[This chapter was so long and so tedious… There was a lot to learn, and two methods stood out as practical. However, there was mostly a lot to learn about what is no longer viable and/or to marvel at how much time and effort they used to invest. If you look at a method, consider two perspectives. One is whether the information would benefit a distiller developing a process. Would they really need to count furfurol and aldehydes? Or, is this for regulators like the IRS that were tasked with evaluation of mystery products for consumer safety or doing the initial basic science to support spirits quality.
The 21rst century makes it a lot easier to obtain pure chemicals and affordable used equipment, but each process is still startlingly involved and makes the birectifier method with organoleptic assay appear—cheap, fast and good! With the birectifier method, I am adding acid titrations and ester determinations. I may try the most simple colormetric determination of fusel oil, using a role model as a reference as opposed to a pure standard. I plan on investigating the use of silver oxide in regards to teasing out rum oil and learning more about sulfuric acid monohydrate (which I think is nothing more than a 1:1 mix with water which likely corresponds to a density).] [Silver oxide was later described as not worth the effort by a helpful professor.]
SPIRITS ANALYSIS (1)
(1) Mariller (Ch.) et Grosfiley (I.) Le contrôle chimique en distillerie. Paris, 1939.
The processes used for the classification and appreciation of spirits can be divided into two main groups: the organoleptic methods, which consist in examining the smell and flavor of the products, and the physico chemical processes, by which the various constituents are measured: ethyl alcohol, extractive substances (sugar, tannin, ash, etc.) and volatile matter foreign to ethyl alcohol (aldehydes, volatile acids, esters, higher alcohols).
“Tasting,” writes Girard and Cuniasse (2), “was formerly the only method used to appreciate the value of wines and spirits. This process then seemed sufficient to discover falsifications in use. In fact, there were only drinks more or less agreeable according to the delicacy of the cru and the success of the harvests; but almost all were natural, and the few falsifications in use were too crude to escape the experienced taster, whose main purpose was to classify wines and spirits, according to their value and origin. In spirits, it was possible to determine approximately the amount of alcohol they contained and to find any added industrial alcohol, then was very poorly rectified and of a special taste”.
(2) Manuel pratique de l’analyse des alcools et des spiritueux, Paris, 1899.
But with the phylloxera crisis, which destroyed part of the French vineyard and reduced the production of wine spirits to an insufficient number, the trade began to sell industrial alcohol-based spirits. The manufacture of adulterated spirits became a thriving industry and falsification a science. To discover falsification, it was then necessary to resort to chemical analysis, the old process of evaluation by tasting, still continued to play a useful role in the interpretation of the results provided by the analysis.
[Most large scale rum distillers I’ve talked to wished they were doing more organoleptic analysis.]
Measurement of ethyl alcohol.
In the past, various empirical methods had been devised to assess the alcoholic richness of eaux-de-vie. The most used was the Holland proof or pearl test. One half filled a small glass bottle with alcoholic liquid, the aperture was closed with the thumb, and it was hurled against the thigh. The appearance of bubbles which formed on the surface of the liquid and their duration, more or less long, indicated the alcoholic strength. The eau-de-vie made the pearl, if its alcohol content was around 50°, and the different denominations used in the liquor trade (proof of Holland, three-six, four-five, five-six , etc.) represented the necessary dilutions to obtain the pearl. The three-six, for example, was a spirit of which it was necessary to add 3 volumes to 6 volumes of alcohol to have about 9 volumes giving the pearl: it corresponded to 85° centesimal.
[This is easy to understand, but strangely I have never heard of it. Kervegant uses the term trois-six else where and I guess it implies 85° spirits or possibly a slang for industrial neutral spirits of that strength?]
In England, the most commonly used proof was to saturate a small amount of gunpowder with the brandy to try, then to approach a flame. If the powder caught fire, the eau-de-vie was said to be proof or over proof; if not, the spirit was under proof. All these processes were not necessarily very precise.
At present, the determination of the alcoholic degree is sometimes made by taking the density of the liquid by means of the pycnometer or Westphal balance and deducing from the density the alcohol content by means of special tables. This procedure is particularly suitable when only small amounts of liquid are available.
More often, the alcoholic strength is obtained directly by means of specially graduated hydrometers, called alcoholometers. There are still other devices: ebullioscopes and ebulliometers, Geissler vaporimeter, Duclaux comptegouttes, immersion refractometers, etc.., which can directly and quickly measure the alcohol contained in wines and spirits, but as they only provide approximate results, they are used only in special cases, when it comes for example to determine small amounts of alcohol (measure of alcohol in musts, vinasse, etc…). The Zeiss immersion refractometer is, however, quite commonly used in the United States. A special table makes it possible to go from the refractive index to the percentage of alcohol in volume.
[I’ve seen the Duclaux comptegouttes used for measuring the alcohol content of wines and it is a surface tension based measurement. The Zeiss refractometers are often called ABBE refractometers and I’ve long wanted one.]
In order to obtain highly accurate results, it is preferable to perform the quantitative determination of alcohol by chemical means (see Chapter XVIII).
The origin of the hydrometer is traced back to Ramnius, who lived under Tiberius and Caligula. This instrument was perfected by Baumé, who entrusted the construction of a liqueur scale to Cartier, a goldsmith, who had it adopted under his name (decision of the Council of State and tables, August 3, 1771). The Cartier apparatus, made of metal, was officially replaced in France by the centesimal alcoholmeter of Gay-Lussac, in 1824; but it continued, however, to be used in the spirits trade and still today it is used in certain countries (South America). Various other models of alcoholometers are used in foreign countries: alcoholometers of Tralles (Germany, Italy), Richter (Germany), Sykes (England), Gendar (United States), etc…
[I have a nice hydrometer collection. I’ve seen early ones made of bone with scales on them. I’ve seen gold plated Cartier’s. Notably, Jamaica used the Arnibaldi scale which was simply calibrated to tropical temperatures.]
Alcoholmeter Gay-Lussac. — This instrument, made of glass, is divided into 100 parts, giving directly the percentage of alcohol in volume (centesimal alcoholometers). It is graduated by successively immersing the apparatus in distilled water (point 0), in absolute alcohol (point 100) and in known mixtures of alcohol and water (95, 90, 85… % alcohol), at a temperature of 15° C.
The French legal alcoholmeter, made obligatory by the law of July 7th, 1881, is the Gay-Lussac alcoholometer, corrected by the National Office of Weights and Measures, according to the evaluations of the densities of the mixtures of alcohol and water made by Mendelejeff. The density 15/15° (i.e., taken at 15° relative to water at 15°) of the absolute alcohol, considered by Gay-Lussac as being 0.7947, was found by Mendelejeff equal to 0.79433 (weighed back to vacuum), a figure confirmed to a very small difference by the work of modern physicists (Young, Winckler, Merriman, etc…) (1). It follows that there is no absolute agreement between the legal alcoholmeter and the Gay-Lussac of 1824: the difference is maximum (0° 43) for the measures of 20 and 21°.
(1) Stastny and Renz, based on the experimental work of Osborne, McKelvey and Bearce (Bur. Of Standards Bull IX, No. 3, 1913), established tables in 1987 giving densities 15/15° and 15/4° mixtures of alcohol and water, as well as the differences between the densities calculated in this way and those indicated by 5 classic tables, including the French legal table. For 100 ° they give: exact density 15/15° 0.794292 and legal density 0.794330, a difference of 0.000038, corresponding to 0°03. At intermediate strengths, the differences are larger: 0.000142, for example, for mixing 2% alcohol by volume, 0.000358 for the 55% (maximum difference).
[Alcoholimetry can start to melt your brain.]
The degrees of the legal alcoholmeter shall be spaced at least 5 mm apart, so that they may be divided into fifths. The full scale is divided into 3 or more usually 5 instruments, so as not to have devices too long and not easy to handle.
The Gay-Lussac alcoholometer is used informally almost everywhere and in official capacity in Belgium, Switzerland, Spain, Portugal, Greece, Turkey and the countries of South America.
Tralles alcoholometer. — This apparatus differs from that of Gay-Lussac only in the density of absolute alcohol (0.7948), which represents the hundredth degree of the scale, and by the temperature of graduation: 15.56/4° (12° 4/9 Reaumur). The indications given by the two instruments are not appreciably different: the difference is a few tenths of a degree at most.
The Tralles alcoholometer often carries a thermometer fused to the lower part of the float, whose tank serves as ballast. The thermometer gives, not thermometric degrees, but the correction that must be made to the Tralles scale to bring it back to what it would be at the temperature of 4°. This correction can only be approximate, the coefficient of expansion of the alcohol varies significantly with the temperature.
Officially employed in Germany until 1889, the Tralles alcoholometer is official in Italy and Russia. He has been replaced in Germany since the 1st of July, 1889, by the Richter weighted alcoholmeter, graduated at a temperature of 15° C, accepting as the specific gravity of absolute alcohol 0.79425. This instrument gives strength in gr. of alcohol for 100 gr. of liquid. The German standard prototypes consist of 6 alcoholometers, divided into tenths of a degree, allowing an approximaton to one thousandth. The Richter alcoholmeter frequently has a thermometer.
Cartier alcoholmeter. — The Cartier hydrometer is graduated arbitrarily, by immersing the apparatus in a saturated solution of sea salt (point 0), at the temperature of 12°5C (10° Reaumur), then in pure water (point 10); the interval is divided into 10 equal parts and the graduation is continued above it by equidistant lines. The device marks 44° in absolute alcohol.
The Cartier alcoholmeter is still used in many Spanish countries, as well as in Cuba, Santo Domingo, Haiti.
