20 A Photoelectric Photometer for the Estimation of Vitamin A in Margarine BY J. L. BOWEN N. T. GRIDGEMAN AND G. F. LONGMAN (Read at the MeetiPzg of the Society on November 7th 1945) INTRODUCTION-Although statutory minima for the vitamin-A content of margarines have been in force in this country for some years a wholly satisfactory method of test is still wanting. The bio-assay is unsuitable for routine estimation of vitamin A not only because of the well-known difficulties in obtaining inter-laboratory agreement but because of its cost and the length of time involved in making the test. It is therefore necessary to make use of the light-absorption characteristics of vitamin A itself or of its chloroform - antimony trichloride reaction product and the literature furnishes a choice of method^,^^^^^^^ each of which however is open to one objection or another.The present paper offers yet another method-one it is hoped that approaches a little nearer the ideals of simplicity inexpensive-ness and reproducibility. Normal trading in vitamin-A oils and concentrates is conducted on the basis of the “E325” test i.e. on the determination of the extinction coefficient referred to a 1% solution, at 325 mp Unfortunately this test cannot be applied to margarine whose normal conten FOR THE ESTIMATION OF VITAMIN A I N MARGARINE 21 of vitamin A is below the critical level of estimation by this means and recourse has to be made to the “spectroscopic blue” test i.e. to the determination of the extinction coefficient, again referred to a 1% solution at 620 mp of the antimony trichloride - vitamin A reaction mixture in chloroform solution.This can be done on a visual spectrophotometer which, however is an expensive instrument that does not usually form part of the equipment of the ordinary routine laboratory. It appeared desirable therefore to evolve a much less costly piece of apparatus that would be simple to use and capable of furnishing results of a high degree of reproducibility. The method about to be described which in essence is the evaluation by means of a photoelectric photometer of the “E620” antimony trichloride test on the unsaponifiable matter was primarily devised for factory control purposes Le. to enable the works laboratory to check the vitamin-A content of the manufactured margarine, and it is from this point of view that the bulk of this paper is written.It is believed however, that the technique is adaptable to the purpose of the outside analyst who has to examine margarine and this application of the method is dealt with separately. It is also hoped that the instrument described will be of general’use in other fields of absorption photometry. The distinction between factory control work and outside analysis lies in the fact that the absorption at 620mp in a margarine test via the unsaponifiable matter is not wholly due to vitamin A ; a portion is contributed by other ingredients and the making of the necessary correction is simplified if the analyst has access to the raw materials. THE INSTRUMENT Yudkin7 and Hoch.s The principle common to all is the measurement by means of a photo-electric cell and a galvanometer of the absorption of filtered light by the test solution-a mixture of antimony trichloride and the unsaponifiable portion of the margarine in chloroform.The total cost of this apparatus* will be of the order of k50 i.e. probably less than a quarter of the cost of a visual spectrophotometer. In this instrument the light source is an adjustable horizontally mounted 6-volt 6-watt automobile lamp run from an accumulator. A rheostat and voltmeter enable the voltage to be kept constant. To minimise reflection from the glass of the bulb the emergent light has first to pass through a circular hole 46mm. in diameter i e . rather larger than the filament. The light beam then enters a simple biconvex lens whose focus is at the lamp filament i.e.about 55 mm. away and immediately beyond the lens is an iris diaphragm. The beam now slightly convergent and of adjustable cross-section traverses the filter and the reaction cell in that order. The filter is a 620 mp Wratten 26 sandwiched between 1 mm. Chance glasses OB.2 and ON.13. (When circumstances permit this filter could advan-tageously be replaced by an all-glass one made from a specially prepared melt exhibiting ab-sorption characteristics the exact inverse of those shown by the antimony trichloride - vitamin A complex.) After leaving the reaction cell (described below) the light impinges on a photo-electric cell whose output (of the order of 2 microamps.) is taken to a light-spot galvanometer fitted at the far end of the