680 Arcai’yst, October, 1968, Vol. 93, $@. 680-687 The Automatic Determination of Original Gravity of Beer Part 11.” The Determination of Alcohol and Gravity Lost BY R. SAWYER AND E. J. DIXON (Laboratory of the Government Chemist, Cornwall House, Stamford Street, London, S.E. 1) The determination of gravity lost on samples of sound beers by an automatic distillation procedure, followed by determination of alcohol in the distillate, is presented and discussed. The correlation between alcohol content and gravity lost, determined by distillation, is assessed, and the resulting relationship is used to determine gravity lost in a series of samples. Summation -of results with those obtained in Part I indicates the accuracy of the determination of total original gravity by the method proposed, and is shown to be in agreement within f2 units of original gravity with the distillation procedure; a mean standard deviation of 0-37 is obtained on replicate analysis.THE reasons for the development of an automatic method for the determination of original gravity of beers have been elaborated in a previous paper1 concerned with the determination of the “extract gravity.” This paper is concerned with the direct automatic determination of alcohol in the sample, the value for which is used to calculate the “gravity lost” during fermentation. The conversion of alcohol content via distillate gravity into “gravity lost,” from the classical distillation method, is defined in the Mean Brewery Table, and is consolidated in the Customs and Excise Act.2 One method for the automatic determination of alcohol in samples of beer has been described by A~hurst.~ A colorimetric method, based on the formation of red, solvent-soluble alcoholates of vanadium 8-hydroxyquinolinate, has been described by Tanaka,4 and this method has been used for determinations of alcohol in b l ~ o d .~ , ~ Enzymatic methods based on the use of alcohol dehydrogenase for automatic determinations of blood alcohol have also been reviewed and The use of such enzymatic techniques was regarded as too costly for work on the scale proposed in our laboratory, and no further consideration was given to this method. PRELIMINARY EXPERIMENTAL WORK- The method of Ashurst, in which the red hexanitratocerate alcoholate colour was used, was considered initially, but the major disadvantage of this method was that a knowledge of the specific gravity of the beer sample was necessary so that the sample could be diluted manually to within a fixed gravity range.This procedure was necessary because carbo- hydrates interfered with the hexanitratocerate reaction. Dilution to a fixed specific gravity range reduced the residual sugar concentration of the sample to a level at which almost complete oxidation of the reducing sugars with alkaline ferricyanide occurred before the reaction with the alcohol reagent removed the interference. The procedure depended on the dialysis of alcohol away from the ferricyanide reaction mixture, used for oxidation of sugars, before the reaction with hexanitratocerate. It was considered that the manual step required in this procedure negated some, if not most, of the advantages of an automatic technique.Attempts were made to modify this procedure to allow samples to be treated wholly by the automatic technique. It was found, however, that, at the increased concentration of alkaline ferricyanide necessary to oxidise the reducing sugars in the high gravity beer samples, a fraction of the ferricyanide reagent dialysed across the dialyser membrane together with the alcohol. This ferricyanide concentration was sufficient to cause variable reaction with the hexanitratocerate; the extent of reaction was inversely proportional to the amount of ferricyanide used in the sugar oxidation. Transfer of colour across the dialyser membrane was also evident with dark samples.For these reasons further work on this method was discontinued. * For details of Part I of this series, see reference list, p. 687. 