Also according to the Cartier scale are graduated Bubbles, employed in the English countries and in Surinam to measure alcoholic strength of liquids flowing from the still. These are small glass spheres, whose volume and weight are adjusted so that they remain immersed in the center of liquids. There are 15 bulbs, numbered from 16 to 30, the number 30 corresponding to 20° Cartier, the number 29 to 21° Cartier, the number 28 to 22° Cartier, etc.
[I have a beautiful complete set of these from Scotland. I have never seen a Carribean set. In another chapter Kervegant describes how these may have been used for small sampling ports on a continuous still that may not have been a full on eprouvettes that could float a hydrometer. If you have a set, please send me a photo. I’ve seen photos of very old ones that even associate the measures with quality like good, bad, etc.]
Dutch Vochmeter. — This alcohol meter, from Van Baumhauer, is used in Holland and in the Dutch colonies (Surinam, etc.). It is divided into 33 parts, the 0 corresponding to pure water at 60° Fahrenheit (15°56 C) and the degree 10 representing the spirits of proof (Vocht proof), at 53°62 G. L.
Sykes hydrometer. — The Sykes hydrometer, used in England and in the English colonies, is an hydrometer of variable weight and volume, constructed of silvered or gilded copper. It consists of a spherical float surmounted by a rectangular rod, divided into 10 equal parts each subdivided into tenths and bearing below a cylindrical rod. Nine small copper weights, numbered 10, 20 … 90, which can be fixed on the latter, accompany the device. The instrument without weight marks 0 in strong alcohol; loaded with the heaviest weight, it scores 10 in pure water at 51° Fahrenheit (10°56 C).
The Sykes hydrometer does not indicate the density or the alcohol content, but the relation which exists between the examined liquid and another alcoholic liquid taken for comparison and which is called “proof spirit”. One chooses the weight that makes the level of the spirits to be tried out in the space between the divisions 0 and 10. One adds to the figure of the weight the number of the divisions which emerge, one notes the degree Fahrenheit and, by referring to the tables from Sykes, we obtain the degree above (over proof) or below the proof (under proof). This degree corresponds to the number of liters of proof that can be formed by a hectolitre of the examined liquid: the number is greater than 100 when the liquid is more alcoholic than the type, and it is smaller than 100 in the opposite case. So 40° O.P. means that 100 liters of spirits contain as much alcohol as 140 liters of “proof”, and 40 U. P. requires 140 liters of spirits to obtain 100 liters of “proof spirit”.
[This is fascinating and I’ve always avoided learning the Sykes system. It makes it seem very practical for on the flay putting volumes to work, either dilution for sale or filling the still. It is a scale tailored to simple work tasks.]
Proof spirit has been defined (Act of the Parliament of July 2, 1816) as being a spirit which, at the temperature of 51° F., has as weight 12/13 of the weight of an equal volume of distilled water (at 51° F), which corresponds, according to modern determinations, to a 57.029% alcohol content by volume (Connah). The degrees Sykes can be converted into centesimal degrees, multiplied by 100 : 175.35 (ratio of the percentage scale to the Sykes scale) and by adding (degrees O.P.) or subtracting (U.P.) from the figure obtained 57.029.
Gendar Hydrometer. — This alcoholmeter, official in the United States is a bicentalesimal device, marking 0° in water at 15°56 and 200 in pure alcohol. The spirit of proof of America contains 50% alcohol in volume and corresponds to the degree 100 of the Gendar graduation. Two degrees correspond very substantially to a degree of the Gay-Lussac scale.
[I have never heard this referred to as a Gendar scale.]
We give in the following table, according to Dujardin-Salleron (1), the correspondences between the density of the mixtures of water and alcohol and the different alcoholic graduations. Note that with regard to the Cartier graduation, it has been published in many tables, with significant differences between them.
(1) Notice sur les instruments de précision appliqués à l’oenologie, 5th ed. Paris, 1923.
We reproduce below the table of correspondences between the weight and the percentage of alcohol in volume at the temperature of 15°56 C. (60° F.), given by the “Bureau of Standards” of the United States (1), for mixtures of alcohol and water.
(1) Bureau of Standards Circular No 19. p. 18, 1924.
Correspondence between weights and volumes of ethyl alcohol.
Apparent degree and real degree.
The indications given by the alcoholometers are accurate only if the measure is taken at the temperature at which the instruments were calibrated, and if the liquid examined consists solely of a mixture of water and alcohol. When these conditions are not realized, it is necessary to make corrections to the apparent degree, to obtain the real alcoholic strength of the liquide, or real degree.
Temperature correction. — For the use of his alcoholmeter, Gay Lussac has established a special table, calculated from degree to degree, called “table of the real strength of liquids”, giving, at various temperatures, the volume of pure alcohol contained in 100 volumes of spirits brought back to the temperature of 15° (1). This gives the real strength of the liquid, that is to say the proportion of alcohol relative to water.
(1) It is also better to avoid corrections, which always lead to errors, operating exactly at 15 ° C. The correction tables, based on the coefficient of expansion of the alcohol, which varies with the temperature and the degree of the alcohol, can not in fact account for the expansion or contraction of the alcohol.
If we want to have the real alcoholic strength, that is to say the quantity of alcohol contained in a given volume of liquid, we must take into account expansion or contraction under the action of temperature. To determine the alcoholic richness, we must multiply the real force by the volume occupied by the liquid at 15° C. The table of Gay-Lussac carries, below the figure which gives the real force, the indication of the real volume that would occupy 1 liter of the examined liquid if it was at 15°.
In order to avoid these calculations, the French Régie has drawn up a table, called “table of alcoholic strength” (2), which takes into account the double correction above. It gives the volume of pure alcohol at 15° which is contained in 100 volumes of the spirits examined, measured at the temperature at which the observation was made. Dujardin has established a rule of alcoholic corrections graduated in tenths of a degree, which avoids the use of the table and renders the sensitivity of the readings much greater, avoiding any calculation when one has to appreciate fractions of degree by interpolation.
(2) Guide pratique d’alcoométrię. Lib. administrative P. Oudin, Poitiers.
The table of alcoholic strength is very convenient for the needs of commerce and the Régie (determination of the volume of pure alcohol contained in a tank), but for the analysis of spirits (determination of the chemical composition) the only one that is suitable is that of real force. It is only at 15° that the numbers of the two tables are identical.
The confusion between the table of real force and that of alcoholic strength often causes errors and disputes. The alcoholometric weighing, used in Germany since 1899 and recommended on various occasions in France, in particular by Barbet since 1893, does not have the drawbacks of volumetric alcoholometry, the alcohol weights being invariable whatever the temperature.
[I do not know enough about alcohol by weight.]
Correction of the dry extract. — Beverage spirits generally contain solids, the proportion of which in some cases can reach 20 to 30 grams per liter. These substances, giving aqueous solutions whose density is greater than 1, increase the density of the eau-de-vie, so that the alcoholmeter indicates an apparent degree lower than the actual degree. To determine the latter, several methods can be used.
In the official French method, 250 cc of eau-de-vie is measured at a temperature as close as possible to 15° C; if the alcoholic strength reaches 65°, take only 200 cc and add 50 cc of water, while if it is less than 50°, 275 cc are distilled. The refrigeration is carried out by means of a pure tin coil having at least 1 m. long, cooled by a stream of cold water. The distillate is collected in a 250 cc graduated flask, in which 10 cc of water is placed. At the end of the refrigerant tube, a tapered glass tube is added, by means of a rubber connector, which is immersed in the water placed at the bottom of the flask, so as to ensure the condensation of the overhead products and in particular aldehydes. Distillation is pushed as far as possible; bring to 250 cc, with distilled water and measure the alcoholic strength.
The long and delicate distillation process is commonly used for wines and beers. But, although official in France, it is not very common in the case of eaux-de-vie, because of the losses of alcohol which can occur especially in hot countries.
b) Tabarié process. — Take the alcoholic strength of the spirit to be tested, then evaporate about 200 cc, liquid in a water bath, until all the alcohol is removed. The residue is then brought to the original volume with distilled water and the density determined (by means of a pycnometer or hydrostatic balance). If this density is equal to 1, the alcoholic degree observed was the real degree; if it is greater than 1, the excess of density is to be subtracted from the density corresponding to the primitive alcoholic degree. The new density thus obtained corresponds to the real degree.
[The American TTB has better protocols for this though few use it because it requires a pricey analytical balance. It is typicaly used to assess obscuration by barrel solids.]
Still more simply, the density of the alcoholic liquid deprived of its extractive matters is obtained by dividing the density of the eau-de-vie by that of the residue dissolved in water. The real degree of alcohol is easily deduced by means of the density tables of the alcoholic mixtures.
This simple and rational way of proceeding is recommended by many authors (Rocques, Deerr, etc …).
c) Blarez’s formula. — The quantity of extractive materials is determined by evaporation, and the degree alcoholic is corrected by applying the formula:
where A is the actual degree, a the apparent degree, E the solids content (grs per liter) and α a variable coefficient with the alcoholic strength.
We make a first calculation, by applying the coefficient to the apparent degree found, provisionally taken as real, and adding the product obtained to the apparent degree; then a second calculation, this time using the coefficient corresponding to the exact real degree. If the extractive material does not exceed 4 or 5 grams per liter, a single calculation is sufficient.
[I uses versions of this idea to measure the sugar content of a liquor by measuring its density and then the effect of the disclosed alcohol content on the density. The different being the increase in density created by sugar and other solids. This is the best way to reversion engineer the sugar content of liquors and also understand the starting proof of the spirit before sugar is added.]
We can compare the formula of Blarez, and that proposed by Malavoine (1), valid for spirits whose alcoholic strength is between 30 and 70°:
(1) Mitt. Lebensm. Hyg. XXII, 145, 1931.
where Dr is the actual degree, Da the apparent degree determined by means of an alcoholmeter and E the dry extract determined by the evaporation of 10 to 20 cc, of the spirits.
