atest amount so devc^loped is found in pasteurized cr(\'un ripcnied with starter and the least amount in the raw cream rii)ened naturally. The condition mentioned above also sup- ports the theoiT of devitalized enzymi^s, alt ho it is moi-e difhcult to explain why approximately the same degree of connnercial lactic acid has a more marked effect than the acid produciMJ by bacteria. It would seem that the structure of the particular lactic acids involved produced different results in this r(>sp(>ct, or that the other acids ])roduccHl by the bacteria are less.inhil)itive than the lactic. Regardless of what the (explanation for the vai'iations may be, the data seem to indicate that there may be a r(>Iationship bc^twecMi the acid value and the fishy flavor, not so nuicli by a constant condition as by a pi'oper balanc(e betwcHMi the J)rogr(>ssi^■e develo])nient of the acid value and some other contributory caus(>. The most favorable condition would seem to be a very gradual increase' in acid value, and one that would be in prop(>r harmou}^ and relationship to some other important and transient factor. If these views can to any degi'ee .s(»rv(> as a basis of (explanation, it is comparatively easy to see how an iiuproper balance of any one of the conditions would determin(> the pr(esence oi- the absence of the fishy rlavor. It might also be c()nc(Mval)k> that the int(Misily of th(> ti'ue flavor would be in invers(e proportion to the degre(e in which thes(> factors wei'e out of ('(luilibrium. Such a coiic(e):)tion could explain the occurrence and the tlisai)i)earanc(^ of the flavor in the same sampk; of butter at different 134 George Cornell Sitpplee times, why tlie fishy, the inclnlhc, and th(> oily flavors schmii to bo closely related, and possibly \vh\' triniethylamine can Ix^ detected in some fishy butters and not in others. trlmethylamixe and a.mmoxlv coxtext of experimental butters The micro method already described was used to ascertain the ti'i- methylamine and annnonia content in the expia'imental butters after difl'erent lenjitlis of \i\uv in stoia X'one 13 .") Xone Xone Xone Xone Xone X'one Xone 3() 17 2 2:; ;', 2'.; 2 Id S 31 20 7 2.') :; 22 4 IS ' 22 1 23 1 17 ;•. 14 2 in :; 1(1.3 Xone Xone Xone Xone X(>u(! 0.4 Noni^ X()n(^ X( )ne Xone Xone Xone 11 S 1.") 2 (; X'one Xone 17 6 ASI{S 8 4 .\SI1. 10 6 ASUS G 8 \UU 30 S \iM!S 18 2 Al'l! 24 4 \P1IS 24 arkn 25 AUUXS 6 8 asi;h 24 4 ASIiBS 12 8 ASIIB . 10 8 ASRL 9.4 ASIILS 10.7 ASIIL \SHLS 12 2 8 8 The I'csults shown in tables IS, 1<1, and 20, alt ho ei'ratic, are of impor- tance as indicatinii' the wiriations in decomposition in the same and in diffcM'ent lots of cre.am, and fui'ther emphasiz(> ili(> complexity of a problem of this nature. The snnie o;en(a-al iv^sults with respect to enzymatic ;icti\ity in sailed and in unsalted l)Utter from law, pastem'ized, and iip(>ned ci'cam ai'c found here as were found in coimection with the acid values of the same butters. Alt ho th(> triniethylamine results arc some- what discordant, a tendency is shown for the presence of this substance The Lecithin Content of Butter 135 to harmonize with tli(> samplps scored as fishy, metalKc, or oily. Of 21 samples in which trimethylamine was found, 15 were assigned one of the characteristic flavors by one or more of the judges at some time during the storage period. In 5 of the; remaining instances, it is to be noted that, while trimethylamine was found in the same type of butter of the same series, its presence did not harmonize with the characteristic flavor in the salted or the vmsalled sampl(\ On the other hand, there were 10 samples of different types of l)u(ter whicli wer(> indicated as having a characteristic flavor by one or more of the judges at some time during the storage period, in which trimethylamine could not be detected. TABLE 19. Tkimethylamink and Ammonia Content of Experimental Butters in B Series after 323 Days in Storage Sample (CIl3)3N NH3 BSR None None 35 2 BSRS . IS BSH BSHS 53 None 8.2 10 G 9.4 4.7 None 35.4 None 5.9 None None None None None 10 2 BIU? 32 8 BI^RS 21.0 BPR 23.8 BPllS 25 BRRN . . 