218 J. CECIL JONES: ESTIMATION OF METHYL ALCOHOL ESTIMATION OF METHYL ALCOHOL IN PRESENCE OF ETHYL ALCOHOL. BY G. CECIL JONES, A.C.G.I., F.I.C. MANY methods for the detection and approximate estimation of methyl alcohol in presence of ethyl alcohol have been based on the oxidation of the former to formal- dehyde, for the detection of which many delicate reagents are available. So delicate are these reagents for formaldehyde that most extravagant claims have been made for the corresponding methods for the detection of methyl alcohol, regardless of the fact that it is impossible to oxidise methyl alcohol to formaldehyde without simul- taneously oxidising ethyl alcohol to acetaldehyde, which reacts with many of the reagents, and difficult to avoid some oxidation of ethyl alcohol to formaldehyde itself.These facts should have been generally known at least since Scudder published his review (J. Amer. Chem. Soc., 1905, 27, 892) of the methods available for the estimation of methyl alcohol in presence of ethyl alcohol, but scarcely a year elapses without the appearance of one or more methods of this type, novel-if at all-only in the choice of reagent for formaldehyde, and generally accompanied by claims to delicacy which cannot be substantiated. Until the appearance of Denigds’ method in 1910, it is doubtful if any of these methods could be depended on to detect lessIN PRESENCE OF ETHYL ALCOHOL 219 than 2 per cent.of methyl alcohol in strong spirit. Hinkel (ANALYST, 1908, 33, 417) put the limit of such methods at 5 per cent.Clearly a method which could only just differentiate between ethyl alcohol and industrial spirit, and could not detect adulteration of the former with 50 per cent. of the latter, was of little practical use. Denigds’ method (Compt. rend., 1910, 150, 832), framed with due regard to the facts mentioned above, constituted a very notable advance. The conditions of oxidation are so chosen that ethyl alcohol never gives rise to enough formaldehyde to affect the reagent used for the detection of that substance, and the final test- with Schifs solution-is made in so strongly acid a medium that acetaldehyde develops no coloration.A much more convenient practice is to dilute the liquid to be examined until it contains 10 per cent, of total alcohols, and to operate on 5 C.C.of this solution, as direcbed by Simmonds (ANALYST, 1912,37,16). This author’s directions for carrying out DenigBs’ method need only trifling amendment, but the writer, who had some difficulty before he discovered the limitations of the method and could confirm the claims made for it, thinks the following notes may prove useful : Simmonds’ directions are very brief, and it will be convenient to reproduce them here.He writes : The alcoholic liquid is first purified, where necessary, either by the method of Thorpe and Holmes or by other suitable means. I t is then diluted with water or mixed with ethyl alcohol until it contains 10 per cent. of total alcohol by volume. “To 6 C.C. of this prepared liquid, contained in a wide test-tube, are added 2-5 C.C.of potassium permanganate solution (2 per cent.), and then 0-2 C.C. of strong sulphuric acid. When the reaction has proceeded about three minutes, 0.5 C.C. of oxalic acid solution (9.6 grm. crystals per 100 c.c.) is added. On shaking, the liquid becomes clear or nearly colourless. One C.C. of strong sulphuric acid is now run in and well mixed with the solution, which is finally treated with 5 C.C.of Schiffs reagent. A violet colour is developed in the course of a few minutes, unless mere traces of methyl alcohol were present, when twenty or thirty minutes may be required. “ A preliminary experiment carried out as described serves to detect the presence of methyl alcohol and to give some idea of the quantity. According to the indications thus obtained, another part of the prepared liquid is further diluted, if necessary, with ethyl alcohol of 10 per cent.strength, until it contains from 0.001 to 0.004 grm. methyl alcohol in 5 c.c., and the experiment is repeated side by side with two or more standards for comparison. These contain 0.001, 0.002, 0-003, etc.: grm. methyl alcohol in 5 C.C. of 10 per cent. ethyl alcohol.” In his original paper (Com,t.rend., 1867, 64, 182), Schiff gave no formula for his reagent beyond that it was a solution of rosaniline sulphite in sulphurous acid. The many different formulae in the textbooks are due to Gayon, Mohler, and others, and usually prescribe the use of fuchsine with varying quantities of sodium bisulphite and mineral acid. Two such solutions, selected a t random from modern textbooks, failed to give any colour when applied to ethyl alcohol containing 5 per cent.of Deniges