[There is lots of roof for error to creep in here. Evaporation has to be followed by desiccation.
Determination of fixed materials
To measure the dry extract, take 25 cc. of spirits that are poured into a platinum capsule or glass tared exactly. Evaporated in a water bath or in a well-controlled oven at 105 or 110°, and cool in a desiccator before weighing the residue. The weight obtained is expressed in grams per liter of eau-de-vie.
The extract can also be obtained by vacuum. For this purpose, it is usual to use glass vacuum bells, with a lapped closure. The capsules containing the alcoholic liquid are placed on a perforated plate, below which is a container containing sulfuric acid monohydrate. After 2 days of exposure under vacuum, the sulfuric acid is replaced by anhydrous phosphoric acid and is left for 2 days. Before opening the bell, it allows air to enter, previously dried by passing through a column of calcium chloride. A difference between the weight of the extract in vacuum and the hot extract indicates the presence of glycerin.
[I need to ponder this. It doesn’t seem to rely on vacuum evaporation so much as desiccation in a vacuum. I don’t understand the change from sulfuric acid as a desiccant to phosphoric acid. Calcium chloride is a just a light duty desiccant capable of drying the introduced air. This is likely neurotic analysis for regulatory agencies.]
When the weight of the extract exceeds 1 gram per liter, it is interesting to make at least a summary review, to recognize the main components: sugars, glycerin, caramel, tannin, etc …
The only sugars that are practically to be found in eaux-de-vie are sucrose and invert sugar.
The presence of the alcohol affects the rotatory power of sugars, it is good to eliminate it by a rapid distillation and bringing the residue to the original volume with distilled water. The aqueous solution is then clarified, if necessary, by means of lead sub-acetate.
[I aim to learn this because it is relevant to molasses analysis. Alcohol will bias a common brix refractometer which is by you must separate it. Here rotary power is measured I think with an ABBE refractometer.]
By means of a qualitative test with the Fehling liquor, the presence of reducing sugars is ensured. If these are absent, sucrose is evaluated by direct polarization and, otherwise, by double polarization, according to ordinary methods. The reducers are evaluated with Fehling’s liquor.
[I will have to get more background here. These are classic tests.]
The presence of glycerine in a significant amount is recognized by the consistency of the dry extract, which does not dry. By heating it with potassium bisulphate, we perceive the characteristic odor of acrolein.
[This is very cool and helpful!]
The search for caramel is most often carried out by the method of Amthor, which is based on the precipitation of this substance by paraldehyde.
10 cc of the eau-de-vie to be examined are introduced in a 100 cc flask and 30 to 50 cc, of paraldehyde, according to the intensity of coloration, and about 15 – 25 cc of absolute alcohol are introduced. Mix and let stand 24 hours. If there is caramel, it forms a more or less dark precipitate, which adheres strongly to the walls of the bottle. The supernatant liquid is decanted, the precipitate is washed with a little alcohol and then dissolved in warm water. After filtration, the solution is brought back to the initial volume by evaporation.
Comparing the colorimeter solution with a caramel solution of known concentration, evaluates the amount of caramel contained in the eau-de-vie. However, the figure obtained is only approximate because the hue of the various caramels is quite variable according to the way they have been prepared. If the amount of caramel is very low, it is good to first concentrate the eau-de-vie, by evaporation preferably under a vacuum, to avoid the formation of caramelization products.
It is possible to recognize the presence of tannic materials in colorations made with ferric chloride or the precipitate which forms, after evaporation of the alcohol, with a solution of gelatin.
To find out if the tannin comes from oak, add to 10 cc of spirits drop by drop and without shaking, a solution of 5 grams of ferric chloride in 100 cc of water: a positive will produce a cerulian black color.
[SOS not positive about the logic of that last line]
In case one suspects the presence of these dyes, very rarely used elsewhere for spirits, we will look for them in the following way:
After evaporation of the alcohol, ammonia is added to the residue and treated with amyl alcohol, who dissolves coal dye and a little caramel. The amyl alcohol is washed and evaporated on a water-bath in a small porcelain dish. A little concentrated sulfuric acid added to the residue, is triturated with a stick, in order to carbonize the caramel entrained by the solvent. It is then mixed with water and treated again with ammonia and amyl alcohol. The latter, after evaporation, leaves as the dyestuffs as residue, sufficiently pure to be revealed by means of their characteristic reactions and by tests of wool dyeing in acidic or alkaline solution.
[Trituration is just grinding the particles smaller with an implement. I have seen sulfuric acid carbonize caramel added to spirits when practicing the version of the sulfuric acid test for rum oil used by Arroyo.]
For the determination of ashes, the dry extract, contained in a platinum capsule, is place in a muffle furnace, and heated moderately.
Eaux-de-vie sometimes contains small quantities of copper, iron, zinc and lead, which comes from the distillation apparatus or containers in which the alcohols are kept. To search for these metals, the ash is treated by the ordinary methods of qualitative and quantitative analysis. The presence of small amounts of copper can be detected, for example, by means of the Maquenne and Demoussy reaction (1), which allows recovery of 1 to 1.5 mgr. of copper in a liter.
(1) C. R. CLXVIII, 489, 1919.
[There are other modern tests for dissolved copper and they are used by some of the best single malt distilleries to spot irregularities. If dissolved copper is normal, they roll with it, otherwise they invest in more time consuming and comprehensive tests isolate the issue. This only works for productions that have evolved to be extremely regular. Elevated copper content could mean elevated volatile acid in the ferment or even issues with the still such as abnormal coolent water or energy input to the boiler.]
Determination of volatile materials other than ethyl alcohol
The method which gives the most detailed indications of the nature and proportions of the volatile substances other than ethyl alcohol contained in spirits is that of fractional distillation.
It consists of distilling the alcoholic liquid, using special laboratory equipment (tubes from Wurtz, Lebel, Henninger, Glynski, Otto; columns from Claudon, Morin, Savalle, etc.), by splitting the volatile products from 10 to 10 degrees. Each portion is then redistilled, setting aside parts that have a fixed boiling point. The intermediate portions are joined at the top, in order to be redistilled together, and the portions which have not given a fixed boiling point are again fractionated, again separating them. Thus, after a certain number of distillations, it is possible to isolate the different constituents from each other, or at least to obtain them with a rate of impurities which is small enough to be able to be characterized by their physical constants and their chemical properties.
[This is different than birectifier distillation and a little more theoretical. Those tubes and columns may in some cases be early vacuum distillation apparatuses. The fractions of birectifier distillation are done by volume over time instead of temperature. The birectifier is more practical in the context of beverage distillation. What Kervegant describes here is more old school wet chemistry examination of unknowns to find clues. As you purify something, you starting to escape its volatility in an alcohol-water matrix and start to look at pure boiling points, often temperatures well of 100°C. I should however be doing some of this to tease out the components in fraction 5 of birectifier distillation which can form droplets.]
This method can only be applied in research laboratories. It is, indeed, long and delicate, and requires implementation of large quantities of the liquid to be tested. In current practice, it is sufficient to evaluate the impurities by chemical functions: acids, aldehydes, esters, higher alcohols, to which furfurol and sometimes methyl alcohol and bases are added. Most of the tests in use have, moreover, only a conventional value, whence the necessity of rigorously following all the details of the operations indicated, to obtain results, if not very exact, at least relatively constant and comparable.
[Chromatography definitely changed a lot of this, but what it didn’t change is our interpretation of the results which is still very poor. GCMS used in working distilleries is not set up to examine high value congeners like rose ketones. Much of what you see in research papers are methods that can only practically be done by universities.]
In different assays, and in particular for the determination of aldehydes, furfurol and higher alcohols by colorimetric methods, it is necessary to operate under conditions of a determined alcoholic concentration, the alcoholic force acting on the sensitivity of the reagents. The measure of 50°, which corresponds to the average strength of eaux-de-vie, is the one generally adopted. Spirits should be brought to this degree, diluted with distilled water, if they are above 50°, or reinforced with pure alcohol (at 90 °), if they have less than 50°. To determine the amount of water or alcohol to add, one must consider the contraction that occurs when mixing alcohol and water. The general formula:
calculates how much x of an a’ degree alcohol and d’ density
(at + 15 ° C) must be added to 100 cc of an alcohol of degree a and of density d, to bring it to degree A, corresponding to the density D. The volume of the liquid obtained is given by the formula:
If a> A, that is to say if the degree is lowered by the addition of water (a ‘= 0, d’ = 1), and if A = 50 °, the preceding formulas become:
and if a < 50°:
Special tables have been set up, so-called mooring tables and spirits, which avoids the previous calculations.
It will obviously be necessary to take into account dilution, in the calculation of the content of impurities. If I represents the weight of the impurity found for the alcohol reduced to 50°, the content in the primary alcohol at t° will be expressed by the formula:
As it is generally agreed to calculate the impurities per 100 absolute alcohol, in order to make comparable spirits of different measure, we will have this proportion by applying the formula:
where A represents the weight of impurities per liter of alcohol at t°.
The determination of acids is carried out by simple alkalimetric titration, generally using phenolphtaleine as an indicator. German chemists, however, preferably use methyl red in alcohol solution at 0.1% and many authors have advocated the replacement of phenolphthalein by phenol red in 0.02% aqueous solution.
[I use phenolphtaleine and am not aware of clear arguments toward using something else.]
It is useful to know not only the total acidity, but also the fixed acidity and the volatile acidity.