instrument.A shutter protects the cell when this is not in use. The galvanometer scale is double one reads normally from 100 to zero and registers per-centage transmission ; the other reads logarithmically in the same direction from zero to infinity and registers optical density. The components are housed in a wooden case with a cast-aluminium base and a top-which serves as an instrument panel-of Tufnol. The reaction cell carriage is also of Tufnol, a material whose particular advantage in this connection is its resistance to stains of the antimony trichloride solution. This carriage is lidded and holds two optical cells side by side and its lateral movement enables either cell to be brought into the light path. The cell holders accommodate either 8 cm.or 1 cm. cells but all the work here reported was done with the former which are about 6 cm. high by 3 cm. wide and in use require not less than 3 ml. of solution. A special funnel is provided to facilitate rapid and splashless filling of the “reaction” cell in the top of which it sits stably and does not need to be removed immediately after use as the hinged lid of the carriage fits comfortably over it. The carriage lid keeps out extraneous light and the instrument as a whole is designed to minimise the entry of stray light. The instrument (see Fig. 1) derives from several predecessors see for instance, * Manufactured by Unicam Instruments Ltd. Arbury Road Cambridge 22 OTHER APPARATUS AND REAGENTS BOWEX GRIDGEMAN AND LONGMAN A PHOTOELECTRIC PHOTOMETER Apparatus 250 ml.round-bottomed narrow-necked flasks fitted with 30 in. ground stoppered air condensers. 1000 ml. glass-stoppered separating funnels. 300 ml. flat-bottomed narrow-necked flasks. Distillation apparatus for ether recovery. This consists of a 300 ml. flat-bottomed wide-necked flask fitted with a 500 ml. separating funnel. The flask is connected through an Evans double-surface condenser to a 2000ml. conical flask which is fitted with a vertical double-surface condenser to prevent escape of distilled ether. 6 x 1 cm. test tubes marked a t the 3 ml. level. Burettes 10 ml. stoppered measuring cylinders and fine pipettes-including at least one 4 ml. pipette graduated in hundredths of a ml. and fitted with a capillary tip 5 cm. long. Reagents 50% aqueous potassium hydroxide solution.Absolute alcohol. Ether (B.P.) distilled fresh from powdered caustic soda as required. Hydroquinone. Chloroform (B.P.). Vitamin-A reagent. Acetic anhydride (pure). Before use wash with water dry with K,CO, filter through a fast paper and This is a 21-23% solution of antimony trichloride in chloroform purified as add 1% of absolute alcohol. above. If obtained ready made it must be repurified as described by Edisbury.l PREPARATION OF UNSAPONIFIABLE MATTER FROM MARGARINE It must be emphasised that the reliability of the technique as a whole depends first and foremost upon the care taken to ensure that all the vitamin A in the sample of margarine is mobilized in the extracted unsaponifiable matter. No detail is therefore spared in the following description of the process of extraction.It is an improved version of that described earlier from this laboratory (Edisburyl) and inter-laboratory trials have led us to believe that it is also an improvement on other methods to be found in the literature already quoted-and some of those methods have theoretically unnecessary or undesirable features such as preliminary removal of non-fatty constituents the addition of a slightly acidic antioxidant, the use of light petroleum or of a large excess of caustic soda (see the evidence of Benham9 on these last two points) and the creation of conditions that increase the ever-present danger of emulsification. Weigh 15 g. taken from the inside of the sample into a 250 ml. flask fitted with a ground glass stoppered air condenser add 10-20mg.of hydroquinone 12ml. of 50% aqueous potassium hydroxide solution and 35ml. of absolute alcohol. Boil gently for 15 minutes on a steam bath cool wash the condenser and stopper with distilled water and pour the contents of the flask into a litre separating funnel using a total of 150 ml. of distilled water for the transfer. Rinse the flask and air condenser three times with small quantities of ether, transfer the washings to the separating funnel and make up the volume of ether in the funnel to 150ml. Invert the funnel and swirl the contents with a single sharp movement of the wrist which should produce an intimate mixing of the two phases and repeat several times. Avoid vigorous shaking. If the phases do not separate within a period of two minutes the test should be