27 BRRNS 32 2 BSRB 29 4 BSRBS... 20 BSHB 23.0 BSHBS 10.4 BSRL 23.0 BSRT.S 17.0 BSHL. . 7.4 BSHLS 11.4 While these results aiv not absoluti^ly coiiclusi\-(\ tii(M'e is ncn^ertheless an indicat'ion that trimethylamine may b(> one of the contributing factors in the develoj^ment of tlic tru(^ fisiu' flavor. It has been shown that this substance is capable of [)roducing a flavor described by butter judges as fishy, this l)eing i)articulai'ly true in I he pres(>n('e of butyric acid. Furthermore, it has been shown that trim(>thyL-unine may be present in tishy-flavored butt(M-. Therefoi'(> it would not ai)pear to be beyond the 136 George Cornell Supplee realm of possibility that the results shown in this paper point to a definite trimethylaniine and acid relationship as l)einji; the cause of that flavor in butter which resembles the fla^-or of herrinj;' or mackerel brine, and that non-typical flavors n^stunbling other fish pi-oducts, or the metallic TABLE 20. TniMKTIlYLAMINK AND AmMONMA CoNTKNT of Kxi'KKIMENTAL BuTTERS IN C Skuiks afteu Variocs Stokaoe Periods Sample After 00 (!a>-s After 310 days fCHalaX XII, (CH3)3N NH3 CS1{ Xonc Xoiu' Xonc Xonc Xonc Xonc Xoiu^ X^oiic Xoiu^ XolU! Xonc Xonc Xonc Xoiu- Xonc Xon CSRBS IS 6 CSHB 31 G CSIIIW 15.4 CSRL CSULS 14.8 10.8 CSHL CSIILS 20. G and the oily flavor, ma>- !)>■ (hie to an unl);ilancin,<; of this r(^lationship, the 0(;currenc(! of these; fiax'ofs beiiio; {\\\c to factoi's in which the presence of trimethylaniine in detectable amounts is in no wa>' contributory. bacteriolocjical studies In an effort to corn^late tlu^ preceding observations with tlu> biological aspects of the j^ro'olem, certain bacteriological studies were carried out. Thest^ included a bactei'ial analysis of fishy- and non-fishy-fiavored butters, and inoculations with pure and mixed cultures into chohne, lecithin, buttei'fat, and ci-eam foi- th(> purpose of finding, if possible, an organism or a group of organisms whicli in some way n)ight contribute to the develop- ment of the fishy flav'or. The Lecithin Content of Butter 137 Bacterial analysis Bacteriological (examinations of fishy- and of normal-flavored butter from various sources seemed to show no characteristic differences in flora, neither were? the quantitative I'esults consistent. F.ven tho the samples examined did not appear to possess marked differences in flora, cultures of the predominating type were isolated from the fishy samples for the purpose of determining a condition under which they might contribute to the characteristic; flavor. The types of l)acteria found included a number of acid-iM'oducing varieties, both coccus and rod forms. Among the species commonly found were Micrococcus lactis acidi, Mic. lactis al- bidus, Bacterium lactis brcins, Bad. aerogenes, and Bact. lactis flocculus. The bacteria content of ccu'tain samples of fishy- and of non-fishy- flavored butter is given in table 21: TABLE 21. Xu.MBEK or Bacteria Found in Various Samples of Fishy- Wl) X(iN-l'ISHY-FLAVORKl) BuTTER Sample Character of flavor Bacteria per gram 1 Strong Fishy I'ishv Fishy Oilv Fishy Fishy Oily Strong TMshy Fislu^ Strong l opportunity for growth and consequent ]iroduction of bj'-products wliich they would have if allowed to grow in the milk or the cream before it is made into butter. There seems to he good evidence that there is a i-apid dying-off of tlie bactei-ia in butter after the first few days of storage. Furthermore, it is well known that low storage temperatures do not entirely prevent enzymatic activity. Rogers (1909) shows that, whili^ low iemi)eratures delay the development of the fishy flavor, they do not entirely prevent it. In accordance with these facts it is conceivable tliat certain enzynu^s winch would be produced by the growth of organisms in the cream, and carried into the butter, would there continue their activit,y, witli the consequent manifestation of certain changes in flavor. Ivcsults obtained in the present investigation indicate also the importance^ of a definite acid relationship. With these factors in mind, tiie inoculation