UC-NRLF 3DS EXCHANGE . A Study of the Highly Unsaturated Fatty Acids Occurring in Fish Oils BY JOHN BERNIS BROWN B.S., University of Illinois, 1915. M.S., University of Illinois, 1917. THESIS hibmitted in Partial Fulfillment of the Requirements for the Degree of DOCTOR OF PHILOSOPHY IN CHEMISTRY IN THE GRADUATE SCHOOL OF THE UNIVERSITY OF ILLINOIS 1921 EASTON, PA.: ESCHBNBACH PRINTING COMPANY 1920 A Study of the Highly Unsaturated Fatty Acids Occurring in Fish Oils BY JOHN BERNIS BROWN B.S., University of Illinois, 1915. M.S., University of Illinois, 1917. THESIS Submitted in Partial Fulfillment of the Requirements for the Degree of DOCTOR OF PHILOSOPHY IN CHEMISTRY IN THE GRADUATE SCHOOL OF THE UNIVERSITY OF ILLINOIS 1921 EASTON, PA.: ESCHENBACH PRINTING COMPANY 1920 EXCHANGE ACKNOWLEDGMENT The writer wishes to express his great indebtedness to Professor G. D. Beal under whose able direction the preparation of this thesis has been carried out, and who has made many valuable suggestions as to methods .of attacking the problems involved. The writer desires also to thank Professor Roger Adams and Doctor C. S. Marvel of the Division of Organic Chemistry for very helpful advice concerning a number of the organic chemical problems involved in the thesis, Mr. Paul Anders for construct- ing the special fractionating apparatus used, and Doctor A. M. Patterson, former Editor of Chemical Abstracts, for offering valuable suggestions concerning the nomenclature of the highly unsaturated acids. THE HIGHLY UNSATURATED FATTY ACIDS OF FISH OILS Introduction While engaged in the preparation of pure fatty acids for use in medical research during the late war under the auspices of the National Research Council, the writers had occasion to prepare pure clupanodonic acid, CisHjsOj. The information contained in the literature regarding this acid was vague, and a number of preliminary studies were made in order to divise a cheap and convenient method for its preparation. The method finally adopted was a modification of that of Riedel. 1 The methyl esters, which were prepared by debromination with zinc of the polybromides of the methyl esters of menhaden oil, had a molecular weight too high for methyl clupanodonate, and furthermore distilled over too wide a range to be a pure compound. These facts indicated that we were dealing with 1 Riedel, Ger. pat., 266,350. some acids other than clupanodonic. Since practically none of the prop- erties of clupanodonic acid has been described, the investigation was con- tinued with the idea of determining the nature of the pure acid, and it has finally developed into an attempt at a characterization of the highly unsaturated acids of fish oils. Hofstaedter in 1854 2 was the first to discuss in detail the presence of an unsaturated acid in a marine oil. Investigations by Fahrion, 3 Bull, 4 Tolman, 6 Twitchell 6 and others have demonstrated beyond question the presence of highly unsaturated acids in both fish and marine animal oils. By brominating the acids from Japanese sardine oil and reducing the resulting bromides with zinc and alcoholic hydrochloric acid, Tsujimoto 7 obtained an acid with an iodine number of 344, which he called clupanodonic acid. Majida and Okada 8 reported that the acids obtained in this fashion had a molecular weight too high for clupanodonic acid, and that on hydrogenation they obtained an ap- parent mixture of arachidic and behenic acids. When the problem had been under way for 18 months, we read the interesting pre- liminary report of Tsujimoto in which he describes the separation of the acids of fish oils by means of variations in the solubility of their lithium salts in acetone. 9 From the analysis of 4 fish oils he concluded that the highly unsaturated acids were chiefly C 2 oH 3 2O 2 and C 2 2H 3 4O2, the latter of which he proposes to call clupanodonic acid instead of that to which he had formerly given the name. 10 In view of the fact that this name has been used for the compound, CisH 2 8O 2 , for sixteen years, we prefer to retain it and are using, therefore, the following nomenclature 11 for the unsaturated acids throughout this paper. Formula Double bonds Acid name Formula Double bonds Acid name Ci 6 H 2 6O 2 3 Hexadecatrienoic C 20 H 3 oO 2 5 Bicosapentenoic CisH 3 oO 2 3 Linolenic C 22 H 3 4O 2 5 Docosapentenoic CisH28C 2 4 Clupanodonic C 22 H3 2 O 2 6 Docosahexenoic C 20 H 32 O 2 4 Arachidonic Experimental Part Attempted Preparation of Clupanodonic Acid from Menhaden Oil Preparation and Bromination of the Esters of Menhaden Oil. In order to compare the methyl, ethyl and butyl esters of menhaden oil and 2 Hofstaedter, Ann., 91, 177 (1854). ' Fahrion, Chem.-Ztg., 17, 521 (1893). 4 Bull, ibid., 23, 996 (1899); Tidskrift Kemi, Farm. Terapi, 14, 1 (1917). 6 Tolman, J. Ind. Eng. Chem., 1, 341 (1909). 8 Twitchell, ibid., 6, 564 (1914); 9, 581 (1917). 7 Tsujimoto, J. Coll. Eng. Imp. Univ. Tokyo, 4, 1 (1906). 8 Majida and Okada, Science Repts., Tohoku Imp. Univ., 3, 1 (1914). 9 Tsujimoto, /. Chem. Ind. (Japan), 23, 1007 (1920). 