rejected.Separate the layers and re-extract the aqueous-alcoholic phase twice with 60ml. quantities of ether. Wash the combined ether extracts four times with 50 ml. quantities of distilled water at about 30" C. using the same swirling motion as described above to mix the two phases. After removal of final wash water decant the ether phase into the separating funnel of the distillation apparatus run the ether solution into the flask at distillation rate i.e. so that the ether is removed at the same rate as the fresh solution enters. When about 1 ml. of solution remains in the funnel rinse with about 5 ml. of absolute alcohol and run it swiftly into the distillation flask. Remove the flask immediately from the apparatus and cork it.The unsaponifiable matter is now in alcoholic solution and may be kept in the dark for two or three hours if necessary before being tested. METHOD OF TEST WHEN CONTROL SAMPLES OF VITAMIN CONCENTRATE AND FAT BLEND ARE AVAILABLE In this section we assume that the analyst has been supplied with samples of vitamin-A concentrate (together with information regarding its potency based on E;trn 325 mp) an FOR THE ESTIMATION OF VITAMIN A I N MARGARINE 23 with the fat blend used in the manufacture of the margarine he is to test together with quantitative details. If the finished margarine contains carotene the appropriate quantity of the carotene concentrate must be mixed in with the control fat blend. Any dye that has been used may also be mixed into the fat blend although it is our experience that none of the margarine dyes known to be in use at the present time contributes any significant absorption in the region of 620mp.The water-soluble ingredients of the margarine are not required, because they are all removed during saponification. Prepare the unsaponifiable fractions from a 15 g portion of the margarine and from a 15-20 g. portion of the corresponding fat blend (plus carotene if present in the margarine) as described above. The instrument in a shaded and vibration-free position is prepared for use as follows: Switch on the current adjust the voltage to 6 and allow 10 minutes’ steadying time. With the photocell screened set the galvanometer spot exactly at “maximum absorption.” Place a reaction cell filled with chloroform in the light path open the screen and adjust the iris diaphragm until the galvanometer spot registers maximum transmission.The instrument is now ready. Remove the alcohol from the solution of unsaponifiable matter by heating on a steam bath in a stream of hydrogen. Still maintaining the gas flow cool the flask and add about 1-5 nil. of chloroform. Pour the solution into a small test tube having a mark at 3 ml. rinse the flask with three 0-5 ml. lots of chloroform and make up the solution to the 3 ml. mark. With a 06ml. pipette graduated in hundredths of a ml. and fitted with a capillary tip 5 cm. long transfer an aliquot of this solution (usually 0.25 to 0.35 ml.) to a clean reaction cell. (After a few tests this cell will become slightly frosted with antimony oxychloride; immediately clean the cell with 4 N hydrochloric acid rinse with distilled water and dry with acetone).Place the cell in the vacant holder in the carriage and move it across into the light path. Fit the funnel into the cell mouth. Allow time for the galvanometer to re-settle. Put 36ml. of antimony trichloride solution and two drops of acetic anhydride into a test tube then pour this solution into the funnel with a rapid smooth motion and quickly and carefully (to prevent vibration) close the lid of the carriage. Note the point of minimum transmission on the galvanometer. If this reading is less than 35 or more than 50 on the transmission scale (i.e. outside the limits 0.456-0.301 on the log or optical density scale) the test must be repeated with an appropriately greater or smaller aliquot of the solution of unsaponifiable matter so that the reading does fall within these limits.Repeat the test twice on each sample checking the zero and maximum transmission points on the galvanometer each time. The three readings for each sample should be very close (see Table I). In the case of the fat blend and in contrast to the vitaminised margarine, it will be observed that the galvanometer spot remains steady at its deflection point and does not almost immediately begin to move back. This is because the absorption is not transient as is that of the vitamin-A reaction mixture. Finally test the vitamin concentrate directly i.e. weigh out 1-2g. and dilute with chloroform to such an extent that an aliquot of less than 0.5 ml.gives a test reading within the limits prescribed above. Interpretation of Readings-The calculation of results proceeds as follows : Let D = reading on the optical density scale of the galvanometer. C = percentage concentration (of margarine or fat blend or vitamin concentrate) V = volume of test solution put in the (4 cm.) reaction cell. R = volume of reagent plus acetic anhydride added to the test solution in the cell. P = given potency of vitamin concentrate in I.U./g. in the chloroform test solution. 