experiments imdertaken in connection with this protilem were carried out in a m:uiner that would alio\v for the manifestation of the possibilities indicated. Nine organisms, all of which were isolated froin samples of fishy butter and Bacteriiun ichlhi/oxun'ux — which Hammer (1917) found would produce the fishy flavor in milk — v.ei-e used in these exjx'rinients. The same original lot of cream w-.is divided into six parts, and (vich of these parts was furtiier divided into ten parts, each of which was inoculated with a specific organism. Different methods of handling the six groups of ten inoculations each were carried out in such a way that the (>ffect of acid in conjunction with the specific organism could b(> determined. Adequate cliecks were made from uninoculated cream. Pasteui'ized sweet cream was used as tlie basis for all inoculations. When the crcvmi was neutralized the acidity was reduced to 0.18 per cent. All samples were made up both salted and unsalted, and were scoi'ed aftci' 2:51 days in storage^ at a temperature of 0° F. or lovicr. The results of th(» expei'iment as regards salted butter are given in table 22. It is charactei'istic of the r(\^ults of this experinunit that none of the .samples of unsalted l)utt('r showed any of tlie characteristic flavors and ai-e therefoi-e not "included in tnble 22. It is believed that the results shown in the table clenrly confirm the o|)inion that l)iological agencies, particiflarly bacterial enzymes, are responsil)l(i to a great degree for the flavors indicated; also, that the fundamental condition necessary for the The Lecithin Content of Butter 139 TABLE 22. Effect on the Flavor of Salted Butte::i, of Inoculating Cream with Specific Organisms under Different Conditions Sample Treatment of cream Acidity of cream* Comments l)y judges 1( + ) 2 ( . + ,) No. 1 No. 2 Xo. 3 No. 4 1 Raw sweet 0.18 0.13 0.3S 2(; 2 3 Pasteurized sweet, ripened Pasteurized swe3t, ripened and neutralized Pasteurized sweet, acidi- fied with lactic acid. . . . ().3s 4 5 o.rvi O.ls o.i;] 0.13 13 Fishy I'isliy Oily Fishy 6 Pasteurizetl sweet, acidi- fied with lactic acid and neutralized 0..-)! D BI Pasteurized sweet, in- oculated with sj)ecifie organisms and churned at once Pasteurized sweet, in- oculated with .■^pecific organisms and held 21 hours before churning Pasteurized sweet, in- oculated with specific organisms, held 21 hours, and neutralizeil before churning D 100 Metallic D 23 D 12 0.13 0.13 13 Tallow^' Metallic Tallowy Oily D 18 D 19 D 8 (> K' Oily D 11 13 13 D 21 D 10 E BI 4.- E 100 41 Fishv Metallic Tallowy E 23 0.4.-, 0.11 0.41 0.43 0.45 0.43 0.4(3 0.43 0.43 0.39 0.41 0.43 4.-! Fishy Oily E 12 E 18 E 19 43 E 8 4(; E 11 43 Oily Oily Fishy E 21 0.43 39 E 10 K BI O.IS O.IS IS Metailic I'ishy Fishy K 100 K 23 Oily K 12 O.IS O.IS IS Butyric Butyric K IS K 19 K 8 IS K 11 0.18 O.IS 0.18 Tallowj- K 21 KIO Fishy *1 ( -(-) = acidity at time of neutralizinfi 2 ( -i-) = acidity at time of cliuniing. 140 Geokgk Cornell Supplee TABLE 22 (,<:oncludcd) Sample Treatment of cream Aciditj- of cream* Comments by judges 1( + ) 2 (+) Xo. 1 Xo. 2 Xo. 3 No. 4 NBI Pasteurized sweet, in- oculated with si)ecifi(' organisms, held 21 hours, neutralized, and ripened Pasteurized sweet, in- oculated with specific organisms, held 2 1 hours, neutralize! 1, ripened, and again neu- tralized Pasteurized sweet, in- oculated with specific 0.15 0.41 0.41 0.43 4.- 0.43 47 43 0.43 0.39 0.3(i 0.2(i 0.38 31 33 0.32 . 29 0.30 0.32 0.29 0.32 0.3(1 N 100 N23 0.3.-) 0.37 0.38 N 12 N 18 N 19 0..3(> 33 N8 N 11 29 N21 39 N 10 0.39 R BI K 100 O.IS 1,S I'ishy Oily Fishy Fishy R 23 IS IS IS Oily R 12 Butyric R IS R 1!) IS Rancid R8 0.18 Fishv R 11 0.18 Fishy K21 0.18 R 10 IS 2!t 4' XBI X 100 ^h'tallic Metallic Tallowy I'Mshv Oily Metallic Strong X23 X 12 organisms, held 21 hours, and ripened !