To determine the total acidity by the official French method, we operate as follows:
25 cc of eau-de-vie are placed in a large flat-bottomed glass flask: 5 drops of 1% phenolphthalein alcohol solution are added and the solution is titrated with N / 20 potassium hydroxide solution. The number of cc of alkaline liquor employed, multiplied by 0.12, gives the acidity per liter expressed as acetic acid.
[I use standardized sodium hydroxide and am not aware of potassium hydroxide usage. I am assuming N/20 is 1/20 or 0.05 Normal.]
If the eau-de-vie contains a substantial quantity of carbonic acid in solution, it is necessary, before carrying out the determination, to heat it up to the beginning of boiling with the ascending refrigerant. If the liquid is very colored, as is usually the case for rums, we will prepare, in a vase identical to the one where the sample to be titrated was placed, an aqueous solution tinted with a suitable amount of brown Bismark, and one can thus, by comparison, be aware of the moment when the turn of the indicator (1) occurs.
(1) It is preferable, if the change of phenolphtaleine is difficult to see, to try tournesol as an indicator. As soon as the halo of the indicator is really blue, we stop the addition of alkaline titrant.
[The modern solution to these problems is degassing with an ultrasonic bath and observing the titration end point using a pH probe instead of a visual indicator.]
In the official American method, about 250 cc of boiled water, which is neutralized, is introduced into a porcelain dish. 25 cc of the spirits to be analyzed are added and titrated with N / 10 sodium hydroxide using approximately 2 cc of phenolphthalein solution as indicator.
[Notice this measure uses a more concentrated titrant. I may try this method because it give me more volume to have my pH prove above my magnetic stirrer. Something to note is that if you use distilled water, you may not exactly get a 7.0 reading from your pH probe.]
In recent years, the pH of the eau-de-vie has also been determined by the potentiometer, particularly in the United States. This gives interesting indications as regards the free acidity and the possible presence of highly ionized acids (sulfuric acid, etc.).
The fixed acidity is measured on the dry extract in the vacuum. In the official French method, 25 cc of eau-de-vie are evaporated in a water bath until approximately 5 cc remain, and the evaporation is completed in a vacuum. The residue is redissolved in distilled water and the titration is carried out as above.
[If you don’t have an aspirator style vacuum pump which can handle the water vapor running through it, I would just use a food dehydrator. The main idea is not to have the temperature of the flask or dish rocket up in temp to a point anything can decompose as the water content hits zero.]
In the official American method, 25-50 cc of spirits to be analyzed are placed in a platinum capsule. Evaporate in a water bath to dryness, then go to the oven at 100° for 1 hour. The residue is taken up with 25-50 cc of neutral alcohol having approximately the same alcohol content as the sample and transferred to a porcelain dish containing 250 cc of neutralized boiled water. The titration is carried out with N/10 sodium hydroxide placed in a 10 cc graduated burette and using 2 cc of phenolphthalein solution as an indicator.
Volatile acidity is generally obtained by the difference between the total acidity and the fixed acidity. It is not advisable to directly determine the free volatile acids on the alcoholic liquid previously distilled, because these acids pass only at the end of the distillation and with difficulty, so that some of them can remain in the residue.
[This is really interesting advice. I know wineries measure volatile acidity with a cash still / RD80 design ($745.00). I know of one brandy distillery that uses it because it is practical and fast. I suspect it gives practical results that sounds decisions can be made off as opposed to absolute accuracy.]
To clarify the nature of volatile acids, the Duclaux method can be used by fractional distillation: 100 cc of spirits are neutralized exactly with an N/20 sodium hydroxide solution and then distilled so as to have only about 10 cc of residue. The latter, which contains all the free acids of the eau-die-vie, has 40 cc of distilled water added and is poured into a 100-110 cc graduated flask. Water from washing the flask having contained the residue is combined with the preceding solution. 1 cc of phosphoric acid is added and made up to 110 cc. The liquid thus obtained is distilled in 10 cc fractions each, which is collected in graduated tubes.
The acidity of each of the fractions is determined with water of lime (or barium) N/50, in the presence of tournesol tincture as an indicator. Let a, b, c, d be the number of cc of lime water needed to saturate the 1st, 2nd, 3rd fraction. We divide each of these numbers by the last, and multiply the result by 100: we thus obtain the ratios in hundredths of the acids passed in the distillates. It will be possible to draw a curve, by bearing in abscissae the number of cc of distillate (10, 20, 30 … 100) and on the ordinate the corresponding ratios. Each volatile fatty acid has characteristic ratios and a characteristic curve, it will be deduced, by comparison with those provided by the spirits examined, the nature of the volatile acids of the latter.
[I first heard of this concept for deducing higher alcohols that were oxidized to their corresponding acids. This may be practical to help someone with no chromatography who is trying to understand the acids they are producing in their ferment (such as exploring a butyricum). This is time consuming, but requires nothing expensive and would likely be a great organoleptic training exercise. Notice a very dilute titrant is used to count very low levels of acids.]
Bordas and Raczkowski (1) have established formulas for easy determination by the Duclaux method, volatile acids produced in an alcoholic fermentation.
(1) J. Pharm. Chim. (6) VII, 479, 1898. [I think I have also seen 20th century data else in the A.O.A.C. Journal.]
The results provided by the method are, however, little used when it is a question of several volatile acids in mixture, the ratios corresponding to determined proportions of two or more given acids being able to be similar to those given by a mixture in proportions different from 2 or several other acids.
The distribution coefficient method, based on the different distribution of fatty acids in admixture between water and a water-immiscible solvent, such as ethyl ether (Behrens) or isopropyl ether (Werkman), presents with respect to that of Duclaux various advantages. It is faster, easier, more accurate and can be used for a mixture of volatile acids and fixed acids.
The following procedure was indicated by Osborne and Werkman (1) for the determination of 3 acids in admixture:
(1) Ind. Eng. Chem. (Anal ed.) III, 264, 1931. [I know where we could find this online if anyone really wants it.]
Take 100 cc of the solution to be analyzed and take 25 cc, which is titrated with 0.1 N potash. Adjust exactly the acidity of the remaining 75 cc between 0.12 and 0.08 N. Introduce 300 cc of the solution into a 100 cc separatory flask with 60 cc. of isopropyl ether. Shake vigorously for 1 minute, allow to stand for 3 minutes and then extract 25 cc of the aqueous layer, the acidity of which is measured with 0.1 N potash using phenolphtaleine as an indicator. Operate in parallel and in the same way on 30 cc of solution, which is treated with 15 cc, of isopropyl ether. It is important that the temperature, during the operation, be kept exactly at 25°.
[This is some old school wet chemistry. I’m not explicitly sure why they are choosing potash. Also, I think there is a big typo. Notice that they are also using 100 cc of the solution to be analyzed, then another 300 cc, and then another 30 cc… I think the 300 cc should really be 30. You are taking two portions of 30 cc, but they are being treated with different amounts of isopropyl ether. First 60 cc of the ether and then again with only 15 cc. These 30 cc volumes are also after they were adjusted to 0.12 and 0.08 N]
Let M be the number of cc of 0.1 N potash necessary to neutralize 25 cc of the original solution, M¹ and M² the number of cc of potash necessary to neutralize 25 cc of each of the aqueous layers obtained by treating 30 cc of the solution of one part with 60 cc and the other part with 15 cc of isopropyl ether, we will have:
K¹ and K² being the constants of the distribution rate.
The distribution coefficients between water and isopropyl ether, at the solvent concentrations indicated above, were determined by Osborne and Werkman for the acetic acid a, propionic acid p and butyric acid b:
We will obtain the percentage of each of these 3 acids in mixture, by solving the system of equations hereafter:
[I think I understand the method, but it is hard to say if it is worth trying in the modern era. A lot can be learned from examining fraction 6,7, and 8 of birectifier distillation and titrating those acids. The number in conjunction with organoleptic assessment can tell you if you are on the path to noble volatile acidity or possibly only pricked with ordinary acetic acid.]
Finally, various methods have been developed for research and determination of certain acids. In particular, Fincke (2), for the determination of formic acid, Grossfeld and Battay (3), for the determination of butyric acid, Grossfeld and Miermeister (4), for the determination of lauric acid and caprylic acid. The determination of high molecular weight acids is of particular interest because of the important role they appear to have in the constitution of the eau-de-vie bouquet.
(2) Z. Unters. Nahr. Genussm. XXV, 589, 1913.
(3) Z. Unters. Lebensm. LXI, 129, 1931.
(4) Z. Unters. Lebensm. LVI, 167, 1988.
Determination of lauric acid. — Distill 500 cc of spirits, placed in a 1-liter flask connected by a bulb tube to an inclined Liebig condenser, heating as evenly as possible. Stop distillation when 450 cc of liquid has been obtained. Then wash the condener and the capital with 50 cc of alcohol, which is added to the distillate. Then treat it with 5 cc of concentrated KOH (75 cc of KOH per 100 cc) and boil with rising refrigerant for 1/2 hour. Evaporate the saponified solution in a water bath solution, then add 25 cc of a buffered sodium acetate solution (obtained by dissolving in water 25 g of crystallized Na acetate, 5 cc, glacial acetic acid and 1 cc, a solution of phenolphtaleine at 1% and increasing to 250 cc). The liquid, which still has a strongly alkaline reaction, is neutralized by dropwise adding dilute acetic acid (30%) until slightly pinkish coloring, and it ends with the addition of sodium hydroxide. Then add a few drops of 0.1 N sodium hydroxide solution until the pink color appears again and filter into a 100 cc beaker, introducing a pinch of Kieselguhr to facilitate filtration.
[So you distill to separate the fixed acids. You also rinse the bulb style still capital and condenser because a lot of these acids can stick to the surface. KOH is potassium hydroxide. Rising refrigerant I think implies total reflux with the goal of saponification. The idea very much parallels the principals of H.H. Cousins High Ether process. The acetic salts are soluble, but the Laurel salts are not will precipitate so they can be filtered. You saponify them and evaporate everything else, then carefully get rid of all the excess KOH by making soluble salts with acetic acid.]