2D(V + R) cv * The term “effective” is used because the light transmission actually recorded includes rather more than that at 620mp-a certain quantity on either side depending upon the quality of the filter. It should be noted that the use of a filter instrument of this type prohibits the direct translation of absorption data to I.U./g.by means of standard conversion factors. If we abbreviate “effective” Et&. 620 mp as E the required result can be written as The “effective” EtZm 620 mp = E(marg.) - E (fat blend) E(conc.) Potency of margarine in I.U./g. = P 24 BOWEX GRIDGEMAN AND LONGMAPI; A PHOTOELECTRIC PHOTOMETEK Example : Weight of sample taken Diluted to . . Therefore concentration Volume of test solution in cell . . . . Volume of reagent and anhydride in cell Galvanometer reading (optical density) Therefore “E” = 2D(V + R)/CV . . An experimental margarine Marg. Fat blend Conc. 15.04 g. 20-26 g. 1-261 g. 3 ml. 3 ml. 25 ml. = C = 501.3% 675.4% 5.044% = v = 0.3 ml. 0.9 ml.0.4 ml. = R = 3.56 ml. 3.56 mi. 3.56 ml. = D = 0.400 0.385 0.455 0.02053 0.00565 1.786 -The given potency of the concentrate was 1560 I.U./g.* The potency of the margarine is therefore calculated as 0.02053 - 0.00565 = 13.0 I.u./g. 1.786 1560 x The convenience of the optical density scale will be apparent in these calculations; the density which is the log of the ratio of the intensities of the incident and transmitted light, is directly proportional to concentration and to cell thickness and density readings are therefore simpler to handle than the transmission percentages given on the ordinary galvano-meter scale. There are nevertheless two limitations to the use of either scale. First the error of reading has a minimum effect on the error of estimation when the transmission is 36.8% corresponding to an optical density of 0.4343; this can easily be derived from first principles-see for instance Twyman and Lothian.lo On either side of this position the influence of the reading error increases.Second it is a characteristic of the reaction mixture that its optical density is not strictly proportional to the concentration in the cell so that unless cell concentration is always confined within certain fairly narrow limits results will not be comparable. These two considerations have led us to define the working limits given above nix. 35% to 50% transmission 0.456 to 0.301 optical density. Accuracy Attainable-In order to assess the reliability of the method several series of margarine samples some from experimental batches and some from ordinary production, were tested.Three preparations of unsaponifiable matter were made from each sample and three or more photometer tests were carried out on each preparation. A typical set of results on one sample is shown in Table I. TABLE I REPLICATION OF ESTIMATES OF I.U./G. IN A SAMPLE OF MARGARINE Each result is calculated from a single reading Unsap. prep. No. . . . . 1 2 3 1st test . . 21.4 22.3 21.8 2nd , . . 21.4 22.0 21.4 3rd , . . 21.8 22.3 2 1.9 4th , . ,. 21-4 22.7 22.0 - - -Means . . 21-5 22.3 21-8 A typical set of results on one series of samples is given in Table 11. These samples were from batches of margarine made in the laboratory under carefully controlled conditions. The figures tabulated are the means of three photometer tests on each preparation of un-saponifiable matter.From all these experiments it has been possible to estimate the accuracy of individual photometer tests one test being defined for this purpose as the whole process from the taking of an aliquot of the solution of the unsaponifiable matter to the recording of the galvanometer reading. At the same time an indirect estimate of the accuracy of the method of extraction of the unsaponifiable matter was made. Statistical analysis revealed that the error of one test defined as above is represented by 52.3% for P = 0.95 while that of preparation of unsaponifiable matter is &4~5”/~ for P = 0.95. I t follows that the error of one reading on one preparation of unsaponifiable matter is i 5.0570 (the square root of the sum of the squares of the two component errors) for P = 0.95.This means that only once in 20 such estimates * The limiting