() 3 development of tht>S(^ flavors, wliieh aw potentially po.ssible from the specific bacteria or en/ymes. In this experiment it is ap|)r(;ciated that tht^ results are obtained by an associative action with the organisms in tlic starter and those surviving pasteurization; this fact, iiowever, do(>s not depreciate the specificity of the jiarticular organisms tliat were inoculated. In reviewing the data from this experi- The Lecitiiix Content of Butter 141 ment, it is interestinc; to note that the sweet pasteurized cream to which lactic acid had been added and which was iininoculated, developed the fishy flavor. The description of the flavor in Die same sample of butter by diffen>nt judges ao;ain calls attention to tlu^ fact that there seem to be some conditions common to the fisliy, the metallic, and the oily flavor. The most consistent comuKMits fi'om specific organisms seem to be from culturc^s BI and 23. Tlu; former- is Bad. ichthyosmius, which was obtained from Hammer; tlie latt(M' is an organism isolated from raw- ripened-cream butt(u- which develoixnl the fishy flavor after two months and retained it for nearly twenty months. Butter samples 1'] 23 and X BI were analyzed for trimethylamine and ammonia. None; of the former substance was found. Sample X BI showed 35.2 parts per million of ammonia, and sample E 23 showed 17 parts per million. loxcxEVIty of bacterium ichthyosmius in butter Preliminary expei'iments with Bacterium ichthyosmius indicated that this oi'ganism might prcxhice the fishy flavor in butter. It seemed desirable, therefore, to determine its longevity in butter made from the inoculated cream. Il(>sults of tlu^ (juantitative d(>terminations of the bacteria in salt(>d butter containing this organism are sliown in talkie 23; results are given only for those sampl(>s to whicli a cliaracteristic flavor was assigned. TABLE 23. liACTERiA Content of Salted Butter :\Iade from Cream Inoculated with Bacterium ichthyo.smius and Stored at a Temperature of 0° I'\ or, T.nAVKii* Ave of sample (davs) Xumbor of bacteria pci ^ .,;...;. KBI RBI X BI 23,400,000 11,400,000 10,4.50,000 8,500,000 6,300,000 1,800.000 1,1.50,000 890,000 750,000 10,400,000 10,100,000 8,300,000 6,700,000 6,350,000 4,000,000 1,3.10,000 21 34,2.50,000 35 28,000,000 49 17,400,000 G3 10,1.50,000 78. . . 11,. 500. 000 91. 11. 201). 000 105 6,400 000 1!') 5,400,000 134 740.000 3.55.000 37.500 148 169 4.600,000 * These results were furnished liy .J. 1 . i usiok. 142 George Cornell Supplee Tlu^ decrease in bacteria content of tlie salted butter containing Bad. ichlhijosniias shows that, ahho this organism may contribute to the dcn'elopment of the fishv' flavor in butter, it does not do so by active muUiphcation in that meihum. FURTHER Snn)II':rt WITH BACTERIUM ICHTHYOSMIUS Tlie resuhs obtained with Baden'uDi /chtlii/osinius seemed to warrant a further study of its n^lationship to the fishy flavor in butter. The following experiment was carried out wil!i the ])urpose of determining the conditions in butter under whi(vh th(^ d(>velopment of the flavor could be accelerated. Pasteurized sweet cream was inoculated with this organism and held for two days at room temperatui-(\ The butter made from this cream was divided into twelve parts, and to each of these parts a different substaric(> was added. The cream at the time of churning contained 0.23 per cent acid. The treatment of this butter, ;md the time of occui'rence of th(> tisliy flavor as determined by two or moi-e judges, aiv shown in t;il)l(' 21: TABIjE 21. KisHY l*'LAV(Hi AS Dev'Klopku in Butteh W'mrn Was Made from Cream IxorUI,ATED WITH IiA("rEHILIM ICHTHYOSMIUS AND TO WhHH V.ARIOUS SUBSTANCES Were Added Sani])l(' Suhstancf added to butttT Days in stora'ie .52 94 136 17.5 328 BI 1 Nnthinn Fishy Fishy Not s(!ored BI 2 lierkl'eldt filtrate from milk culture of Bdd. icbtluiD.'itiiius Choliue, 0.0118 per cent Calcium caseiuate BI 'A Oily BI 4 BI.-) BI (') Lactic acid, 0.117 per cent Lecithin from butter Fishy Fishy Fishy No , scored Fi.shv BI 7 Lactic acitl and choliiu; Bdct. Inctis (icid> starter I'ishy BI S Slightly fishy BI !» Berkfeldt filtrate aud caseiuate Berkfeldt filtrate and choline. BI 10 BI 11 Berkfeldt filtrate, lactic acid, and choline EI 12 Berkfeldt filtrate made alkaline Oily Fishy No J scored AH of tlie butters intlicated in table 24 possessed a very disagrei^able flaN'or Mild odo!