10 Since the above was written, Professor Tsujimoto, writing of Japanese sardine oil, has again given the name clupanodonic acid to the compound C 22 H 3 4O2 [ Chem. Umschau Fette, Oele, Wachse Harz, 33, 261 (1922)]. Since the acid Ci 8 H 28 O 2t to which the name was originally given, undoubtedly occurs in oils from the herring family, we can see no reason for the change in the use of the name. We feel that more confusion will be caused by a change now, since a number of authors, among others MacArthur and Burton [J. Am. Chem. Soc., 38, 1375 (1916)] have reported clupanodonic acid, Ci8H 28 O 2 , as a constituent of animal tissues. 11 Dr. A. M. Patterson has kindly verified the nomenclature of these acids. their behavior towards bromination, they were prepared by a modified form of Haller's method 12 of alcoholysis. Menhaden oil was refluxed with 1.5 times its weight of absolute methyl alcohol or an equivalent amount of the other alcohols, containing about 2% of dry hydrogen chloride, for 24 hours. The mixture was cooled and washed with strong brine to remove the ex- cess of alcohol, the liberated glycerol and hydrochloric acid, the esters then being distilled under reduced pressure. The esters were dissolved in 9 times their weight of dry ether and bromine slowly added in 10% excess, while the mixture was kept at a temper- ature of 10 to 5 by a bath of ice and salt. After standing for 6 hours the precipi- tated bromides were washed with ether by decantation until freed from bromine, then dried in air. These methods of esterification and bromination were used generally throughout this investigation. The percentage yield of bromides based upon the original weight of ester, or polybromide number, and the percentage of bromine in the bromides were determined. The mixed butyl esters of the acids of linseed oil were carried through a like process to determine whether evidences of substitution would appear. Determination of Bromine in the Polybromides. The bromine content of the ester polybromides was determined by the aid of the Parr peroxide bomb, using the authors' modification of the procedure outlined by Lemp arid Broderson. 13 A sample of about 0.3 g. was placed in the fusion cup together with 1 g. of accelerator, in this case sodium nitrate, 0.5 g. of granulated sugar and a measure of sodium peroxide. After ignition, either by a Bunsen burner or the improved electrical method, the fusion mixture was cooled and dissolved in 150 cc. of water. The~solution was acidified with 22 cc. of cone, nitric acid, an excess of 0.1 N silver nitrate solution added and the solution heated to boiling; 10 cc. of a 4% solution of hydrazine sulfate was added to reduce any bromate ions present and boiling continued until the precipitate had coagulated thor- oughly. After the mixture had boiled, 5 cc. of a saturated solution of ferric alum was added and the excess of silver nitrate titrated with 0.05 N ammonium thiocyanate solu- tion. A blank determination was made with the reagents and sugar alone. TABLE I A COMPARISON OF THE BEHAVIOR OF THE METHYL, ETHYL AND BUTYL ESTERS OF MENHADEN OIL TOWARDS BROMINATION Boiling Wt. * Wt. Calc. for point ester bromides Polybro- Bromine clupano- Ester at 15 mm. G. G. mide No. % donate Menhaden Oil Methyl 195-240 2100 805 38.3 68.31 68.79 Ethyl* 195-240 600 195 32.5 67.43 67.84 Ethyl 195-240 1270 430 33.8 67.48 67.84 Ethyl 240-265 409 367 85.4 69.49 67.84 w-Butyl 190-245 305 96 31 .5 67.64 65.82 Linseed Acids n-Butyl 190-240 100 51 51 .0 59.30 59.38* a Two lots of ethyl esters were made and distilled into 2 fractions as shown above. The higher-boiling fractions were combined. 6 59.38% is the theoretical bromine content of butyl hexabromo-stearate. This compound sintered at 157 and melted at 160. It has not been previously described. 12 Haller, Compt. rend., 146, 250 (1908). 13 Lemp and Broderson, J. Am. Chem. Soc., 39, 2069 (1917). It has been suggested that sodium bisulfite solution be used as a reducing agent in- stead of the hydrazine sulfate. This must be added and the excess oxidized with potas- sium permanganate solution before the addition of the silver nitrate solution. The results with this procedure were not always of the best. One of us (J. B. B.) has had an illuminating experience in the determination of iodine by this method. The peroxide reaction apparently gave only iodate since silver nitrate in the acid solution formed no precipitate. Immediately on adding the hydrazine sulfate solution silver iodide was precipitated and bubbles of nitrogen gas were evolved. There was no appearance of free iodine. When sodium bisulfite was used as the reducing agent iodine invariably ap- peared. Was Pure Clupanodonic Acid Obtained? The percentage of bromine in the polybromides obtained from the higher-boiling fraction of the ethyl esters and from the butyl esters is an indication of the presence of more highly unsaturated acids than clupanodonic. These acids are further- more of higher molecular weight, since they are in the higher-boiling fraction. Qualitative tests indicate that the solubility of the polybromides of the esters increases with an increase in the molecular weight of the alcohol radical. The absence of more than traces of substitution in the method of bromination employed is indicated by the fact that butyl hexabromo-stearate as prepared had practically the calculated bromine content, and no evidence of formation of hydrobromic acid could be ob- tained. Denomination ""of the bromides of the fractions of the lower- boiling ethyl ester in absolute ethyl alcohol with zinc dust gave an ester with an iodine number of 333. The