worked on 0.1 C.C. of strong spirit. In the first place, “Schiffs reagent” is an ambiguous term.220 J. CECIL JONES: ESTIMATION OF METHYL ALCOHOL methyl alcohol that had been oxidised as directed by Simmonds. By reducing the amount of acid directed by Simmonds to be added immediately before the Schiffs reagent, they might have been made to work, but this was not followed up, as the stock Schiff solution of the writer’s laboratory was found to serve well.It is important, however, to note that the formula of the Schiffs reagent used has a marked influence on the sensitiveness of the test. In the writer’s laboratory Schiffs solution is prepared by dissolving 0.2 grm.of magenta base in 10 C.C. of a freshly prepared, cold, saturated aqueous solution of sulphur dioxide, and after twenty-four hours diluting the solution to 200 C.C. with water. Such a solution is very sensitive, and will develop a distinct colour within a minute or so, even with such substances as vanillin or salicylic aldehyde. I n the notes that follow the use of Schiff’s solution of this composition is assumed.The amount of permanganate used must be rigidly adhered to in quantitative work, since it determines the amount of formaldehyde formed and the final colour intensity. The use of 2 C.C. in place of 2.5 C.C. reduces the final colour intensity about 30 per cent., whilst the use of 5 C.C. more than doubles the sensitiveness of the test, when all other conditions are kept the same, except that the use of 5 C.C. of permanganate necessitates the subsequent use of nearly 1 C.C.instead of 0.5 C.C. of oxalic acid solution. Provided all the other standard conditions are rigidly adhered to, this is the simplest and safest way of increasing the sensitiveness of the test, as ethyl alcohol under these conditions yields no formaldehyde and no colour, but the permissible latitude in some of the other measurements then becomes narrower.The amount of sulphuric acid added with the permanganate is less important. The amount of formaldehyde formed is greater with more acid, but, within the limits 0.1 to 0.3 c.c., the final results are indistinguishable, owing to the compensating effect’of the higher final concentration of acid reducing the intensity of colour due to a definite quantity of formaldehyde. Larger quantities, however, might lead to the formation of formaldehyde from ethyl alcohol itself.The exact concentration of acid necessary to bring this about was not determined, but Denighs’ original paper shows that if 0.7 C.C. of sulphuric acid were present in a total volume of 8 c.c., formaldehyde would certainly be produced from ethyl alcohol.The time allowed for oxidation, provided it be not less than three minutes, appears to be without effect. The influence of temperature may not be negligible, but the point was not investigated, as the temperature might be expected to be constant in any one set of experiments, and duplicate experiments run simul- taneously without any special precautions to insure uniformity of temperature invariably gave identical results.Of all the reagents used, oxalic acid is the one of which the exact concentration is least important. For some reason, not plainly evident, ‘Simmonds gives the concentration of this reagent with exceptional precision-namely, 9.6 per cent. As a, matter of fact, solutions of oxalic acid of this strength deposit crystals at ordinary temperatures.The writer uses 0.5 (3.0. of a cold, saturated solution. Not much less will reduce the excess of permanganate in the cold, faintly acid solution, but larger quantities, up to 1 c.c., appear to have no measurable influence on the results.IN PRESENCE OF ETHYL ALCOHOL 221 The subsequent addition of sulphuric acid must be as nearly as possible the same in any one set of experiments, and is most conveniently made 1 c.c., as directed by Simmonds.With only 0.6 C.C. added, the acetaldehyde derived from the ethyl alcohol will give a distinct coloration with Schiffs solution of the character described. With 0.75 C.C. or more, pure ethyl alcohol gives no purple or even blue colour, provided all the other standard conditions are rigidly adhered to, but it is unwise to reduce the amount below 1 C.C.except in very special circumstances, where it, is necessary to make the test as sensitive as possible. In such cases the sensitiveness of the test can be increased 40 per cent. by using only 0.75 C.C. of acid, but very careful control experiments become necessary.On the other hand, not more than 1 C.C. of acid should be used, as the use of so much as 1-25 C.C. reduces the sensitiveness of the test about 30 per cent., whilst 1.5 C.C. reduces it 50 per cent., and 2 C.C. nearly 90 per cent. If 10 