factor in this method of arriving a t the potency of the margarine is clearly the relia-bility of the ascribed potency of the concentrate. If an ultra-violet spectrophotometer is available con-centrates can of course be checked for their E:2m. 325 mp values on which unitages are based. If not i t is advisable to calibrate the galvanometer scale with concentrates of warranted potency as described in the latter part of this paper FOR THE ESTIMATION OF VITAMIN A IN MARGARINE 26 should we be more than 5% away from truth. Normally of course more than one reading is taken and preferably more than one preparation of unsaponifiable matter is tested.TABLE I1 REPLICATION OF ESTIMATES OF I.U./G. IN FOUR SAMPLES OF LABORATORY-MADE MARGARINE Each estimate is derived from the mean of three photometer tests Vitamin A Vitamin A found I.U./g. added I A -l Marg. No. I.U./g. Unsap. 1 Unsap. 2 Unsap. 3 Mean I 10 9.9 9.2 8.9 9.3 I1 16 14.8 14.7 14.1 14.6 I11 20 19.0 19-2 18.8 19.0 IV 25 25.9 26.6 25.3 25-6 The influence of the two kinds of replication on the error of the mean result is shown in Table 111. TABLE I11 INFLUENCE OF THE NUMBER OF PREPARATIONS OF UNSAPONIFIABLE MATTER AND THE NUMBER OF TESTS ON EACH ON THE P 0.95 PERCENTAGE ERROR OF~THE MEAN RESULT No. of unsaps. 1 2 3 5 10 00 1 6.05 3.57 2-92 2.26 1.60 0 2 4-78 3-38 2.76 2.14 1.51 0 3 4.69 3-32 2.7 1 2-10 1.48 0 10 4-66 3-22 2-63 2-04 1.44 0 03 4.60 3.18 2-60 2.01 1.42 0 No.of tests on each unsap. It is suggested that 3 tests on each of 2 preparations of unsaponifiable matter giving an error of about -+3.3y0 for P = 0.95 should be made standard practice. The accuracy of the final result is also dependent on the error of the observations on the fat blend and the con-centrate both of which are less susceptible to error than is the actual margarine test and the former of which makes only a small contribution to the final error. If the margarine test is replicated as suggested above if three readings are taken on one preparation of fat blend unsaponifiable matter and if three readings are taken on the concentrate the error of the final result should be of the order of f 5y0 with f 6% the maximum range.METHOD OF TEST WHEN MARGARINE ALONE IS AVAILABLE If E:?= 620 mp is estimated from the antimony trichloride reaction with the unsaponi-fiable fraction of a margarine and no further information is available the question arises, how much of the absorption can be attributed to vitamin A? Experience has shown that constituents other than vitamin A may contribute anything up to in exceptional cases, 50% of the total E value. Fortunately the absorption due to the vitamin-A reaction product has a peculiar characteristic that can be used to distinguish it quantitatively from the irrele-vant non-vitamin-A absorption. The characteristic is that the absorption is transient ; it declines rapidly at first and more slowly later eventually reaching a minimum that is almost invariably 10% of its initial maximum value.The fading period is about 20-30 minutes. The difference between the “gross” initial absorption and the minimum absorption is known as the “nett” absorption; thus for pure vitamin A and high-potency concentrates the nett value can be taken as 90% of the gross value. But for low-potency materials in which there is a significant non-vitamin-A contribution to the absorption at 620 mp the value is a little less than goyo for two reasons (a) the recorded gross value is not the simple sum of the values for the vitamin-A and non-vitamin-A fractions; there is a slight inhibition of total colour development when the two fractions conjoin and (b) the non-vitamin-A con-tribution increases slightly during the fading period.The accompanying diagram (Fig. 2) illustrates these points. The above applies to the absorption at 620 mp only i e . as estimated on a spectrophoto-meter. When a light-filter instrument is employed the recorded absorption embraces a small region of the spectrum maximised at 620mp. The correction factors will therefore not necessarily be the same and each instrument must be calibrated. For this purpose three or four vitamin-A concentrates of the type used in margarine are required. The potency of each in I.U./g. based on E:& 325 mp and a conversion factor of 1600 must be knowii 26 BOWEN GRIDGEMAN AND LOSGMAN A PHOTOELECTRIC PHOTOMETER Because the “fading” test necessitates the taking of galvanometer readings well outside the recommended limits it is advisable to use the concentrates to calibrate the whole scale-or at least between the 30% and 90% transmission points.