acid mixture as obtained by saponifi- cation had an iodine number of 342 and a molecular weight by titration of 305, which values are not in agreement with those for clupanodonic acid. The Effect of Heat on the Fatty Acids of Menhaden Oil. The free acids of fish oils could be distilled only with difficulty, since a thick viscous liquid remained in the distilling flasks. To test the effect of heat a quan- tity of mixed fatty acids from menhaden oil was placed in a Claisen flask and heated to 240. To prevent oxidation a gentle stream of carbon dioxide was passed over the acids. Samples were withdrawn from time to time and the molecular weight was determined by titration. TABUS II THE EFFECT OF HEAT ON THE FATTY ACIDS OF MENHADEN OIL Time of heating, min 00 60 105 165 Mol. wt 289.7 297.4 303.4 313.2 The maximum temperature to which the fractions are subjected under the conditions of fractional distillation under vacuum is 240. A similar experiment conducted with the esters of these same acids showed practically no change in molecular weight. This apparent polymerization may ex- plain why large quantities of tarry residue are found in the distilling flask on fractionating the free acids, while the esters may be distilled 9 almost to dryness. All subsequent distillations, when possible, were of esters. Comparative Analysis of Five Commercial Fish Oils From the data given above it was evident that in the sample of men- haden oil used there were present highly unsaturated acids other than clupanodonic acid. Five fish oils which were available were analyzed to determine the probable nature of the highly unsaturated acids. The constants of the oils are given in Table III. TABLE III ANALYTICAL CONSTANTS OF COMMERCIAL FISH OILS Oil Sap. No. Iodine No. n Menhaden 191.2 151.7 1.4778 Salmon 185.0 137.2 1.4768 Cod 186.9 151.0 1.4770 Herring 186.5 139.8 1.4765 Sardine 187.3 158.1 1.4791 One kg. of each oil was esterified by the method mentioned previously and the methyl esters were distilled into fractions over a 10 range under 15 mm. pressure. In the case of menhaden oil 3 kg. was used and the dis- tillation repeated thrice. The fractions were then analyzed for free acid, mean molecular weight of the acids, index of refraction, iodine number, polybromide number and percentage of bromine in the bromides, while the mean molecular weight of the acids was calculated from the molecular weight of the esters as determined by saponification. The fish oils examined were found to be decidedly similar in character. In general the iodine number, polybromide number and percentage of bromine in the bromides increased with the boiling point, the main di- vergence being in the last fraction, which might easily contain decomposed material. The oils as a class contain acids whose esters distil at much too high a temperature for the Ci 8 series. Among these higher acids are those which are more highly unsaturated than clupanodonic acid. Separation of the Highly Unsaturated Acids From the facts mentioned above it is quite probable that the clupano- donic acid previously described in the literature was not a pure compound, but rather a mixture of a number of acids. Accordingly a number of attempts were made to separate the fatty acids of menhaden oil, the most successful procedure being applied to cod and herring oils. Separation by Fractional Distillation. The esters from 3 kg. of menhaden oil obtained in the previous experiment were fractionated 6 times. The pressure during each distillation was maintained at 15 mm. and on the last distillation the fractions were collected in such a way that certain 10 ones would represent as far as possible only acids of the same carbon con- tent. Although the intermediate fractions had such properties as to indi- cate that the principal fractions were not chemical individuals, the boiling point, index of refraction, molecular weight and iodine number indicated the predominance in each of these fractions of acids of a certain group. TABLE IV RESULTS OF ANALYSIS OF THE FRACTIONS FROM THE SIXTH DISTILLATION OF THE METHYL ESTERS FROM THREE KILOGRAMS OF MENHADEN OIL 15 mm. range Wt. M. m. wt. Iodine Probable series Cl4 C 18 C 20 C 22 From the series of analyses just tabulated the following conclusions may be drawn: (1) It is possible by fractionation of the methyl esters to make a rough separation of acids according to molecular weight. (2) The important acids have a carbon content ranging from CH to C 22 . Those of the lower molecular weights are largely saturated. (3) The iodine numbers and mean molecular weights of the fractions rise rapidly with the boiling points. The index of refraction of the esters was also found to increase regularly. When the index of refraction was plotted against iodine num- bers practically a straight line was obtained, the main divergence being with the last point. The same general relationship is seen when the index of refraction is plotted against molecular weight, the breaks at the beginning and end of the series giving the graph the form of the sign of integration. (4) The polybromide number of the esters rises with the boiling point, while the bromine content of the derivatives also increases. Unsaturation undoubtedly increases with an increase in molecular weight. Separation by Metallic Salts. It is apparent from the data