Bring the filtrate to the boil, and drop by drop 5 cc of a solution of magnesium sulphate (150: 1,000). It prevents the pink color from disappearing during cooling, adding a few drops of the sodium acetate solution. After 24 hours, after marking the liquid level on the beaker, filter on a Gooch crucible filled with asbestos (previously washed with magnesium laurate solution), and dry in an oven at 100° to constant weight, the aforementioned Mg laurate retained on the filter. Weigh and incinerate on a mushroom burner and weigh again.
[A mushroom burner is just a burner with a head to diffuse the flame. I think you have to incinerate because the measure of acid is so small you cannot titrate. You likely must also weight with a 0.0001 balance. This all sounds annoying.]
The difference between the two weighings gives the quantity of lauric anhydride, which is transformed into laurate of magnesia by multiplying by the factor 1,105. All the laurate not being precipitated after 24 hours, a correction coefficient must be applied, to account for the amount remaining in solution. Add, accordingly, to the figure found 0.0126 mg per cc of solution. The amount of lauric acid is calculated from Mg laurate, multiplying by the factor 0.9472.
[I will not be trying any of this.]
Determination of intermediate fatty acids (caproic acid, pelargonic acid, caprylic acid, capric acid). — Take 100 cc of spirits and distill 95 cc. Add to distillate 1 cc. 50% potash liquor and boil with constant reflux for 1/2 hour, then evaporate to dryness in a water bath. Take up the residue with 10 cc of water, add 1 drop of phenolphtaleine and neutralize until low pink color, with diluted acetic acid (30%). Add enough water to bring the volume of liquid to 50 cc and mix with 25 cc of magnesium sulphate solution (15 g per liter). After 24 hours, filter and, for 50 cc of the filtrate, stir in 10 cc of a solution of buffered Cu sulphate (containing per Liter: 50 g of crystallized sodium acetate, 3.12 g of crystallized Cu sulphate and 5 cc of 20% acetic acid). The next day, filter on a Gooch crucible filled with asbestos, dry the precipitate in the oven and weigh it.
A certain quantity of laurate remaining in solution, it is necessary to make a correction, subtracting from the figure found 0.7 mg. The amount of intermediate fatty acids, evaluated in mg. of caprylic acid per 100 cc of the examined liquid is obtained by multiplying by the factor 1,251.
[Keep in mind that this also looks at the acids that comprise esters in a spirit.]
Aldehydes can be assayed by colorimetric or volumetric methods. These are the first ones that are most commonly used in the analysis of eaux-de-vie. However, when the aldehyde level exceeds 1 gr. per liter, volumetric methods are preferred. Some authors (Bonis) recommend, however, even in this case, the colorimetric method, and advise to dilute the spirit with neutral alcohol, before proceeding to the test, so as to lower its concentration of aldehydes below 1%.
There are many substances the give color reaction with aldehydes, likely to be used for qualitative research and determination of bodies in this group. In particular, mention may be made of: fuschine bisulfite (indicated by Schiff), which gives a violet color; metaphenylenediamine hydrochloride (recommended by Windisch), which provides a yellow color with green fluorescence; sodium nitroprusside (Bela de Bitto), which gives a red-yellow to red-violet coloring; hydroquinone in sulfuric solution (yellow-orange coloring) etc.
Certain phenols (phenic acid, a – naphthol, b – naphthol, resorcinol, hydroquinone, pyrogallol, phloroglucine, guaiacol, etc.) give colored reactions, which vary according to the nature of the aldehydes: their use has been proposed for the differentiation of these (Barbet and Jandrier, Istrati).
It is fuschine bisulfit, or bisulphite of resaniline, which is most commonly used for the determination of aldehydes, except in Germany and Switzerland, where metaphenylenediamine hydrochloride is used instead.
Recoloration, in the presence of the aldehydes, of fuchsin previously discolored by sulphurous acid was reported by Schiff in 1865 (so-called Schiff reaction), and recommended by Schmidt, in 1881, to detect the traces of aldehydes, then by Gayon (1), to perform the evaluation of these. It has been attributed (Wieland and Schleuing (2), Rumpf (3)) to the simultaneous action of SO2 and aldehyde on a pale yellow para-fuschine-leucosulfonic acid, with the formation of a soluble violet dye resistant to discoloration.
(1) C. R. CV, 1182, 1887. The reagent proposed by Gayon was composed as follows: aqueous solution of fuchsin 1/1000 1000 cc – sodium bisulfite at 30 ° Baumé 20 cc. – pure and concentrated HCL 10 cc.
(2) Ber. Deut. Chem. Ges. LIV, 2527, 1921.
(3) Bull. Soc. Claim. LI, 508, 1933.
Purple coloration takes about 15 minutes to reach the maximum intensity and, after having persisted for some time, it gradually disappears. This is so at least in the case of pure acetic aldehyde. In fact, the duration of color development depends on the nature of the aldehydes. For example, Quantin (4) observed that for some rums, the maximum intensity was obtained after 10 minutes, while for others it was only after 25 minutes. The result is indecision over the amount of time that must elapse before colorimetric observations are made, and a cause of error that is difficult to escape.
(4) C. R. 4. Cong. Chim. Appl. II, 241, 1900.
Of the different aldehydes which may be found in eau-de-vie, it is acetic aldehyde which has the most effect on fuschine bisulfite. The others develop, at equal weight, a lower color intensity. However, the differences are quite small (except for furfurol (5) and butyric aldehyde), so that the evaluation of the total sum of aldehydes as acetic aldehyde can be considered close to reality, especially as this the last is the one that predominates in spirits.
(5) 70 parts of furfurol give the same coloring intensity as 1 part of acetaldehyde (J. Paul).
The reaction is also influenced by the relative proportions of fuschine and sulfurous gas, by alcoholic concentration and temperature. Coloring is all the more intense as the fuchsin content is high and that of SO2 is lower. If there is little SO2, the reagent is colored even in the presence of pure alcohol: this is what happens in particular with Gayon’s reagent, on everything when it has been prepared for some time. Raising the temperature promotes the reaction and may, if it is above 30° C, cause coloration in an alcohol free of aldehydes. The same is true of the alcohol concentration: at very high concentrations, the reaction can be positive in the absence of aldehydes. This is why it will be important to operate on the sample and standard solution under identical temperature conditions (the official American method specifies 15° C) and alcoholic concentration (50° generally).
Finally, it should be pointed out that the intensity of the coloration obtained is not proportional to the content of aldehydes, or at least that this proportionality exists only between very small limits. It will therefore be necessary to bring the two solutions to color intensity as close as possible.
French official method (Gayon – Rocques) – Reagents. — a) Titrated solution of ethyl aldehyde. Pure aldehydrate of pure ammonia is first purified from commerce, by grinding it several times in a mortar with anhydrous ether and decanting the solvent each time. The aldehydrate is then dried in the open air and then in vacuo on sulfuric acid. Weigh 1.386 grams of dry aldehydrate (amount corresponding to 1 gr of aldehyde); the material is introduced into a small 100 cc. flask, and about 50 cc of 95° pure alcohol is cold-dissolved. When the solution is operated, 22.7 cc of normal sulfuric acid is added to the pure alcohol at 95°. A precipitate of sulphate of Am is immediately produced. The volume is made up to 100 cc, with pure alcohol at 95°; then add, in addition, 0.8 cc. of alcohol, so as to compensate for the volume occupied by the Am sulphate formed. Stir, leave overnight and filter. This gives a solution of 1% aldehyde in pure alcohol at 95° C. It is then diluted with the amount of pure alcohol at 50° necessary to obtain a solution of 100 mg per liter of alcohol at 50°.
b) Rosaniline bisulfite solution. In a 250 cc graduated flask are poured: 30 cc of 1/1000 fuchsin solution in 95° pure alcohol, 15 cc. of sodium bisulphite at 36° Baum, 30 cc of water. The vial is capped, shaken and allowed to stand for one hour. At the end of this time, 15 cc of sulfuric acid is added at 1/3, and then it is made up to 250 cc with pure alcohol at 50°. This solution is slightly colored yellow when it has just been prepared; it disappears completely after a while.
Operating procedure — Test tubes of 20 cc capacity are used, stuffed with emery and bearing a mark of 10 cc. 10 cc of standard aldehyde solution at 100 per liter and in another 10 cc of the alcohol to be tested are introduced into a tube (distilled and brought to 50°). 4 cc of reagent are added to each tube: stir and wait for 20 minutes. At the end of this time, the colorimetric test will be carried out. The typical liquid is examined with the colorimeter to a thickness of 10 mm; determine the concentration of the other liquid necessary to obtain the equality of hue. If the liquid is slightly colored, the thickness of the type is lowered to 5 mm. for testing, then multiply the number read by 2.
[Se we are taking a known concentration and figuring out how much of it we need to make in color an unknown concentration while holding other variables constant.]
If the intensity of the coloration was directly proportional to the aldehyde content, the quantity of aldehydes contained in the spirits sample would be given by the formula:
where a represents the richness of the typical solution, H the height of the standard solution and h the height of the sample. But proportionality exists only within very restricted limits, when the difference between H and h is only 3 or 4 mm for standard solutions at 0.050 or 0.100 (Girard and Cuniasse). The actual content will be greater than the content thus calculated, if H is substantially greater than h and lower in the opposite case. It will then be necessary to make one or more successive colorimetric examinations, after having diluted with pure alcohol to 50° either the type or the sample, in order to make the colorations as close as possible.