(Incidentally as a calibration curve has to be used for the interpretation of readings in this type of test the straight trans-mission scale only need be used.) At the same time the nett gross ratio (0.9 1.0 for pure vitamin A and E 620 mp only) for the instrument can be ascertained. A weighed amount of each concentrate should be diluted in two stages to the strength required to produce a galvanometer reading of about 30 and the exact reading noted. Leave the cell in position,-preferably with a small glass cover to prevent evaporation losses-for NETT VALUE -Time Fig. 2. Typical course of absorption of SbCl reaction on the unsap. of an A-vitaminised margarine.about half-an-hour during which time the transmissibility will rise to about 80 i e . the colour and absorption of the solution fade to a minimum at this point. Then make up a series of weaker solutions (from the original first-stage dilutions)-say SOYO SOYo 40% and 10% of the strengths that gave readings of about 30. Only the initial gross readings need be recorded for these. From all the gross readings a mean calibration curve may now be con-structed relating gross readings to I.U./ml. of cell solution. A curve such as that in‘Fig. 3 will be obtained. This curve may now be used to check the value of the nett figures obtained by fading tests on the concentrates in terms of the gross figures; as already indicated they should be approximately 90%.Let us suppose that a mean value for all the concentrates tested of S9yo is obtained. This value can then be applied to the “fading” test on a margarine as shown in the following example : The unsaponifiable matter from 15 g. of margarine was dissolved to 3 ml. in chloroform. A reaction cell containing 0.2 ml. of this solution plus 3.5 ml. of reagen FOR THE ESTIMATION OF VITAMIX A IN MAKGARINE 27 On From the plus 0.06ml. (2 drops) of acetic anhydride gave a galvanometer reading of 42. fading the furthest point on the scale that the light spot reached was 76-5. graph (Fig. 3) these values were read off: Equivalent I. U. Transmission per ml. in cell 42 6.56 76.5 1.70 4.86 -I 1 I I t 1 I ¶ I 1 30 40 50 60 70 80 9 Percentage Transmission and strength of Vitamin A in reaction cell.Fig. 3. Relation between galvanometer readings (yo light transmission) This difference 4.86 I.U. per ml. of cell solution is clearly the “nett” which as already indicated corresponds to a true vitamin-A-only gross value of 4-86/0439 I.U. per ml in the cell. Now the cell contained altogether 3-76 ml. of which 0.2 ml. was an aliquot of the 3 ml. solution of the unsaponifiable matter from 15 g. of margarine. Therefore the required potency of the margarine is 4086 3*76 = 20.6 I.U,/g. 0.89 x 0.2 X 15 It should be noted that we have made no allowance for the slight over-rating of the nett/gross factor (here 0439) due to inhibition etc. the reason being that we cannot be con-fident of its precise value in any particular margarine.By ignoring it however we are unlikely to underestimate by more than 5% a figure that is within the experimental error of the method. SWMMARY Vitamin-A potency is normally expressed on the basis of Ei:m. 325 mp but this value cannot be successfully determined on margarine. The best alternative is to measure the absorption at 620 mp of the antimony trichloride - vitamin A complex and an improved type of photoelectric photometer has been desiwed for this purpose. The margarine is tested via the unsaponifiable matter and the instrument is calibrated with margarine-type vit amin-A concentrates 28 BOWEN GRIDGEMAN AND LONGMAN A PHOTOELECTRIC PHOTOMETER The principal case considered is that in which control samples of the particular con-centrate and the unvitaminised fat blend are available.The other case is when the margarine sample alone is available and is met by the “fading technique,” i e . advantage is taken of the fact that of the gross registered absorption at 620 mp the fraction due to vitamin A declines by about 90% over a period of approximately 20-30 minutes. The authors wish to express thanks to Dr. J. R. Edisbury Mr. R. J. Taylor and Mr. A. L. Wilkie of these laboratories for their help; to Professor R. A. Morton and Mr. T. W. Goodwin, of the University of Liverpool for much useful discussion; and to the directors of Lever Brothers & Unilever Limited for permission to publish this paper. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. REFERENCES Edisbury J. R. ANALYST 1940 65 484. Oser B. L. Melnick D.and Pader M. I n d . Eng. Chem. Anal. Ed. 1943 15 724. Neal R. H. and Luckman F. H. Ibid. 1944 16 358. Wilkie J . B. and De Witt J. B. J.A.O.A.C. 1945 28 174. Evelyn K. A. J . Biol. Chem. 1936 115 63. Dann W. J. and Evelyn K. A. Biochem. J . 