previously given that a number of highly unsaturated acids occur in fish oils. It is furthermore probably true that no one has succeeded in isolating any of them in the free state. The clupanodonic acids of Tsujimoto and of Riedel, previously referred to, were undoubtedly complex mixtures. Var- Fraction c. G. gii acids no. 1 156-166 21.5 1.4415 234.1 18.58 2 166-170 26 1.4408 233.7 14.73 3 170-180 129 1.4425 239.5 27.52 4 180-194 257 1.4470 255.2 53.2 5 194-198 500 1.4500 260.5 72.74 6 198-210 208 1.4546 268.9 101.8 7 210-216 191 1.4610 276.6 142.25 8 216-226 237 1.4670 283.2 185.2 9 226-230 112 1.4765 293.7 245.4 10 230-235 66 1.4810 300.1 277.0 11 235-240 94 1.4858 305.1 299.7 12 240-245 107 1.4895 309.9 306.5 13 245-250 76 1.4934 317.2 306.6 14 250-255 31 1.4960 324.9 304.1 15 255-260 22 1.4979 326.4 284.3 11 ious methods have been proposed for the separation of the saturated and unsaturated fatty acids. Most of the methods depend upon differences in the solubility of metallic salts of the unsaturated acids in organic sol- vents, those of the unsaturated acids being most soluble. Such sepa- rations are not quantitative, due to the tendency of the slightly unsatur- ated acids to follow the saturated acids in their behavior, and also to an apparent mutual solubility of the salts of the two types, one in the other. The Lead Soap-ether Method. Five hundred g. of refined menhaden oil (iodine number 183.9; saponification number 191.8) was treated ac- cording to the usual procedure for the lead soap-ether method 14>15 using only such modifications as were necessitated by the large size of the sample. The lead salts so separated were decomposed as usual with- hydrochloric acid and the solid and liquid acids examined. The liquid acids weighed 300 g., showed an average molecular weight of 328.6 (the abnormally high value is probably due to the presence of a small amount of retained ether) and an iodine number of 237.9; whereas the solid acids weighed 150 g. and showed a molecular weight of 264.6 and an iodine number of 55.3. The liquid acids were then converted into then- methyl esters and dis- tilled under 15 mm. pressure. TABLE V THE DISTILLATION OF THE METHYL ESTERS OF THE UNSATURATED ACIDS OF MENHADEN OIL AS OBTAINED BY THE LEAD SOAP-ETHER METHOD Range Wt. Fraction C. G. Iodine No. 1 180-200 55 131.7 1.4560 2 200-210 36 166.5 1.4622 3 210-220 49 210.9 1.4683 4 220-230 37 266.0 1.4771 5 230-240 48 321.0 1.4860 This distillation was twice repeated, when the fraction boiling at 235- 250 had an iodine number of 334.1 and a molecular weight of 326.5. Since, however, the lower fractions gave evidence of the presence of sat- urated acids as impurities and since the method applied to large quantities was so tedious, further investigation by this procedure was discontinued. By the Barium Soap-benzene Method. The application of the barium soap-benzene method 16 was attempted. This is based on the solubility of the barium soaps of fatty acids with more than one double bond in cold benzene containing about 5% of 95% alcohol, added to bring a trace of moisture into the benzene, while the barium salts of the acids of the oleic and the saturated series will dissolve in the hot benzene-alcohol but crystal- lize on cooling. 14 Gusserow, Ann., 27, 153 (1828). 16 Varrentrapp, ibid., 35, 197 (1840). 16 Farnsteiner, Z. Nahr. Genussm., [2] 1, 390 (1898). 12 The sodium soaps of menhaden oil were neutralized with acetic acid and precipi- tated in the cold with an excess of barium chloride solution. The soaps so precipitated were filtered on a Biichner funnel, and when transferred to a large flask and heated on a steam-bath for a few minutes so agglutinated and shrunk that practically all of the occluded water could be decanted. The soaps were then treated with the benzene- alcohol mixture and refluxed for 30 minutes on a steam-bath. The solvent was poured off and the residual soaps were treated with a fresh portion of the solvent. This treatment was repeated until the soaps were completely disintegrated and practically all dissolved. The, combined solutions were then allowed to cool overnight, and filtered from the crys- talline precipitate. About 8 liters of solvent was required for 500 g. of the oil. The acids from both filtrate and precipitate were recovered by decomposing the soaps with hydrochloric acid and distilling the benzene. The last traces of benzene were removed by heating in a vacuum. The results obtained in the first trial distillation of 500 g. were so satisfactory that 3 additional distillations were made, the products being combined for analysis. No. of distillation Character G. Mol. wt. no. 20 1 Liquid 260 306.0 267.0 1.4845 2-3-4 1 Wt. Iodine Character G. Mol. wt. no. Liquid 260 306.0 267.0 1 Liquid 790 305.9 261.4 Solid 144 281.6 53.7 Solid 420 Constants not determined One kg. of these liquid acids was converted into their methyl esters and distilled thrice under 15 mm. pressure. A special Claisen flask was made of Pyrex glass for this distillation by Mr. Paul Anders. The flask had an especially wide neck and side tube, to prevent the liquid from bumping over, and the side arm was inserted a short distance into the upright side tube in order to prevent esters condensing on the stopper from running into the receiver. The side arm was also especially long so that it might be surrounded by a small Pyrex condenser while it conducted the distillate directly into a Raikow re- ceiver which permitted five fractions to be taken without breaking the vacuum. The rubber stoppers were protected from the hot ester vapors by pinning a thin sheet of cork over their lower surfaces. TABLE VI ANALYTICAL CONSTANTS OF THE METHYL ESTERS OF THE LIQUID ACIDS OF MENHADEN OIL OBTAINED BY THE BARIUM SOAP SEPARATION, AFTER THREE DISTILLATIONS Range (15 mm.) Wt. M. m. wt. Iodine no. Fraction C. G. acids of ester 18 1 170-180 8 ... ... 