- wIkmi fresh l)ut they seemed to improve in quality during storage. The deNclopment of the fishy tiavor in certain samples shows some v?]y interesting features. In reviewing the results of this experi- The Lecitiiix Context of Butter 143 ment it must he homo in mind that tho cream fi'om which they were made contained the prochicts of two days jirowth of Bdd. ichthijosmius. The development of the fishy flavor in the sami)le to which nothing was added, is thxM-efore significant. The earlier occurrence of tlie flavor in the sampk^ to which lactic acid was added is significant in that it confirms certain observations alr(>ady noted. The development of the flavor in the sample containing tho. alkaline Herkfeldt filtrate seems to be about sinuiltaneous with its development in the sample to which nothing was ad alone oi- in combination with other sub- stances, might api)ear to be contradict oiy to the (^nzyniatic idea pre- viously expressed. It, is beli(>ved, however, that this is more than offset by the other data, which i)oint to \]\v necessity of a definite set of con- ditions that must l)e met in ordei- lo produce the flavor. Such being the case, the absence of the flavor when the filtrate was added may be explain(>d on the basis that the proper e(iiiilibi-iiun had b(>en disturbed. Th(^ final occurrence of fishiness in the sample cotitaining lecithin is of importance as indicating that this may be the mother substance of the material causing the flavor. Other scattering results do not merit particular discussion at this time. The trimethylamine and anunonia content of the samples shown in table 24, and their acid value, arc; giv(!n in tabk; 25: TABLE 25. TuiMirniYhAMrxK .v\*i) Ammonia Content and Acid Value of Butteu S.\MPLEs Which Were .\Iai>e from Cream Inoculated with Bacterium ichthyosmujs AND to Which Other Substances Were Subsequently Added Sample Age of sjiiiiple (daysj (CH3)3N (parts per milliou) xib (parts per niillionj Acid value BI 1 13G 328 328 328 94 328 328 328 328 328 328 136 7.2 0.4 5.8 4.6 4.7 9.4 8,2 4 (■) ") , S ""us 20 7 30 2 28 4 26.8 12 7 25.4 24 22 4 25^8 "l7's 8.7 BI 2 9.5 BI 3 8 5 BI 4 7.8 BI 5 8.2 BI (•) 8.0 BI 7 8.9 BI S 8.5 BI <» BI 10 9.2 BI 11 . . 9.4 p: 12 9.3 144 George Cornell Supplee The relatively constant trimethylaniine results shown in table 25 indicate^ strongly that this substance has been produced in the cream by F>a d . ichth yosm ius . trimethylamine and am.moxlv prodfctiox by bacterium ichthyosmius i.\ milk axd cream In order to detorniine tlie triniethyianiine production by Bacterium ichthno.smius in milk and cream, oO-cubic-centim(>ter (juantities of these substances, sterihztMl, \ver(> inoculated with the organism alone and in c()ml)ination with a lactic-acid stai'tca*. Tlu^ inoculations were held for foiiy hours at 'M)° ('., and the triin(>thvlaniine and anunonia were then determined in 2()-cuhic-ccntimeter ciuantities. TIk^ results of these, detei'minations are shown in tal)le 2f): TABLE 20. Amotn't ok 'I'ltiMKinYLAMiNE AXD Ammonia Produced in Rkimmilk and i\ CuKAM BY Bacterium ichthyosmius Inoculalion liaclcrluni IchifiiidsrHiit^t and starter Bad. ichthi/osmiit.s and starter . . . . Bart, irhlhui^sniius B(ul. ichlhijosDiius Mateiial inocHiiated Skiininilk Oeam . . , Skiminiliv C'reaiii . . (CHslsN (parts per milliou) Xone 204.0 04.4 74.7 NH3 (parts per million) 84 88 125 78 The results presented in tahle 2(5 are of gr(>at interest as showing beyond a doubt that the fishy flnvor pi'oduced in milk and cream by Bdcl. iclithno.stniiis is due to t fimethylamine. This being the case, it is obvious that this substanc(> would \)v carried into the butter, and ther(% under proper conditions which have ah'i'ady been pointecl »out , be r(>sponsil)le for the characteristic flavoi- in that material W'itii r(>spect to the pro- duction of trimethylamine in cream and in milk by this organism, it is desiral)le to again call attention to the ol)S(^ivations of the authoi', in which the evolution of a fishy flavor was noted on the addition of alkali to sweet