0 SAC; Crown Copyright Reserved.SAWYER AND DIXON 681 Further attention was then paid to the Tanaka method, which depends on the transfer of alcohol from the aqueous phase into a solvent containing vanadium 8-hydroxyquinolinate where the red alcoholate is formed from the blue - black reagent. Under the conditions necessary to measure alcohol derived from beer in any of the solvents (nitrobenzene, dichloro- benzene, chloroform) compatible with the system, it was necessary to heat the two-phase reaction mixture t o 50" C to produce a measurable colour in the organic phase in a practicable time. On separation of the phases following the heating cycle, removal of dissolved water from the organic phase was incomplete and this caused noisy traces.Addition of acetone or dioxan to the organic layer after separation rendered the water droplets miscible with the solvent but also introduced residual sample colour. The use of automatic vapour sampling techniques has been described in the determina- tion of hydrogen fluoride after di~tillation,~ metabolic carbon dioxidelo and carbon dioxide derived from antibiotic assays with micro organisms.11 The method of continuous distilla- tion of volatile aldehydes and ketones described by Duncornbe and Shawl2 has been adapted to suit our needs. EXPERIMENTAL APPARATUS- In addition to the units elaborated in Part I,l the apparatus required for the operation of the combined manifold comprises a distillation unit mounted in a 95" C heating bath, a proportioning pump, a colorimeter with an 8-mm flow cell and 610-mp filters, a range expander to second pen on a two-pen recorder and two standard delay coils.-€== Cool i ng water 1( Segmented sample Water Waste t Glass bead: 0.025 inch i.d. pump tube ,030 x0-048 inch polyth,ene Re-sample t o pump Heating bath 95°C .Air 300 ml/min Fig. 1. Distillation flow path The distillation unit developed is illustrated in Fig. 1. The assembly consists of three vertically mounted glass coils, the lower ends of which meet at a T-junction in a 95" C oil-bath. A stream of nitrogen is introduced via one coil and meets a hot segmented stream of liquid set up from a series of samples in the second coil.A flash distillation of alcohol is achieved, and the vapour is carried through the third beaded coil to the outside of the heating bath. The vapour and residue are then separated in the glass separator illustrated, and a sample of vapour is carried by the gas stream to a condensing train fed at a T-junction with distilled water. The sample vapour is condensed in a standard coiled condenser and the emergent gas and liquid are separated The undistilled residue and some gas are rejected to waste.682 SAWYER AND DIXON : AUTOMATIC DETERMINATION [A%a@Si!, VOl. 93 3.9 Sample 30/hr Debubble to waste --- A I d r I---- - --- Time delay coils I Waste Colorimeter 6 I0 m.p, 0.8-cm flow cell c I I I 0.8 Air 8 1 4 3 *4 Pumps mt/min Aand B Range Recorder expander Fig.2. Manifold for alcohol 4 4 at a modified Technicon B2 micro trap; all the gas and some condensate are rejected to waste and a sample aliquot is pumped back into the manifold for colorimetric analysis with an acid dichromate finish. In normal operation the wash chamber of the sampler is supplied with distilled water. This ensures that the distillation unit is cleaned between samples and that no insoluble residue is deposited in the glass coils. A diagram of the manifold as a separate assembly is shown in Fig. 2. In the full method a combined manifold incorporating the I) r b Range expanders Colorimeter stack Recorder and volt sta bi I iser on lower rack Distillation unit Pump A Fig. 3. Block schematic layout of operating units of AutoAnalyzer assembly for combined manifoldOctober, 19681 OF ORIGINAL GRAVITY OF BEER.PART I1 683 sugar determination is used and operated from a common sample line to a stream splitter; the distribution of pumping tubes between three pumps, A, B and C, is indicated. A