1.4627 2 180-200 98 255.2 111.4 1.4536 3 200-210 100 257.9 142.2 1.4580 4 210-220 147 260.7 181.0 1.4640 5 220-225 49 ... 231.2 1.4718 6 225-230 86 ... 249.3 1.4740 7 230-235 66 ... 280.1 1.4791 8 235-240 95 304.0 316.5 1.4845 9 240-245 65 308.5 334.6 1.4890 10 245-250 43 312.6 347.7 1.4930 11 250-255 58 321.6 348.4 1.4960 12 255-260 34 336.7 330.9 1.4980 Fractions 8-12, boiling over 5 ranges from 235260, had mean molec- ular weights ranging from 304 to 336 and iodine numbers for the esters 13 ranging from 316 to 348. Fraction 8 gave values corresponding to the theoretical values for the methyl ester of arachidonic acid. Probably the higher-boiling fractions contain docosapentenoic and docosahexenoic acids. The polybromide numbers of Fractions 8 and 11 were determined by the following procedure. Place 1-2 g. of the ester in a weighed 50cc. centrifuge tube and add 35 cc. of an- hydrous ether. Place the tube in a cooling bath kept below and add bromine, with vigorous stirring to a distinct excess or until the solution is colored red. Allow the tube and contents to stand in an icebox overnight in the dark, centrifuge, decant the ether, and wash the precipitate 4 times by centrifuging with 40cc. portions of ether. Dry the tube and contents at 60 for 2 hours and weigh. The weight of bromides, divided by the weight of ester and multiplied by 100, gives the "polybromide number." In order to deter- mine the nature of the residual acids in the ether nitrates, they were shaken with sodium thiosulfate solution to remove the excess of bromine, then dried with calcium chloride, the ether was allowed to evaporate and the residue finally dried in a vacuum oven at 60. Fraction 8 showed a polybromide number of 97.90, giving 69.61% of bromine in the bromides, and 56.56% of bromine in the ether-soluble portions, while for Fraction 11 the polybromide number was 101.13 and the bromine percentages 70.07 and 56.05 respectively. Fraction 8, above, could not be pure methyl arachidonate since this would yield a bromine derivative with 66.78% of bromine, whereas the derivative formed actually contained 69.61%. If this fraction contained a con- siderable quantity of methyl docosapentenoate, which gives a polybromide with 71.66% of bromine, such results might be obtained. However, the substitution of this acid in part or entirely would give iodine numbers which would be much higher than those obtained. It is more probable that the fraction is a mixture of methyl arachidonate and methyl doco- sapentenoate with small quantities of the esters of less unsaturated acids. An examination of the ether-soluble bromides leads to one of two con- clusions: either the liquid bromides are not completely saturated with bromine, because of some sort of steric hindrance, or they are isomeric liquid bromides mixed with esters of the saturated acids or with bromides of less unsaturated acids. Separation by Reduction of the Polybromide s. The method used in the preparation of these polybromides was essentially the same as that which had previously been employed. A sample of commercial menhaden oil, dark reddish-amber in color, was available. -Two portions of 3 kg. each were converted into their methyl esters. The first lot was re- fluxed with the acid-alcohol mixture for 12 hours and the second for 24 hours. The yields of esters on distillation, between the usual range of 190-250 under 15 mm. pres- sure were 2000 g. of ester for the first trial and 2700 g. for the second. The methyl esters were brominated in kilogram quantities in ether solution at a temperature which was kept below 0, as has been previously described. In all, 1772 g. of polybromides was ob- tained. Three portions of the esters were brominated in ether containing 10% of glacial acetic acid, but no appreciable difference could be observed in either the yield or the percentage of bromine taken up. 14 Previous experiments have been rewarded with low yields when the polybromides were reduced. 2 This was thought to be due possibly to the insolubility of the reacting sub- stances. Accordingly normal butyl alcohol and benzyl alcohol were tried as solvents, since the polybromides appeared to be much more soluble in these liquids. Boiling benzyl alcohol dissolves the polybromides to the extent of 100 g. per liter, giving a dark colored solution from which the white polybromides deposit on cooling. Reduction in Benzyl Alcohol. One hundred g. of the polybromides and an excess of zinc dust were placed in a Pyrex flask with about 200 cc. of benzyl alcohol. The mixture was carefully heated, but as soon as the reaction had begun it progressed violently, and the heat developed was sufficient to break the flask. A second lOOg. portion was mixed in the dry condition with 100 g. of zinc dust and added in small portions