cream. Thes(> results ai'e of further importance in that the cream inoculatcnl with start(>r and Bact. i('hth!!Osmh(s contained a greatei' amount of trimethylaniine^ than did tlie cream inoculated with the organism alone. This in(Ucatc>s that an aciti condition is most favorable for this particular The Lkcithin Content of Buto:r 145 formentation, whicli would be in hai'mony with tho idea that lecithin fvn-nishes the souivf^ of tii(> triniethylamine ])rodii('(Hl by the organism. The results are support(Ml also by the fact that lecithin is largely asso- ciated with th(> fat, and that according to Hannnarsten and Hcnlin (1915) lecithin is decomposed by dilute acids. Such being the case, it is readily seen that this fermentation brought about by Bod. ichfhyosmius would be greatly enhanced by the presence of acid. Th(> ])resence of triniethylamine in skimmilk inoculated with the oi-ganism alone might be explained on the basis that the oi-ganism was able to produce this substance from [)i'oteins as well as from lecithin. Certain data not included in this paper, however, indicate that therc^ is a certain amount of lecithin present in skimmilk. Just why there is no ti'imethylamine in skimmilk inoculated with Ihe starter and th(> organism, is more difficult to explain. It may be that tlu^ gnvitei- acidity in thc> skinnnilk has inhibited the particular factor responsible for tiimethylamin(> production. PRODUCTION OF THIMETHVLAMIXE FROM LECITIIIN AND CHOLINE BY BACTEItlAL ACTION In order to determine, if jwssible, whether certain organisms found in milk and in butter were capable of decoinjwsing k>cithin or chohne into triniethylamine, a series of inoculation experiments were carried out. Lecithin alone in 0.3 per cent concentration, and in the presence of lactic acid and salt, was inoculated with a number of organisms, some of wliich were obtained from milk, souh^ from fishy loutter, and some from decomposed egg yolk which had d(>V(>loi)ed th(> fishy Havor. The following known species w{M-e also used: Haderiuin lad is ncidi, Bad. oerogenes, Bacillus procligiosus, B. ))roteus, Baden'iun ichlfii/osmius, Pscudomonas liquejaciens fluoresce ns, Oidium lad/x. All organisms wcw inoculated singly and in various combinations, and the cultures were held at 20° C. for approximately n'uw months. At the end of that timci the cultures were tested for the presence of trimethylamine by berating with alkali. Negative results were obtained from all of the lecithin inoculations tested. Unfortunately, many of the; culturc^s werc^ contaminated with mold, and, since the results could not b(^ considered trustworthy, the}^ were discarded. The same series of exfx'riments was repeated by inoculating 0.1-per- cent choline alone and in the presence of lactic acid and salt. These cultures were held under the same conditions as were the U'cithin inocula- 146 George Cornell SiprLEE tions. Trimoth,ylainin(' \v;is found A\h('i-(> Baderiuin icJilhi/os/nius was inoculated alone, in coinbinatiou witli Oidium lacti.'i, and with Bacterium ladis acidi. The pi'esence of salt did not seem to ])rev{nit the productiori of trimethylaniine. Two organisms wliich were isolated from milk gave a pronounced test from the choline inoculation, but gave negative results in the presence of lactic acid and sjilt ; Bdcillus produjiosus gave a positive reaction from the ch()lin(> alonc^; and Bavierium acrogcnes gave a non- typical test under the same conditions, as did Psciidomonas liquefaciens fluaresccns. All other iiioculatif)ns gave negative results. It would appear fi'om tlie I'esults of th(> inoculation experiments that since trimethylaniine is pi'oduced from chohne by Bacterium ichthyosmius and cei'tain organisms found in milk, it is (juile possibk^ that the fishy flavor and odor found in miUv and in butter may be due to this substance's having been ])i-oduced from ilu> clioline of the l(>cithin molecule. The fact that the two organisms isolated from milk gave a positive reaction and that they were selected at random, indicates that such a fermentation might be found fairly oft(>n. These results would tluMx^forc* seem to ))oint to Ixicterial agenci(^s as the cause of the fiho\vn in table 27 are of interest only to the extent that they show the variation in acid valu(> caused by different species of bacteria. Inasmuch as the sanipl(>s were placed in storage ininie(liat(>ly after being inoculated, it is probahle that the chang(>s are the result of bacterial enzynu^s liberated by autolysis, because it has been reix-atedly shown that little or no growth takes place during storage. 