block schematic layout of the various units is shown in Fig 3; pump A controls the sampling and primary diluting stages, pump B controls the alcohol end method, pump C controls the sugar end method. The principles of manifold design elaborated in Part I are also applied in this method to obtain good discrimination at 30 samples per hour. METHOD REAGENTS- Acid dichromate-Suspend 600 g of analytical-reagent grade potassium dichromate in 6 litres of distilled water; add to this 2.5 litres of concentrated analytical-reagent grade nitric acid.Shake to dissolve and dilute to 10 litres with distilled water. Standard solzctiorts-Prepare standard solutions from analytical-reagent grade absolute ethanol in the range 2 to 10 per cent. v/v ethanol. The specific gravity of a 10 per cent. v/v solution, 0.98659 6c 9 is used as a check on quality. In practice these standards also contain sugar for the sugar method. PROCEDURE- The samples are prepared as indicated in Part I1; an aliquot of the sample is taken from a stream splitter, which is connected to the sample probe of the sampler and to the sugar manifold. The aliquot for alcohol analysis is pumped at the rate of 3.9 ml per minute; this is immediately re-cycled via a de-bubbler to remove the air pulse obtained when the probe is out of solution.A 2.9 ml per minute aliquot is taken, segmented and passed to the 95" C distillation unit (Fig. 1). The sample stream is met by an air stream at 300 ml per minute, which is obtained from a cylinder supply through a reducing valve at 40 p.s.i. and stabilised by passing through a molecular-sieve filter. The vapour generated in this system is passed through an ascending coil filled with glass beads and emerges at the lower jet of the separator as a spray that is directed to the lower part of the bulb. The undistilled residue is withdrawn with some gas from the bottom end of the separator via a cooling coil and pumped to waste at 6.8 ml per minute. The separated gas and vapour stream is collected at the jet, directed towards the dome of the separator, and is passed to a T-piece where it meets a stream of de-ionised water pumped at 7.8 ml per minute.This mixed stream is condensed and passed to a second gas - liquid separator, which is a modified Technicon B2 micro seDarator unit. The excess of gas is vented to waste via the internallv sealed jet (60 F) togetherAwith most of the condensed lcquid. A sample for analysis is withdra& from ihe Wavelength, mp Fig. 4. Spectrum of alcohol reagent system for A, reagents against water; B, reagents pZus 11 per cent. of ethanol against water in a 1-cm cell684 SAWYER AND DIXON : AUTOMATIC DETERMINATION [Analyst, Vol. 93 70 I 6.0 b pool of liquid at the lower end of the separator. This pool is of small volume (less than 0.2 ml) and is continuously flushed by the condensate and gas to the waste stream; the concentration of alcohol remains representative of the original samples in sequence.A sample aliquot is withdrawn at 0-16 ml per minute and this is passed via an H, connector to an air-segmented stream of acid dichromate pumped at the rate 2.9 ml per minute, and the colour is developed and read at 610mp in an 8-mm flow cell. Fig. 4 illustrates the spectra of the reagent- sample mixtures and indicates that at the wavelength chosen the reagent system alone has minimal absorbance. The reaction time path is adjusted to give near coincident emergence 4.0 3.0 2-0 80 70 60 J 50 - 40 - 30 .20 - - 11.0 I Fig. 5. Ethanol standards of the alcohol and sugar peaks for each sample, the overriding factor being the time taken for the sugar method.A series of standard ethanol peaks is illustrated in Fig. 5, and a typical calibration on x2 range expansion is shown in Fig. 6. A total analysis time of 25-minutes from sampler to recorder is required for both channels. Alcohol, per cent.v/v Fig. 6. Calibration curve of ethanol, per cent., against full scale deflectionOctober, 19681 OF ORIGINAL GRAVITY OF BEER. PART I1 685 