In order to avoid the long manipulations and the causes of error produced by successive dilutions, we have drawn up graphs which give, for a first test, the corrected result. It will be good however to check the figure thus obtained, in the using to make a new dilution.
Cuniasse (1) drew curves for standard solutions of 0.050 and 0.100, plotting on the ordinate the apparent content calculated from the formula above and in abscissae the actual content. The official French method also gives some values corresponding to various thicknesses of the sample, to allow the construction of a curve of correction:
(1) Kling (A) — Méthodes actuelles d’expertises employées au laboratoire municipal de Paris. T. IV : Alcools et spiritueu, par M. Cuniasse. Paris, 1922.
Bonis (1) indicates slightly different figures for the establishment of the curve:
(1) Ann. Falsif. I. 86, 1908.
Beyond 30 gr of aldehydes per hectolitre of alcohol at 100°, Bonis advises to dilute the sample to be examined with pure alcohol at 50° before proceeding to the colorimetric test, the interpretation of the colorimetric readings for color intensities deviating substantially from that of the type being difficult and may give rise to significant differences depending on the operators. It also recommends that each chemist experimentally establish his correction curve, the conclusions of different operators being, in terms of colorimetric observations, slightly divergent.
This way of proceeding allowed Bonis to obtain, even with marc spirits containing 300 gr and more aldehydes per hectolitre, results never having deviations greater than 1% compared to those provided by the volumetric method of Rocques.
Since the Gayon reagent sometimes gives rosy colorings with an alcohol free of aldehydes, other formulas have been proposed that do not have this drawback. Here is the one recommended by Mohler (2):
(2) C. R. CXI. 187, 1890.
This reagent should be used immediately after its preparation, in the proportion of 4 cc. for 10 cc. alcohol to test. Although less sensitive than that of Gayon, it still allows to detect 1:100,000 ordinary aldehyde (instead of 1:500,000 with the reagent of Gayon).
As it has been found that the reagents based on sodium bisulphite do not always provide, for a given quantity of aldehyde, the same intensity of coloration, which appears to come from the fact that the bisulphite solutions of the same specific gravity do not present the same composition, some authors have proposed to replace bisulphite with a solution of sulphurous gas, the exact measure of which can be determined iodometrically. J. Paul (3) recommended a reagent consisting of 0.05 gr. pure fuschine and 0.5 gr. SO2 in 100 cc of water, a formula that has been retained in the American official method, otherwise similar to the French method.
(3) Z. Anal. Chem. XXXV. 647, 1896.
Swiss official method (Enz. Method) (4). — Add to 10 cc of distillate, reduced to 40°, 1 cc. freshly prepared aqueous solution of 10% metaperrylenediamine hydrochloride (Merek). Shake, let stand 15 minutes, then compare the resulting color with that of standard solutions placed in testing glasses of the same diameter.
(4) Schweiz Lebensm. III, 341, 1017.
These solutions are prepared in the following way (kept in the dark, they are kept for months). Dissolve 0.1 gr. of 00 (Merck) in 500 cc of water. Leave to deposit and prepare, with the liquid that floats above the sediment, the following dilutions, corresponding to the quantities of aldehydes given opposite:
If the content of aldehydes is greater than 1 p. 1000 in vol., Dilute with 40 or 96% alcohol, free of aldehydes and repeat the operation. According to Muller (1), the results obtained by this method are consistent with those given by the Hoepner method.
(1) Mitt. Lebensm. Hyg. XIV, 1923.
[I do not think I will be trying any of these fussy methods.]
The volumetric methods used for the determination of aldehydes in spirits are based for the most part on the principle that aldehydes combine with sulphurous gas or bisulphites, to give additional compounds. The amount of uncombined acid, after contact of the spirit with a standard solution of SO2 or alkaline bisulfite, is determined by iodine. The results provided by these methods are also subject to inaccuracies, because the reaction between the aldehydes and SO2 is reversible, the balance achieved depending on the nature of the aldehydes, the reaction of the medium (pH), the temperature of the the concentration of the solution and the excess reagent.
French official method (Rocques). — Reagents: a) Sodium sulfite solution. Dissolve in 400 cc of water 12.6 g of pure and dry sodium sulphite, then add 100 g. of normal sulfuric acid and make up the volume to one liter with 95% pure alcohol. If crystals of sulphate of soda are deposited, filter them to separate them.
b) Iodine solution. N/10 iodine solution in potassium iodide. 1 cc of this solution corresponds to 0.0032 gr of sulphurous acid, or 0.0022 of ethyl aldehyde. The liquor of soda is titrated by means of the iodine liquor; if the sulphite of soda employed is pure, 10 cc of the first require 20 cc of the second.
Operating procedure – To dose the aldehyde, we introduce, in a volumetric flask of 100 cc with a long neck, the solution to be titrated (10 cc if it contains 5 to 10% aldehyde, 20 cc if it contains 2 to 5% and 50 cc if it contains only 1 to 2%): 50 cc of solution of sulphite: the volume of 100 cc is completed with pure alcohol at 50°: the balloon is shaken and firmly sealed with a cork. A control flask, similar to the preceding one, is prepared in which 50 cc of sulphite solution is introduced; the volume is also completed to 100 cc and stirred.
The two flasks are placed in a water bath heated to 50° C: they are left there for 4 hours, maintaining this temperature. At the end of this time, the mixture is cooled, stirred again and 50 cc of each of the two liquids are taken out, on which the sulfurous acid is titrated by means of the iodine liquor. Care should be taken to add about 50 cc of water, then a little starch solution, otherwise, in the presence of alcohol, the final coloring is dirty red-brown, instead of being a beautiful blue.
Let A be the number of cc of iodine liquor required by the 50 cc of liquid of the control flask, and a the number of cc required by the flask containing the aldehyde solution; the aldehyde content, per liter of the latter, will be:
Jaulmes and Espézel method (1). This method is a refinement of that of Rocques, which is rendered more rapid by a systematic study of the conditions to be fulfilled so that the reactions are complete in the least amount of time.
(1) Ann. Falsif. XXVIII, 325, 1905.
Operating procedure — In a 250 cc conical flask, mix 50 cc of neutral buffer solution, 10 cc of sodium bisulphite solution and a number of cc of the test liquid containing 0.01 to 0.03 gr of aldehyde. Stopper the bottle, shake and let stand 20 minutes.
Add 1-2 cc of starch paste (at 2 per 1,000), 100 cc of distilled water and 10 cc of acid solution. Then pour, with a burette, N/10 iodine liquor, until blue coloring. Add 1-2 cc of phenolphlatein and 100 cc of alkaline solution. Titrate with the iodine liquor until the blue color reappears. If n cc of iodine N/10 has been poured, the quantity of aldehyde contained in the test portion will, in mgr, be 2.2 n.
American official method. — Reagents: a) 0.05 N sodium hyposulphite solution. Perform the titration, using a solution of 0.05 N K-dichromate as follows. Introduce 20 cc. of the dichromate solution in an emery-capped vial and add 5 cc. 15% K iodide solution. Add 2.5 cc of HCL and dilute with 100 cc of CO2-free water and immediately titrate the iodine released with the hyposulphite solution until the yellow color has almost disappeared. Add 1-2 cc of starch indicator and continue, stirring continuously, the addition of hyposulphite until disappearance of the blue color.
b) 0.05 N iodine solution. Titrate this solution with that of hyposulphite.
c) 0.05 N sodium bisulfite solution
Operating procedure — Add 50 cc of the test sample to an Erlenmeyer flask and add 10 cc of water, with some carborundum fragments to obtain a regular boil. Distill 50 cc or a little more into an emery-sealed flask, using a delivery tube dipping into 100 cc of boiled water. Using a pipette, add 25 cc of the bisulfite solution and allow to stand for about 30 minutes, stirring occasionally. Introduce an excess (about 30 cc) of iodine liquor, titrate this excess with the hyposulphite solution and calculate in acetaldehyde, 1 cc of 0.05 N solution = 0.00011 gr of acetaldehyde. The difference between the 2 titrations of hyposulphite (0.05 N bisulfite employed) x 0.00011 = gr. acetaldehyde.
Notes. — Introduce the starch indicator only when the yellow color of the iodine solution has almost disappeared. As the end of the reaction approaches, the solution becomes purple rather than blue. If there is a duplicate for this purpose, add a little more starch indicator. The bluish purple color indicates that the point is not reached. Always perform a blank test of the bisulphite solution for each aldehyde assay using the same amount of bisulfite and iodine tester as for the treatment of the spirits sample.
Hoepner method (1). This method, used by the German government, is based on the property possessed by aldehydes of giving in the presence of hydroxylamine hydrochloride aldoximes, with release of HCl, which is titrated by a soda liquor.
(1) Z. Unters. Nahr. Genusm. XXXIV. 483, 1917.
300 cc of Erlenmeyer flask are charged with 10 cc of spirits reduced to 30° alcoholic strength, then 7.5 cc of a solution of 10% hydroxylamine hydrochloride. The vial is capped, shaken with care and allowed to stand for half an hour at laboratory temperature. The acidity is titrated to 0.1 N sodium hydroxide, with the methyl orange indicator, until it turns yellow. It is necessary to determine in advance, by a blank test, the pure acidity of the 100 cc of hydrochloride, that one discards the figure found. 1 cc of sodium hydroxide N/10 corresponds to 0.0044 gr. acetaldehyde.
[Of all the aldehyde methods, this one is the most appealing.]
According to Mariller, this method would not be advisable for alcohols with low aldehyde content, but would be suitable for very impure liquids.
[Nothing I have come across after a wide variety of role model analysis with the birectifier has convinced me that aldehyde analysis is worth the significant effort. I have experienced excessive aldehydes by distilling fully oxidized desert wines which is a good exercise to gain organoleptic training with acetaldehyde.]