148 George Cornell Supplee SUMMARY The data prcs(>nt(Hl in this pajxT show beyond a doubt that there is in jiornial butter a sufficient amount of lecithin to yield, on decom- position, small quant it i(>s of trimethylamine, and it is shown also that small quantities of this substance are essential for the manifestation of a fishy odoi'. Furthermore, it is shown that when this substance is worked into butt(^r under the proper conditions, it produces a flavor d(\scribed as fisliy. These K^sults are most uniform when trimethylamine butyrate is used. An associative fermentation in butter or in cream, with the ultimate foi-mation of this substance, is quite possible. As to whether or not this or some otlier volatile and unstable combination of trimeth,ylamine is tlie cause of th(> natural fisliy flavor, remains to be shown more conclusively. Certain data do indicate that trimethylamine is found in soni(> samples of fishy-flavored buttei' but not in others. Altho it is possible that its presence is incidcMital in such samples, that is not believed to be tiie case. In this ccMinection it is worth while to call atten- tion to the confusion betw(>en th(^ fishy, oily, and metallic flavors when they are pi'escnt to only a slifrht initiation of tlie development of these flavors depends on a conunon fundamental factor. AMiethei' or not any particular one of tlu>m develops to its typical flavor would dcpeMid on tlie presence of ('(M'tain conditions which wei'o siHH'ific for that flavoi'. With this possil)ihty in view, it would be logical to assume that tiimethylamine is responsibl(> foi- tli(> typical luM'rinjj;, or mack(M'el, lla/or and odor in ])utter, and tliat the absence of this su])stance would result in the manif(\station of similar but non-typical flavors. Tliere sec^ns to l)e no doui)t that the ])res(>nc(> of a definite acid con- dition in the ])utter is essential for thc^ (l(>vei()pment of the fishy flavor. This condition is best obtained when butter is made from cream con- taining lactic acid, regardless of whether this is developed by bacteria or added to thi^ cream in the form of the commercial product. Further- more, the results indicate that, while a definite acid condition is essential, it must be accompanied by some other equally important factor. The data show tliat this factor is determined by biological agencies. It appears that both th(>se factors nuist (>xist in a definite and delicate relationship, and that if the proper equilibi'ium is disturbed, the characteristic flavor is not manifest. Numerous results and observations indicate that the imknown transient factor is trimethylamine. The Lecithin Content of Butter 149 The bacterioloo;ical aspects of the problem seem to involve the deter- mination of the relationship already mentioned. It is shown that the acid value of butter is to a certain extent regulated by biological factors, probably enzjmies. It is shown also that trimethylamine may be produced in milk and in cream, probabl}^ to some extent from lecithin, with the consequent production of the fishy flavor in those products. Furthermore, it is shown that Bacterium ichthyo.smius, which producc^d the flavor in those substances, would produce^ the flavor in ])utt('r also under certain conditions. It would therefore schmu possible that other species of microorganisms might bring about tlie same type of change. It seems highly probable that the growth of bacteria in the cream before it is made into butter determines the conditions necessary for the later mani- festation of the fishy flavor. The data dealing with lecithin as tlu^ source of trimethylamine in milk products arc too meager to warrant definit(! conclusions at this time. However, the results presented herein, taken together with what is known regarding tliis substance, indicate that this is one of the most logical sources. ackxowledg:\iexts The author wishes to express his keen appreciation and gratitude to Dr. E. S. Guthrie, Mr. W. E. Ayres, and Mr. H. C. Jackson, of the Depart- ment of Dairy Industry at Cornell University, to all other members of this Department who have been kind enough to aid in the scoring of butter, and to all who have furnished samples for study. 