RESULTS With the manifold assembly detailed operating at optimum sampling speed of 30 samples per hour, a study has been made of the correlation between the “gravity lost,” as deter- mined by the distillation procedure, and the alcohol, as determined by our method. A total of 80 samples of different beers was used with a range of “gravity lost” from 13.8 (2.2 per cent. v/v of alcohol) to 54.9 (8.6 per cent.v/v of alcohol) and the relationship between “gravity lost” and percentage of alcohol found to be- where G = gravity lost expressed in the 1000 notation and The statistical constants found for this line are regression coefficient r = 0.996 and standard error of the estimate of gravity lost from alcohol = 0.79. The range of differences between calculated gravity lost and the observed figure by the distillation procedure for the set of samples was -1.7 to +16. TABLE I COMPARISON OF RESULTS OF THE DETERMINATION OF GRAVITY LOST (G) BY THE PROPOSED G = 6.56 A - 0.3 A = percentage of alcohol. AND THE OFFICIAL METHODS Mean gravity lost Mean difference Standard f-*-, between pairs deviation samples from alcohol distillation observed differences Number of Calculated Observed by calculated - of Type Milk stout .. .. 6 28.1 28.9 - 0.8* 0.37 stout . . .. .. 5 20.0 19-6 + 0.4 0.20 stout . . * . .. 5 17.3 16.8 +0*5t 0.13 Brownale .. .. 5 16.6 16.4 + 0.2 0.19 Continental lager . . 11 34.2 33-8 + 0-4 0.48 Continental lager . . 12 18.4 18.8 - 0.4t 0.13 Continental lager . . 12 28.1 27.9 + 0.2 0.32 U.K. lager . . .. 12 29.3 29.5 - 0.2 0-29 0.37 ::::i 0.46 Pale ale . . .. . . 12 29.4 28.6 Strong ale . . .. 6 58.9 57.3 * Significant a t 1 per cent. probability level. t Significant at 0.1 per cent. probability level. Following the scheme outlined in Part I, a series of samples was chosen to test the difference between calculated and observed values for particular varieties, and the results of this examination are shown in Table I, together with the standard deviation for the differences in each series.Probability levels of significance of differences obtained by use of Student’s t-test are also indicated. Again there is evidence of a difference characteristic TABLE I1 COMPARISON OF RESULTS OF THE DETERMINATION OF ORIGINAL GRAVITY BY THE PROPOSED AND THE OFFICIAL METHODS Mean original gravity f-*7 Mean difference *Calculated between pairs Number of from sugar Observed by calculated - Type samples and alcohol distillation observed Milk stout . . .. 6 56-3 57-0 -0.7t stout . . .. .. 5 46-3 44.4 + 1-93 stout . . .. .. 5 34.0 33.2 +@a$ Continental lager . . 11 52.4 51.5 + o.9t Continental lager . . 12 29.2 29.3 - 0.1 Continental lager . . 12 42-4 41-7 +0*7$ Pale ale .. .. .. 12 46.0 44.9 +1-1$ Brown ale . . .. 5 33.6 33.0 + 0.6 U.K. lager . . .. 12 44.7 44.1 +0*6$ Strong ale . . .. 6 86.5 87.0 -0*5$ * Significant figures on the 1000 notation scale are quoted (e.g., 56.3 = 1056.3). t Significant a t 1 per cent. probability level. Significant a t 0.1 per cent. probability level. Standard deviation of differences 0.34 0.33 0.19 0-46 0-50 0.23 0.34 0-40 0.58 0.17686 SAWYER AND DIXON : AUTOMATIC DETERMINATION [Analyst, Vol. 93 of a particular label. The sums of the two component parameters of the original gravity determination derived from those values shown in Part I for “extract weight” and for “gravity lost” are shown in Table 11, together with the standard deviation of the differences between the original gravity by direct distillation and original gravity calculated from sugar and alcohol contents.The sum of the two parameters lies between +2-0 units of original gravity in all these cases, and the weighted mean standard deviation of the differences is 0.37. As may be expected, a few types of beer fall outside the limits k2.0 units of original gravity indicated above; these are types that are infrequently met in practice and include such extreme varieties as the dry diabetic lagers and some high gravity porters and stouts. These samples