The determination of furturol, or pyromucic aldehyde, is done with great guarantees of accuracy by colorimetry, using the very sensitive reaction of aniline acetate, indicated by Jorissen in 1882. This reagent gives in the presence of furfurol a beautiful grenadine red coloring, still sensitive if the furfurol is diluted to 1 / 10,000,000. The intensity of the coloration is proportional to the furfurol content. It also has a certain constancy, so that one can use, if one has many evaluations to perform, colored types, prepared for example with a solution of safranin (Swiss official method).
When spirits are colored with caramel, it is advisable to carry out their preliminary distillation.
French official method. — 10 cc of alcohol reduced to 50° are introduced into a test tube with 0.5 cc of freshly distilled aniline and 2 cc of glacial acetic acid free of furfurol. At the same time, a comparative test is carried out with 10 cc of standard liquor containing 0.010 gr of furfurol in pure alcohol at 50°. After 20 minutes, the liquors are examined by comparison with a colorimeter, using a thickness of 10 mm.
The furfurol content per liter of alcohol at 50° is given by the formula 0.010 X H/h where H is the height of the standard liquid and h is that of the sample.
[I think height is height on the scale and may be more intuitive if you are familiar with those devices.]
Acetic acid often contains furfurol, or at least one substance giving the same reaction with aniline on the one hand, and the intensity of the coloring being influenced on the other hand by the temperature (Tolman), it is replaced in the official American method acetic acid with hydrochloric acid, and one operates at a constant temperature of 15 ° C.
[These days I suspect we would be buying very pure glacial acetic acid.]
American official method. — Reagents: Typical solution of furfurol: Discolve 1 gr of redistilled furfurol in 100 cc of alcohol at 95°. Prepare the types, diluting 1 cc of the previous solution, with alcohol at 50° (most suitable for common use are those containing 0.05, 0.10, 0.20, 0.25 and 0.30 of furfurol).
The concentrated solution of furfurol retains its strength, but not the diluted solutions.
Operating procedure. — 10-20 cc of spirits, previously distilled as in the case of the ester assay, are diluted to 50 cc with 50° alcohol free of furfurol. 2 cc of colorless aniline and 0.5 cc of HCl (sp. 1.125) are added and the mixture is kept at 15° C for 15 minutes. The standard solutions are treated in the same way, and for the colorimetric examination the color closest to the sample is used.
[Furfurol can be observed organoleptically in the very last fractions of the birectifier. I’ve found no need to quantify it. An explicit count doesn’t help the distiller so much as spirits buyer trying to understand how much pot sitll spirit is in their blend. Furfurol however, could be a proxy for still operation and imply how much time under heat a spirit saw and how deep into the tail an operator went. However, there are a lot of other proxies, proof, volume, and copper content being others.
The esters are usually evaluated by saponification (after saturation of the free acids), by means of a titrated liquor of potash or soda, and then determining the quantity of alkali absorbed. The results are calculated as ethyl acetate, 1 cc. of N/10 potassium hydroxide corresponding to 0.0088 of saponified ethyl acetate. The alcoholic measure being without influence on the absorption of the alkali, the operation can be made without prior dilution of the spirit.
[Good advice here I’ve not seen elsewhere about being able to ignore alcohol content.]
Various causes of error may occur in the assay, which lead to often irregular results. According to Duchemin and Dourlen (1), during the saponification, oxidation of the alcohol under the action of oxygen in the air would take place, with formation of acetic acid, which would give rather large differences in the saponification number depending on the duration and speed of boiling. To avoid this cause of error, the previous authors have proposed a method of operating the saponification in a vacuum.
(1) Bull Ass. Chim. XXIII, 109, 1903.
[I have not seen any modern protocol take this into account.]
At the same time, the saponification figure is influenced by the presence of aldehydes. These are resinified under the action of bases, which leads to a loss of variable alkali and determines a final variable ester measure too high. To remedy this, Barbet and Jandrier (2) proposed to use as saponifiers lime sucrate, which has no action on aldehydes. However, according to Mohler, the presence of aldehydes would only appreciably distort the results if their content exceeds 1 gr per liter, which is rare. In this case, it would also be possible, as Rocques proposes, to dispose of the aldehydes beforehand using aniline phosphate or metaphenylenediamine hydrochloride.
(2) C. R. 2 Cong. Int. Chim. Appl. I, 464, 1886.
[This I think is all theoretical and to my knowledge no modern protocol takes this into account. GCMS made a lot of these techniques some what obsolete, but my understanding is that sometime titration methods are used to practical build models for inline spectroscopy.]
Finally, according to Tolman and Trescot (3), caramel and invert sugar present in spirits absorb a large quantity of alkali. The error from this interference could reach 100% in the case of highly caramelized spirits. It is therefore essential to distill the spirits, before proceeding to the determination of the esters.
(3) J. Am. Cherm. Soc. XXVIII, 1619, 1906.
French official method. — In a 150 cc round glass flask, 100 cc of alcohol are introduced at 50°, 2 grains of pumice stone and 5 drops of 1% phenolphthalene alcoholic solution. The free acids are neutralized exactly with N/10 sodium hydroxide solution (liquor freshly prepared and free of carbonates); 20 cc of N/10 alkaline liquor are then added and it is boiled for one hour under total reflux; let cool, then 20 cc of N/10 sulfuric acid are added and the excess acid is titrated with N/10 sodium hydroxide. Let n be the number of N/10 employed, n x 17.6 gives the content of esters (evaluated as ethyl acetate) per hectolitre of alcohol.
[This is unique because it does not ask you to prepare a blank and is very specific about the amount of sulfuric acid. You over shoot with alkaline to split the esters, then you undue that with a matching amount of acid, and you are left acidic by the acids free from the esters, then you titrate that.]
When the alcohol to be analyzed contains an appreciable proportion of aldehydes, the saponification is carried out by boiling for 2 hours with a titrated liquor of lime sucrate.
US official method. — Introduce 100 – 200 cc of spirits to be analyzed in a flask, add 12.5 – 25 cc of water and distill slowly 100 – 200 cc (according to the volume of the sample taken), using a mercury valve to prevent alcohol loss. Neutralize exactly the free acidity in 50 cc of the distillate, with 0.1 N sodium hydroxide or potassium hydroxide, and add a measured excess of 25 – 50 cc of 0.1 N alkaline liquor. Then, boil for 1 hour at full reflux, and titrate with 0.1 N acid; or leave the solution at rest for one night in a stoppered bottle, heat with a tubular condenser for 30 minutes at a temperature below the boiling point, cool and titrate. 1 cc of 0.1 N alkaline liquor = 0.0088 g of ethyl acetate. Make a blank test, using water instead of distillate, and make the necessary correction.
[This is interesting and I think this protocol has evolved, but I cannot point to the most current. First you distill which seems like it could augment the esters. I don’t know why that wouldn’t be reserved for situations where excess sugar is present. Then you neutralize and then go alkaline. Then you back titrate with acid. The difference becomes what was absorbed by the acids liberated from the esters. What they call a tubular condenser is called an air condenser these days. It is nice to know you can start and stop this if you need to do it over a two day period.]
Nitrogenous substances and bases.
Nitrogen materials can be determined by applying the Kjeldahl method, as proposed by Lindet, or that of Wanklyn and Chapman, for the determination of nitrogenous matter in water, as recommended by Mohler.
It is also rare that the evaluation of bases is currently included in the results of current analysis. Towards the end of the last century, however, it was a quite common operation.
Lindet process (1). — 500 cc or 1 liter of spirits are treated with 20 cc of sulfuric acid. It is distilled to remove water and alcohol, and it is heated in a sand bath to burn organic matter. After one hour of heating, 0.5 gr of metallic mercury is added and the mixture is heated until the liquid has become clear. The operation is then continued as if it were a nitrogen evaluation, that is to say that we dilute with distilled water, add a solution of potash or soda, to thoroughly alkalize the liquid, and add a few drops of a solution of potassium or sodium sulphide, which precipitates mercury; the ammonia is distilled which is collected in a solution of sulfuric acid and titrated.
(1) C. R. CVI, 280, 1898.
[I cannot say I completely understand this operation. I think they are trying to free base various substances and collection them by distillation. They may have to be quickly absorbed by the acid so they don’t react else? I bet by the end of the chapter we find a method that uses all of mercury, lead, and asbestos.]
The bases are expressed in ammonia. Morin proposed the coefficient 100 / 23.5, to calculate bases boiling from 178 to 180°.
Mohler process. — In this process which makes it possible to separately obtain saline ammonia and amides on the one hand, ammonia bases and alkaloids on the other hand, Nessler’s reagent (double iodide of mercury and potassium hydroxide) is used to characterize ammonia, which gives in the presence of this body a yellow-brown coloring.
100 cc of the spirits to be tested are poured into a flat-bottomed flask; 2 cc of phosphoric acid is added at 45 ° B, and then evaporated until complete disappearance of the alcohol. A residue containing the phosphoric solution of the bases is thus obtained, which is subjected to the treatment recommended by Wanklyn and Chapman.
In a 2-liter flask, one liter of distilled water is heated with 20 g of sodium carbonate and then distilled until the condensed liquid no longer gives color with the Nessler reagent. After cooling, the residue of the evaporation of the alcohol is poured into the flask and boiled again, adjusting the heating so that the condensed liquid flows drop by drop. In this operation, the bases are decomposed into ammonia, which distils with water and which generally passes entirely into the first 250 cc; this is assured by treating a small portion of the water which continues to distil with Nessler’s reagent, and if there is still ammonia, the distillation is continued. The condensed liquids are combined in a graduated cylinder and the volume is recorded.