to 250 cc. of benzyl alcohol boiling in a 500 cc. flask under an air condenser. No heat was applied after the addition of the first portions of the mixture, the heat of the reaction being suffi- cient to keep the mixture refluxing vigorously. After cooling, the mixture was a very thick viscous liquid from which nothing could be separated. It was thought that the benzyl alcohol had possibly entered into the reaction. Reduction in Butyl Alcohol. A mixture of the polybromides with zinc dust was re- fluxed for 10 hours in the presence of normal butyl alcohol, the alcohol filtered off and the residue washed twice with fresh portions of butyl alcohol. After the alcohol was removed an attempt was made to distil the residue, but nothing passed over even under 15 mm. pressure. After the distillation flask had been removed from the heat for 5 minutes a strongly exothermic reaction set in which resulted in the formation of a mass of tar in the flask. Reduction in Methyl Alcohol. Equal parts of the polybromides and zinc dust were refluxed in methyl alcohol for 48 hours, the alcoholic layer was filtered off and the solid residue washed thrice with alcohol, the alcoholic filtrates being combined and the alcohol distilled under atmospheric pressure. The ester-alcohol mixture remaining was agitated with warm dil. hydrochloric acid to remove the remaining alcohol and to decompose any zinc soaps present. The ester layer was then separated and refluxed for 12 hours with twice its weight of absolute methyl alcohol and hydrochloric acid, under the usual conditions for methanolysis, this operation being made necessary by the formation of some free acid during the debromination. The esters which were finally recovered were dis- tilled under 15 mm. pressure. The first two runs of 500 g. each gave a total of 137 g. of re- duced ester; yield, 45%. A third run of 1050 g. of bromides gave 163 g. of esters; yield, 50.9% ; iodine number of esters, 368.5; mean molecular weight of acids, 307.9; w 20 , 1.4910. The esters distilled between 215 and 250 under 15 mm. pressure. There was practically no residue in the distilling flask after these esters were distilled, a fact ac- counted for by the absence of free acid carried into the alcoholic solution as zinc salt. Comparison of the Acids of Cod and Herring Oils The highly unsaturated esters from cod and herring oils were prepared in a similar manner; 1620 g. of the methyl esters from 2000 g. of cod oil gave 480 g. of bromides, con- taining 70.75% of bromine. These when debrominated gave 93 g. of esters ; yield, 67.65% . 1875 g. of esters from 2200 g. of herring oil gave 517 g. of bromides, with a bromine con- tent of 69.29% . These yielded 83 g. of esters, or 52.3% . The highly unsaturated esters of menhaden oil were distilled thrice under diminished pressure, while those of cod and herring oils were distilled only once, on account of the small quantity available. For the second and third distillations of the menhaden oil esters, the fractions were introduced in order into the distilling flask by means of a sep- aratory funnel, the stem of which had been drawn out to a fine tip. All of the temper- atures were corrected to 15 mm. pressure, including the stem correction for the ther- mometer. The last 4 fractions of the esters of cod oil showed abnormal values for all but molecular weight, due to accidental overheating. The analyses follow. 15 TABLE VII THE CONSTANTS OF THE FRACTIONATED METHYL ESTERS OF THE HIGHLY UNSATURATED FATTY ACIDS OF MENHADEN, COD AND HERRING OILS, PREPARED BY REDUCTION OF THE POLYBROMIDES Menhaden oil, distilled thrice Fraction Range (15 mm.) wt. G. M. m. wt. Iodine acids no. * Poly- bromide no. Br Mixed esters (215-250) 307 .9 368.5 1 .4910 105. 33 69. 40 1 below 203 9.5 322 .0 260.6 1 .4753 . 2 203-213 30.0 274 .1 334.8 1 .4800 . 3 213-218 16.0 280 .8 348.8 1 .4860 94. 67 68. 02 4 218-223 22.0 284 .7 359.4 1 .4888 96. 62 69. 71 5 223-228 47.0 292 .2 363.8 1 .4907 101. 9 69. 78 6 228-233 35.5 298 .1 372.8 1 .4919 101. 69. 61 7 233-238 33.0 307 .5 376.3 1 .4950 101. 21 70. 45 8 238-243 31.0 315 .8 379.2 1 .4970 97. 56 70. 40 9 243-248 17.0 317 .4 373.2 1 .4980 76. 02 70. 48 10 248-255 11.0 324 .7 357.9 1 .4987 64. 90 70. 99 Cod oil, distilled once Mixed esters (210-247) 298 .0 383.5 1 .4912 125. 8 70. 4 1 below 213 11.0 279 .4 362.2 1 .4865 104. 68. 70 2 213-218 11.0 286 7 369.8 1 .4881 109. 8 69. 21 3 218-223 18.0 293 .0 380.0 1 .4895 113. 69. 56 4 223-228 11.5 303 .3 305.9 1 .4940 . . . 5 228-233 4.0 307 .3 277.9 1 .4985 . . . , . . ~ 6 233-250 5.5 313 .0 257.5 1 .4992 . . . m . . , 7 250-270 10.0 323 .5 232.5 1 .4993 .. . Herring oil, distilled once Mixed esters (215-250) 299 .2 373.9 1.4910 113. 25 69. 66 1 below 213 6.0 277 .8 354.0 1 .4860 97. 7 68. 99 2 213-218 7.0 283 .4 361.8 1 .4873 101. 1 68. 81 3 218-223 9.0 286 .0 366.3 1 .4888 103. 69. 05 4 223-228 11.5 294 .0 369.5 'l .4900 111. 69. 45 5 228-2-33 16.5 299 .0 376.0 1 .4928 106. 5 69. 69 6 233-238 11.0 309 .3 379.4 1 .4960 102. 70. 12 7 238-245 4.5 315 .0 372.0 1 .4992 94. 70.61 Menhaden oil, faction 2, redistilled 2a 180-195 9.0 266 .8 316.5 1 .4800 72. 78 68. 59 2b 195-205 7.5 274 .5 333.7 1 .4829 . . . . 