150 Geokge Cornell Supplee BIBLIOGRAPHY Barger, George. T1u> simpler natural bases, p. 1-215. (Reference on p. 12.) 1914. BoRDAS, F., AND 1{a('zkowski, Sig. i)E. Variation de I'acide phos- phorique suivant I'age du lait. Acad. Sci. [Paris]. Compt. rend. 135.302-303. 1902. Browne, C. A., Jr. A conti-il)ntion to the chemistry of butter-fat. II. The chemical composition of buttcn-fat. Amer. Chem. Soc. Journ. 21:807-827. 1899. Bltdai (Bau(M-). Koloman. Methode zur fiuantitativ(Mi Bestimmuns des Annnoniaks und Trimcthvlamins. Ztsclu-. jihysiol. Chem. 86:107- 121. 1913. Davis, W. A. .Vmincs and aunnonium bases. /// Allen's commercial orva.oorated milk. lo-.va Agr. Exp. Sta. Research bul. 38 : 2;'>3 240. 1917.' ILvRDiNfi, H. A., RofiER.s, L. A., and Smith, G. A. Notes on some dairy troubl(>s. New York ((i(>neva) Agr. F.xp. Sta. Bul. 183:173-193. (Refcn-ence on p. 181.) 1900. Harrison, F. C. Rej^ort of Professor of BactfM'iology. Ontario Agr. Coll. and Exp. Farm. Ann. rept. 27 (1901) : 74-81. (Reference on p. 79.) 1902. Hasehuoek, Karl. Ueber das Schicksal des Lecithins im Korper, und eine Beziehung dessellxMi zinn Sumpfgas im Darmcanal. Ztschr. physiol. (4iem. I2:148-l()2. 188S. Hunziker, O. F. Investigations in dairy manufactures. In Report of the Depaitment of Daii'v Husbandry. Purdue Univ. Agr. Exd. Sta. Ann. rept. 28 (1915) :39. " 1910. The Lecithin Content ok Butter 151 Koch, Waldemar. Die Leeithane und ihrc Bcdeutung fiir die lebcnde Zelle. Ztschr. physiol. Chem. 37:181-188. 1902-03.' Leathes, J. B. The fats, p. 1-138. (Reference on p. 4o.) 1913. Long, J. H. Observations on the stability of lecithin. Arner. Chem Soc. Journ. 30: 881-895. 1908. MacLean, Iluon. ITiitersuchungen iibei' l''ifi-e!b-Lecilhin. Ztsc-hr. phys- iol. Chem. 59:223-229. 1909. Nerkincj, ,]., AND Haensel, E. Der L(>citliinsehah diM- Milch. Biochem. Ztschr. 13:348-353. 1908. O'Callaghan, ]\r. A. Fishy-tlavorcHl })utfer. The cause and remedy. Agr. gaz. New Soiith Wales 12(1 90 1 ) : ;M 1 -340. 1 902. — — ■ — Butter classification. The s(>i<>ntific examination of butt(M- for export to l^ngland. Agi-. gaz. Xew South Wales 18 (1907 j : 223-227 1908. Piffard, II. Ci. Fishy flavor in but tea*. Xcnv York pi-od. rev. and Amer. creamer}' i3-:20. 1901. Reakes, C. ,]., Cuddie, D., axd Riad, II. A. Fishy flavour in butter. A preliminaiy note on an incjuirv into the cause of the defect. Journ. New Zealand Dept. Agr. 4:1-0. 1912. Rogers, L. A. Fishy flavor in butter. U. S. Bur. Anini. Indus. Cir. 146:1-20. 1909. The develoimient of flsliy fla\-or in ])utter. Wisconsin Bu.tter- makers Assn. Proc. 13:70. 1911a. The fishy flavor in butt (>r. 4'he milk dealer 3"^: 10-12. 1914 b. Stanek, Vladi]\iih. Uber das Cholin])erjodid und die quantitative Falhing von Cholin durch Kaliumhijoilid. Ztschr. phvsiol. Cluan. 46:280-285. 1905. Washbx'rn, R. M., and DAHLiiERG, A. (■. The influence of salt on the changes taking place in storage buttei-. Joui-n. dairv sci. i: 114-12(5. 1918. . " Memoir 24, A Study of the Plant Lice Injurinri the Foliaqc and Fruit of the Apple, the fifth precedinp; mimbnr in this soiies of publications, was mailed on October 7, 1910. Memoir 25. Tlic Crmn-FHiK of Xeiv York. Purl I . I)istrihi,t,on and TaxoJiomy of the Adult Flies, i\\c fourth i)rec('(iinsi number iu thi,* aeries of publications, was mailed on October 23, 1919. Memoir 2(), The Dri/ Root-Hot of the Bean, the third preceding number in this series of publications, was mailed on October 23, 1919. 000 885 973 5