C.C. be taken instead of 5 c.c., the acid concentration is so much reduced that even the acetaldehyde from ethyl alcohol develops a colour; On the other hand, the use of only 2 C.C.in place of 5 C.C. reduces the amount of colour developed by a fixed quantity of formaldehyde about 90 per cent. The method is thus eeen to be very sensitive to small variations in the experi- mental conditions, a fact that cannot have escaped Simmonds, but which is not emphasised in his paper. I t is, however, quite easy to secure sufficient uniformity in the preparation of the assay and comparison liquids, and it is quite worth while, for the method is incomparably quicker than any other with any pretence to accuracy, and, if conducted with care, is at least as accurate as any.As described above, it will not detect much less than 0.2 per cent, of methyl alcohol expressed as a per- centage of the.tota1 alcohols.This is sensitive enough for almost any purpose, but if greater sensitiveness be needed, the foregoing notes indicate the directions in which the test should be modified so as to show 0.1 per cent., or even less. They also show that it is scarcely advisable to strive after the utmost sensitiveness unless some substantial reason exists for doing so. Proportions of the order of 1 per cent.can be estimated with an error not exceeding 1 part in 20. Finally, the amount of Schiff’s solution taken is not without influence. DISCUSSION. The PRESIDENT, in inviting discussion, remarked that, particularly when the quantities of methyl alcohol were small, processes depending on oxidation under strictly defined conditions were liable either to under-estimate or to over-estimate the quantity of methyl alcohol to a considerable extent, and for that reason he felt that Mr.Jones had done good service in investigating what was perhaps one of the best of such methods, and in emphasising the need for close adherence to the condi- tions laid down by the original author. Mr. A. E. PARKES said that in his experience even small quantities of perman- ganate would oxidise ordinary ethyl alcohol, and he asked whether, under the condi- tions of this process, it was possible to oxidise methyl alcohol without oxidising the222 ESTIMATION OF METHYL ALCOHOL IN PRESENCE OF ETHYL ALCOHOL ethyl alcohol.Any acetaldehyde that might be formed would, of course, give a strong reaction with Schiff’s reagent, just as formaldehyde would.Mr. F. H. LEES said that it seemed difficult to understand why the sensitiveness should be increased by the use of a larger quantity of permanganate. One explana- tion that suggested itself was that more acetaldehyde wag produced when more permanganate was used, and that the final acidity of the solution to which Schiff’s reagent was added was then not suffioient to prevent the acetaldehyde from producing a coloration.Mr. W. T. BURGESS asked what precautions were taken in the measurement of the small quantities in whioh some of the reagents were added. In measuring accurately a liquid like strong sulphuric acid, he should expect Bome difficulty to be experienced when the quantity was as small as 1 C.C. Mr. JONES, in reply, said that permanganate under these conditions certainly did oxidise ethyl alcohol to acetaldehyde, but it would be found that, in presence of the quantity of sulphuric acid which was added at the end, the acetaldehyde would give no coloration with the Schiffs reagent described in the paper or with that described in the British Pharmacopoeia.No formaldehyde was formed from ethyl alcohol under these conditions.He thought that the increased sensitiveness when more permanganate was used was due to the production of more formaldehyde. Although this method of estimating methyl alcohol was more or less quantitative, the oxidation of methyl alcohol to formaldehyde was by no means quantitative. With regard to the question of measuring, since, in the case of the permanganate solution, the use of 5 c.c., instead of 24 c.c., would result in an over-estimate of 100 per cent., if 2.6 C.C. were used the error would be 4 per cent., and it should certainly be possible to measure such a quantity with a smaller error than 1 in 25. For the sulphuric acid he used a 1 C.C. graduated pipette with a very fine orifice, giving about 5 drope to 0-1 C.C. I t was only necessary to add the same number of drops in the control experiment as in the other, and that could certainly be done with all the accuracy that the test required. If 12 C.C. of sulphuric acid were used, the sensitiveness would be increased by 30 per cent. It was quite easy to measure the 1 C.C. so closely that the method was accurate within 5 per cent., and more than this he did not expect from it.