are again consistent in their behaviour and due allowance can be made in any screening technique applied. TABLE I11 RESULTS OF ANALYSIS OF ROUTINE SAMPLES BY THE TWO METHODS Type Continental lagers U.K. lager . . .. Pils .. .. .. Brown ale . ... stout . . .. Ales . . .. .. Number of samples with same *Nominal label 10 1 1 3 1 4 4 11 3 10 3 1 1 2 6 4 1 20 1 3 1 1 1 2 1 3 2 1 1 2 2 1 1 1 1 1 gravity 30 30 42 30 48 81 44 44 45 33 34 39 37 57 44 38 48.7 55 45 42 41 56 68 48 47 45.5 35 37 45.5 40 44 56 63 76 87 100 *Gravity found by traditional method 1 Range 29.2 to 30.0 - - 30.3 to 30.8 77.9 to 79.9 42.5 to 43-5 43.5 to 45.1 43.8 to 44.4 31.8 to 33-8 33-0 to 33-52 - - - 55-6 to 57.6 43.2 to 44.1 36-1 to 37.4 52.7 to 55.4 41.3 to 42.0 - - - - - 46.5 to 47.6 44-2 to 45.0 34.7 36.6 43- 1 39.8 to 40.0 43.7 to 43.9 - - - - - - Average 29.71 28.6t 42.4 30.5 42.8 78.6t 43.0 44.5 44.lt 32.8 33.1 40.7 36.6 56.6 43-4 36.5 46.8 54.3 45.6 41.8 40.6 55.9 66.4 47-0 47-8 44-5 34.7 - - 39-9 43.8 55.9 62.4 75.8 85-8 99-9 *Gravity found by proposed method r Range 28.5 to 29.5 - - 29-5 to 30.3 78.6 to 80.4 43.6 to 44.8 43.5 to 44-6 43.1 to 44.1 31.9 to 33.5 34.7 to 35.1 - - - 64-5 to 56.2 42.9 to 44.3 36.52 to 37.2 63.2 to 55-0 41.9 to 42.9 - - - - - 47-9 to 48.5 44.4 to 45.4 34.4 to 34.7 - - - 40.0 to 40.4 44.0 to 44.2 - - - - - * Significant figures on the 1000 notation scale are quoted (e.g., 30 = 1030).t Determined by distillation. Average 29.3 27 9 41.8 30.0 42.1 79.5 44.4 44.0 43.6 32-8 34.9 39.1 35.1 55.4 43.5 36-8 46-6 64.0 44.5 42.4 41.3 57.1 67.6 48.2 46.9 44.9 34-6 37.1 42.9 40.2 44.1 57.1 61.6 74.4 86.6 96-4 Table I11 shows a summary of a typical set of analyses on 112 routine samples examined by the proposed method and by the normal distillation or refractometric13~14 methods of analysis; declared values of original gravity are also shown for comparison.With one exception, a heavy gravity ale, the agreement between the two methods is within the limits expected; it is also noteworthy that the two methods of analysis indicate the one sample in the series for which declaration is grossly in error.October, 19681 OF ORIGINAL GRAVITY OF BEER. PART I1 687 Further work on the automatic reading of peak heights and automatic calculation of the results of analysis is in hand and will be reported in a further paper. The authors wish to thank Miss L. M. Grisley for much of the experimental work on the manual methods, and the Government Chemist for his permission to publish this paper. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. REFERENCES Sawyer, R., and Dixon, E. J., Analyst, 1968, 93, 669. Customs and Excise Act 1952, The Beer Regulations 1952, S.I. 2232, H.M. Stationery Office, Ashurst, P. R., J. Inst. Brew., 1963, 69, 457. Tanaka, M., Talanta, 1960, 5, 162. Eskes, D., Mikrochim. Acta, 1965, 5, 1065. Van Gent, P. K., and Kerrich, J. E., Analyst, 1965, 90, 335. Bonniehsen, R., in Bergmeyer, H. V., Editor, “Methods of Enzymatic Analysis,” Academic Press, Goldberg, L., and Rydberg, U., in “Automation in Analytical Chemistry, 1965,” Technicon Mandl, R. H., Weinstein, L. H., Jacobson, J. S., McCune, D. C., and Hitchcock, A. E., in op. cit., Becattini, U., Citi, S., Cangi, G., Gabbrielli, G. F., and Loddi, L., in “Automation in Analytical Shaw, W. H. C., and Duncombe, R. E., Analyst, 1963, 88, 694. Duncombe, R. E., and Shaw, W. H. C., in “Automation in Analytical Chemistry, 1966,” Volume 11, Simmonds, C., “Alcohol : Its Production, Properties, Chemistry and Industrial Applications,” Schild, Von E., and Irrgang, G., Brauwissenschaft, 1956, 9, 314. London. New York and London, 1965, p. 285. Symposia, Mediad Inc., New York, 1966, p. 595. p. 270. Chemistry, 1966,” Volume 11, Technicon Symposia, Mediad Inc., New York, 1967, p. 257. Technicon Symposia, Mediad Inc., New York, 1967, p. 15. Macmillan and Co., London, 1919, p. 473. NOTE-Reference 1 is to Part I of this series. First received August 25th, 1967 Amended October 16th, 1967