2c 205-212 5.5 282 .7 344.7 1 .4896 . . . .' The Nature of the Fatty Acid The following conclusions may be drawn from the examination of the ester fractions of the various oils in the preceding table. The highly unsaturated acids would seem to contain 16, 18, 20 and 22 carbon atoms, the molecular weights of the tetra-unsaturated acids being 248, 276, 304 and 332, respectively. The lowest mean molecular weight determined was 266.8 (see below), which is 9 points lower than that for 16 clupanodonic acid. The highest value found was 324.7, which is only 7 units below that for the C 2 2 acid. From the examination of the iodine numbers it will be seen that up to the last two fractions the unsaturation increases with the boiling point and molecular weight. The calculated iodine numbers of the methyl esters of the acids which may be present are: hexadecatrienoate, 288.5; clupano- donate, 350.1; arachidonate, 319.3; eicosapentenoate, 401.7; docosatetren- oate, 293.5; docosapentenoate, 369.0; docosahexenoate, 445.2. Fraction 2 when slowly distilled into 3 sub-fractions gave a fraction having an iodine number of 316.5 and a molecular weight for its acids of 266.8, which data indicate a mixture composed of 2 / 3 methyl clupanodonate and l / 8 methyl hexadecatrienoate. The data for Fraction 3 are very close to the theoretical values for methyl clupanodonate. The data for Fractions 6 and 7 indicate mixtures of methyl arachidonate and eicosapentenoate, while those for Fraction 9 indicate methyl docosatetrenoate, docosapentenoate and docosahexenoate . The percentages of bromine in the polybromides of the various fractions show that the degree of unsaturation increases with an increase in the boiling point of the fraction and, therefore, with the molecular weight. One of the sub-fractions obtained from Fraction 2 and also Fraction 3 yielded bromides with 68.59 and 68.6% of bromine which are almost in exact agreement with the calculated value for methyl octobromostearate. This compound has not been previously prepared. It is a white amor- phous solid which melts at 240, uncorr. The following percentage values for bromine content have been calcu- lated: octobromostearate, 68.79; octobromo-arachidate, 66.78; decabromo- arachidate, 71.66; octobromobehenate, 64.90; decabromobehenate, 69.90; dodecabromobehenate, 73.72. Fractions 6 and 7 would seem from this also to consist of methyl arachidonate and eicosapentenoate, while Frac- tions 9 and 10 are probably chiefly methyl docosapentenoate. Since Fraction 10 yields a bromide which contains more bromine than Fraction 9, the former undoubtedly contains some of the methyl docosahexenoate. This is the principal evidence which we have of the presence of this acid. The polybromide numbers of the fractions furnish little evidence of the nature of the fatty acids. The very important fact which should be observed here is that they are only about 1 / 8 of the theoretical value, which is always in the vicinity of 300. On the basis of the analyses of Fraction 3, it is believed that this fraction is almost pure methyl clupanodonate. It is further believed that this represents the first time that a derivative of clupanodonic acid has been prepared in a reasonably pure state. The constants of methyl clupano- donate based on this sample are, iodine number (Wijs) 348.8; b. p. (15 mm.) 215; n zo , 1.4860. The octobromide is described above. r The data in Table VII for cod and herring oils offer no evidence to con- tradict our conclusions based on the more complete distillation of the esters of menhaden oil, but rather confirm most of these data. The poly- bromide numbers of the mixed esters of the 3 oils are 105.3, 125.8 and 113.3, values which are about 1 / 3 of those calculated. The Nature of the Bromination Reaction A low yield of polybromides upon bromination of the pure, highly uii- saturated esters of linolenic and linolic acids has been reported by Rollett 17 and Erdmann. 18 The most satisfactory explanation of the liquid bromides is that perhaps they are isomeric soluble bromides with correspondingly different properties. To furnish further light on this problem the follow- ing experiment was carried out. Starting with menhaden oil, the methyl esters were prepared and these reduced to the esters of the unsaturated acids. Then, under quantitative conditions, the supposedly pure un- saturated acids were prepared by saponification and brominated, and both the liquid and solid bromides weighed, reduced and again brominated. Suitable analyses were made at each step. The diagram, Table VIII, TABUS VIII QUANTITATIVE BROMINATION AND REDUCTION OF MENHADEN Oil, ESTERS 1900 g. of menhaden oil-*- 1337 g. of methyl esters-*- 500 g. of polybromides (68.62% of bromine)*- 47 g. of methyl esters * 40 g. of fatty acids. Iodine number, 360.5. Molecular weight, 304.2 33.19 g. brominated ,/ \ 23.37 g. insol. bromides (69.53% of Br) polybromide no., 70.44 I 5.5 g. of fatty acid iodine no., 380.6 mol. wt., 317 brominated 2.56 g. crystallized 2 g. 66.36% mol. wt., 667 sol. bromides 6.11 g. 65.93% of Br insol. bromides, 2.65 g polybromide no., 103.7 70.38% of Br sol. bromides liquid, 70 g. 62.53% of Br reduced iodine no., 278.5 mol. wt., 301 3.12 g. brominated I sol. bromides 60.76% Br insol. bromides 0.66 g. polybromide no., 21.3 67.45% Br The acid obtained here represents that from an additional 7 g. of solid bromides from the same point in a parallel experiment. shows the result of this experiment. Although no final conclusion may be drawn from these data, the following suggestions may be made. The liquid bromides seem to contain in general considerably less bromine than the solid. When reduced, the former give a product with an iodine 17 Rollett, Z. physiol. Ghent., 62, 410, 421 (1909). Erdmann and Bedford, Ber., 42, 1324 (1909). 