1938 32 1008. Yudkin S. Ibid. 1941 35 551. Hoch H. Ibid. 1943 37 425. Benham G. H. Canad. J . Res. 1944 22 21. Twyman F. and Lothian G. F. Proc. Phys. SOC. (Lond.) 1933 46 643. CENTRAL TECHNICAL DEPARTMENT LEVER BROTHERS & UNILEVER LIMITED PORT SUNLIGHT CHESHIRE DISCUSSION ON THE PRECEDING TWO PAPERS Mr. J . I. M. JONES said that the rationale of the conversion factor merited careful consideration. The margarine manufacturer based his calculation of the amount of vitamin concentrate to be added to his margarine on a potency in international units given by the supplier of the vitamin concentrate.This potency was determined from E:2m. 328 mp on the whole concentrate. When the margarine was analysed it was subjected to saponification and the Carr-Price reaction was used for estimating the vitamin A from E;?,.,,. 617 mp. It was therefore necessary to relate the result obtained by the latter process to that obtained by the former. This had been done by testing the vitamin mixture as added to the fat blend by both methods. The factor necessary to convert the E:trn. 617mp values to yield the same value in inter-national units as those given by E::m. 328mp x 1600 was then found to vary between 560 and 590 on different concentrates and to average about 565 and gave values of approximately 18 I.U.of vitamin A per gram of fresh margarine in samples examined in his laboratory. A considerable amount of work had been done on loss in the analytical procedure which it was hoped to publish in due course. Mr. Jones enquired what steps had been taken in the instrument shown to ensure stability of the light source and absence of drift in the galvanometer. Dr. H. WILKINSON said he would like to agree with the remarks made by Mr. Jones concerning con-version factors. A t Port Sunlight they had found after examining a large number of maigarine con-centrates that the average E-620/E-325 ratio was 2.85. Professor Morton had used the ratio 3.1 which obtains for pure vitamin A in calculating his 510 conversion factor for the gross and nett E-620 value. If one took the ratio 2-85 and used the conversion factor 1600 for the E-325 \,slue which is standard commercial practice then one obtained a conversion factor 570 for the gross 13-620 which was very close to the 565 quoted by Mr.Jones. They had found in their laboratories that for the nett E-620 value determined on the unsaponifiable matter of margarine a factor of 680 gave the best results. This value was attained by consideration of the fact that pure vitamin A gives a nett figure only 89% of the gross E-620 value and also by making an allowance of 74% for inhibition and contribution; the 73% value was at present sub judice as they have not sufficient results available a t the present tiine to consider it absolute. One other point he would like to mention concerned the method of chromatographic.separation of carotene and vitamin A on de-fatted bone meal. When Mr. Mann published his method they used it on the unsaponifiable matter from margarine and found that the dye associated itself with carotene and they were unable to obtain quantitative elution of the vitamin A. I t should be noted that these results refer only to tests with the unsaponifiable matter. He wondered whrther Professor Morton had any details which he could give on the matter although he (nr. M ilkinson) felt sure that Professor Morton would not recommend the chromatographic separation as regular routine but would prefer to determine the vitamin A on a separate portion of the unsaponifiable matter. Lastly he would like to take this opportunity of publicly thanking Professor Morton end Mr.Goodwin, on behalf of his colleagues and himself for the help they had given them in designing the instrument which his colleague had just described. Dr. H. E. Cox asked whether much destruction of vitamin A could occur during the spectrographic observations owing to the presence of ultra-\iolet light. Mr. J . V SMART enquired if the authors had had any cxperience of a new reagent for vitamin A, g1ycero1-1~3-dichlorohydrin (cj. this Vol. p. 281 which was said to give a purple colour much more stable than the colour given by antimony trichloride. More work was at the present moment in progress FOR THE ESTIMATION OF VITAMIN A IN MARGARINE 29 Mr. J. H. HIGH asked whether the Spekker absorptiometer could be used. Rapidicy of setting and reading elimination of trouble from fluctuations in the light source and ability to allow for “drift” (.