18 number considerably less than the original acid, while the molecular weight is only slightly less. These in turn yield only a small amount of solid bromides on bromination. The low bromine content of the liquid bromides and the decrease in iodine number suggest a change in the degree of unsaturation. There seem to be four possibilities of change here. (1) One double bond may have been saturated during the reduction. (2) There may have been a rearrangement leaving one of the unsaturations in a position where it cannot be brominated, since it is known that an -/? unsaturation causes a reduction in the iodine number of an acid. (3) Oxidation may have oc- curred. (4) A double bond may have disappeared by ring formation as from a 1-5 di-ene grouping. The second and fourth possibilities are the ' most reasonable. In support of the second it has been noticed repeatedly in this Laboratory that at bromine is not added to the unsaturated acids with the same ease as iodine monochloride or monobromide in at- tempts at the determination of the bromine absorption values of drying vegetable and fish oils. In support of the possibility of ring formation we may recall the fact that when the highly unsaturated esters of cod oil were accidentally overheated, there was a rapid lowering of the iodine and polybromide numbers with no change in molecular weight. Bromination might catalyze a similar change. Whatever may be the answer to the question of the liquid bromides, apparently 1 / 8 or less of the ester brominated goes to form the normal ether-insoluble bromides, while the other 2 / 3 forms a product with 4 to 8% less bromine, which is soluble in ether. The liquid bromides on debromination yield a product from which a smaller proportion of poly- bromides may be obtained on rebromination in ether solution. Presumably the material soluble in ether after bromination consists of two portions: the solid bromides which are dissolved in the ether solu- tion of the liquid bromides, and the true liquid bromides, either isomeric with the others or formed from altered unsaturated acids. Summary The methyl, ethyl and normal butyl esters of the acids of menhaden oil and their bromination products have been prepared and analyzed. The presence of more highly unsaturated acids than clupanodonic is indi- cated. The analysis of the methyl ester fractions from 5 fish oils indicates the presence of acids more highly unsaturated than clupanodonic. By esterifying menhaden oil and fractionating the esters 6 times by distillation under diminished pressure, collecting 16 fractions, the presence of acids with 16, 18, 20 and 22 atoms of carbon is indicated. The unsaturated acids obtained by the lead soap-ether and barium soap- 19 benzene separations have been esterified and fractionated and their bromine addition products analyzed. The ether-insoluble bromine addition products of the unsaturated acids have been debrominated and fractionated and their bromine addition products analyzed. We believe that we have demonstrated the presence in fish oils of myris- tic, palmitic, palmitolic (hexadecatrienoic), clupanodonic (octodecate- trenoic), arachidonic, eicosapentenoic, docosapentenoic and docosa- hexenoic acids. VITA The writer of this thesis received his high school education in the Sterling Township High School, Sterling, Illinois during the years 1907 to 1911. He entered the University of Illinois in the fall of 1911 and pursued the usual four year course in chemistry. He received the B.S. degree in 1915 and entered the Graduate School of the University of Illinois the same year. He was a graduate assistant in chemistry from 1915-1917, when he received the M.S. degree. He was assistant in chemistry for a short time in the fall of 1917, when he resigned to enter the army. He was commissioned and detailed to work under Prof. G. D. Beal on Chemical War Problem No. 112, National Research Council, which work was carried out at the University of Illinois the last three months of 1918 for which residence credit in the Graduate School was allowed. From January, 1919, to the spring of 1920 he held the position of full time assistant in chemistry. He held the Dupont Fellowship in chemistry during 1920- 1921. The writer is the author of "The Composition of the Body Fat of the Common Woodchuck (Marmota Monax)," master's thesis, unpublished, and "The Preparation of the Pure Fatty Acids and their Cholesteryl Esters," a report on Chemical War Problem No. 112, unpublished. THIS BOOK IS DUE ON THE LAST DATE STAMPED BELOW AN INITIAL FINE OF 25 CENTS WILL BE ASSESSED FOR FAILURE TO RETURN THIS BOOK ON THE DATE DUE. THE PENALTY WILL INCREASE TO 5O CENTS ON THE FOURTH DAY AND TO $f.OO ON THE SEVENTH DAY OVERDUE. LD 21-100m-7,'33 OF CALIFORNIA LIBRARY THIS BOOK IS DUE ON THE LAST DATE STAMPED BELOW AN INITIAL FINE OF 25 CENTS WILL BE ASSESSED FOR FAILURE TO RETURN THIS BOOK ON THE DATE DUE. THE PENALTY WILL INCREASE TO SO CENTS ON THE FOURTH DAY AND TO $1.OO ON THE SEVENTH DAY OVERDUE. MAR 13193;* APR 251939