~NALYST 1943 68 78) were its chief advantages.Mr R. A. C. ISBELL said there was no fundamental reason why the Spekker absorptiometer should not be used for vitamin A or carotene determinations with suitable filters. Hilgers did not specifically recommend i t for this purpose as i t seemed unnecessarily expensive for an instrument for a limited use. They had therefore developed a simplified version of the “Spekker,” which is a single beam type similar in some respects to the instrument which had been described. The light source is a 6 volt 18 watt headlamp energised from a high capacity accumulator or a stabilising transformer. By carefully selecting filters it has been found possible to make a combination with sufficient transmission to enable a Unipivot type needle galvanometer to be used.Instead of an iris diaphragm they preferred to use a potentiometer form of control for setting the full scale deflection of the galvanometer as this has been found to give smoother variation. With regard to the drift of the photocell mentioned by an earlier speaker, it is possible to select photocells with minimum fatigue. The type of photocell used may have either positive or negative drift and i t follows that cells can be found in which this effect is very slight. They had for some years selected such cells for special purposes. Mr. K. A. WILLIAMS remarked that the instrument demonstrated by Mr. Gridgeman and that men-tioned by Mr. Isbell were almost exactly the same as the one described to the Society ten years before by E.R. Bolton and himself (ANALYST 1935 60 447). He had used that instrument very successfully for blue value determinations for some years and could confirm that i t possessed great advantages over the visual method not only in speed of operation but also by making i t possible to work near to the head of the absorption $and of the test solution. In the early days of the use of photoelectric cells in this way some doubt had arisen as to the validity of results owing to the limitations imposed on selecting the operating wavelength through the use of dyed gelatin filters. He would be interested to hear whether it had proved possible for the authors to narrow the spectral range of their filter more than had earlier been possible.He noted that the vitamin A content of a margarine showed an initial falling off which might well be attributed to effects of the manufacturing process; this was followed by a comparatively long period in which there was little change and this period seemed to correspond to the known long period in which edible fats remained sweet; it was probably the result of the action of the natural or added inhibitors present. Finally there appeared a period in which the vitamin content dropped quickly and he thought that this would be found to correspond to the final deterioration of the fat and onset of rancidity. The industry could congratulate itself that during the difficult time of the war fats had proved as well refined as in peace time and to have very similar keeping properties Where they had shown themselves to keep a little less well it was probably due to the great changes that had taken place in transport and storage; this was perhaps the price that had to be paid for progress in such matters.Mr. J. G. LUNT noted that Mr. Goodwin had suggested the possibility of selecting the constituent oils so as to achieve good keeping quality and would like to ask on what basis such selection should be made. His firm were continually selecting new oils and having trouble with their suppliers on that account. Mr. A. E. PARKES enquired if any special precautions had to be observed in the saponification process, and why acetic anhydride was used with the antimony trichloride reagent. Professor MORTON replying to Dr. Cox said that the risk of destroying vitamin A photochemically during the spectrographic examination was not serious. If the potency is approximately determined first (e.g. by means of the colour test) the optimum concentration and cell thickness for the ultra-violet test can be foreseen sufficiently closely to limit the period of exposure to ultra-violet light to a short time. The proposed new test referred to by Mr. Smart had not been tried yet a t the Liverpool laboratory. The question of light filters arose frequently in discussions concerning photoelectric colorimetry. The interference filters associated with Geffcken transmitted very narrow strips of the spectrum and it was hoped that before long such filters would become more easily available. Mr. GOODWIN in reply to Mr. Parkes said that acetic anhydride (2 or 3 drops in 4 ml.) is used in the test solutions to remove any water which might make the solutions turbid. Even a slight cloudiness introduces errors. Its chief disadvantage was that of cost