To evaluate the amount of ammonia that has passed through the distillation, 50 cc of ammonia water is poured into a Nessler tube and, in another similar tube, 500 cc distilled water free of ammonia. Two 50 cc tubes of Nessler reagent are added. It develops in the first a clear yellow-brown color proportional to the ammonia content. We then sink in the second tube, drop by drop, a standard solution of ammonia hydrochloride containing 0.1 gr of ammonia per liter. The flow is stopped at the color identity and one or more further tests are made by pouring this amount into the tube before the introduction of the Nessler reagent. until one obtains a perfect equality of coloring. If X represents the number of cc of titrated solution used and V the volume of the condensed liquid, the ammonia content, in gr. per liter, will be:
This gives the weight of saline ammonia and amides. To obtain the ammonia of the pyridine bases and alkaloids, 80 cc of a solution containing 8 g of potassium permanganate and 200 g of potassium hydroxide are added to the residue remaining in the flask. The distillation is resumed by collecting 250 cc of liquid approximately, and then operate as before.
When eaux-de-vie contain a certain quantity of extract, it happens that, during the distillation, the action of the alkali on the sugars determines the formation of ulmic substances; the distilled water is then colored a cloudy lemon yellow by the Nessler reagent. Care must be taken not to attribute this color to the presence of ammonia. Mohler pointed out that in the case of ammonia, the intensity of the coloration is maxima in a few seconds, whereas the coloration due to the volatile compounds from the action of the potash on the sugary matters only occurs after a half minute. It will therefore be necessary to observe the intensity of the reaction immediately after the addition of the Nessler reagent.
The determination of the higher alcohols is the one which presents the greatest difficulty, since there is no clear and sensitive colored reaction for this group of impurities, nor a precise and easily applicable volumetric or weighing method.
The process which seems to be at least the least imperfect is based on the very careful fractionation of the products of distillation (Lindet method, for example), but as it requires a large volume of the sample (corresponding to 10 – 20 liters of absolute alcohol), it is rarely applicable in practice, where one has at most 1 liter of liquid. The more or less empirical methodological differences which have been proposed for the current analysis do not give very exact results. The most widely used are the colorimetric methods of Mohler-Rocques and Komarowky, and the volumetric methods of Röse and Allen-Marquardt, variously modified.
These methods are based on the formation of colored products, as a result of the interaction of higher alcohols and cyclic aldehydes in the presence of concentrated sulfuric acid. According to Von Fellenberg (1), the higher alcohols would be converted by sulfuric acid into unsaturated hydrocarbons, which are then combined with aldehyde to give colored products. In fact, the mechanism of the reaction would be more complicated, according to Bleyer, Diemair and Frank (2).
(1) Mitt, Lebensm. Hyg. I, 311, 1910.
(2) Z. Unters. Lebensm. LXVI, 389, 1938.
The reaction is not specific for higher alcohols. It occurs generally on the one hand with unsaturated hydrocarbons and their derivatives, on the other hand with the various substances which, treated with concentrated sulfuric acid, are converted into unsaturated hydrocarbons. As a result, some aldehydes, ketones, acetones and terpenes may be involved in the reaction. It will obviously be important to eliminate these substances, to perform the determination of higher alcohols.
In the original form of the method, the eau-de-vie was treated with sulfuric acid monohydrate alone, which on contact with the impurities of the ethyl alcohol, gives a color varying from yellow to dark brown. This reaction, reported by Dumas was used by Savalle for the rapid determination of the impurities of alcohols (Savalle test). It has been applied to the determination of higher alcohols, after elimination of aldehydes, by Mohler, Girard and Rocques, and it still constitutes the official French method.
Saglier, to increase the sensitivity of the reaction, proposed adding to the mixture of alcohol and acid a few drops of a solution of furfurol, reagent already used by Udransky (1) to detect fusel oil. Komarowsky (2) observed that various spirits samples did not give a regular coloration with furfurol, replaced it with salicylic aldehyde, whose light yellow tint with pure alcohol changes to more or less dark red in the presence of higher alcohols. Since then, various other aldehydes have been used; benzoic aldehydes, paraoxybenzoic, p-dimethylaminobenzoic, veratric, vanillin, it would be p-dimethylaminobenzoic aldehyde which would give the most sensitive and regular color reaction with higher alcohols (Bleyer, Diemair and Frank).
(1) Z. f. Phys. Chem. XII, 355, 1889.
(2) Chem. Ztg. XXVII, 1086, 1903.
Several processes have been indicated for eliminating spirits products giving with the cyclic aldehydes the same color reactions as the higher alcohols. It has been proposed in particular to treat the spirits with aniline acid phosphate (Mohler), m-phenylenediamine hydrochloride (Girard and Rocques), phenylhydrazine p-sulfonate (Schidrowitz and Kaye), which form with the aldehydes of the stable combinations, which can then be separated by distillation. But the most effective way seems to be to saponify the eau-de-vie in the presence of silver oxide: the aldehydes are thus resinified and the terpenes decomposed (Fellenberg).
[Silver oxide seems common and safe enough that it may have some value in testing for rum oil with the birectifier.]
The colorimetric methods used for the determination of higher alcohols have several causes of error. The intensity of the coloration developed depends, in fact, many factors: concentration of spirits in ethyl alcohol, amounts of aldehyde and sulfuric acid used, temperature at which the reaction takes place (degree of temperature reached and duration of heating), especially the nature of the higher alcohols constituting the fusel oil of the eaux-de-vie. In addition, for a given higher alcohol, this intensity is not proportional to the content of this product, except within very narrow limits.
The stainings provided by the sulfuric reagent alone would be in the following ratios for various higher alcohols in a 1 p. 1000 solution in pure alcohol at 50°, according to Mohler:
Rocques found that the ratio of the color intensities developed respectively by amyl alcohol and isobutyl alcohol in solution in pure ethyl alcohol at 67°7, varied with the level of higher alcohols, but was on average 6 / 10.
With Saglier’s reagent (10 drops of furfurol at 1/1000 per cc of spirits reduced to 50 °), the color intensities would be as follows (Saglier):
Von Fellenberg, with Komarowsky’s reagent, found:
Penniman, Smith and Lawshé represented in the graph below the relative color intensities provided, with various reagents, by the higher alcohols in alcohol solution 1 p. 1,000.
In the graph, the higher alcohols are in the following order: i-amyl, n-amyl, i-butyl, n-butyl, i-propyl, n-propyl alcohols.
With dimethylaminobenzoic aldehyde, Bleyer, Diemair and Frank found the following intensities:
It can be seen that if, under rigorously determined conditions of operative technique and by subjecting the sample to be analyzed and the standard solution to exactly the same treatment, it is possible to eliminate the causes of error due to alcoholic concentration, reaction temperature, etc., on the other hand, it is impossible to avoid those resulting from the use of a given higher alcohol as a type of comparison. This would be possible only by using as a type a mixture of higher alcohols having the same composition as the fusel of the spirits examined, something difficult to achieve, especially in the case of rums, which offer particularly large compositional variations. Depending on the methods, isobutyl alcohol, amyl alcohol or a mixture of these two products is used as the types.
[Basically, comparing to a standard of known composition is a bitch because fusel oil is a mixture of higher alcohols and they all product varying color reactions. I suspect it may be possible to only make comparisons against role models aided by the birectifier which isolates roughly 75% of fusel oil in fraction 5. Your role model would likely have to be within your category of spirit produced, but in the modern era there are lots of remarkably consistent spirits to use as references for what is commercially acceptable. There may also be image recognition tools for colorimetry. A spirit may be its own role model as a fermentation variable changes. With practical effort, we may at least know which direction fusel oil went. All the reagents may be easy to purchase.]
The magnitude of the error due to the type can be reduced by carrying out several determinations with different aldehydic reagents. The observed variations in color intensity will at the same time reveal the nature of the higher alcohols existing in the spirits under examination (Penniman, Smith and Lawshe).
French official method (Mohler-Rocques). — In a 250 cc flask, 100 cc of eau-de-vie is added to analyze, previously distilled and brought exactly to the alcoholic strength of 50°. 1 cc of pure aniline, 1 cc of pure syrupy phosphoric acid and a few grains of pumice are added, the the liquid gently boiled with complete reflux for one hour. At the end of this time, the heating is stopped and when the liquid is cooled, it is distilled. Care must be taken in this distillation to incline the flask to about 45° and connect it to a glass coil by a fairly wide and beveled tube. The refrigerant must be well cooled and about 1 meter long, so that the distilled liquid flows at room temperature. In a small flask, exactly 75 cc of liquid is collected, which contains all the alcohol and marks 66°7 at the alcohol meter. This mixture is made homogeneous by stirring.
[The 45° angle likely reduces reflux as much as possible to make sure all the higher alcohols come across. I’m not sure why the condenser has to be so long. One would think it would risk higher alcohols sticking to its surface area.]
The sulfuric acid is made to act on this liquid. For this we use small test glasses with a capacity of 100 cc, whose neck is cut so that it is 20 cm long. With a pipette, exactly 10 cc of pure and colorless sulfuric acid monohydrate is measured, which is poured along the wall of the flask, so that it meets at the bottom; the alcohol and the acid are then strongly mixed, and the mixture is heated at 120° for one hour in a bath of calcium chloride boiling at this temperature and kept constant by a supply tank filled with water. At the same time as the alcohol or the alcohols to be tested, a flask containing 10 cc of standard liquor is placed in the bath. containing 0.667 g of pure isobutyl alcohol per liter of pure alcohol at 66°2, and 10 cc of sulfuric acid.
[I think the test glasses are boiled open (without re-condensing) because a reaction is taking place and there is no danger of the higher alcohols evaporating. The glasses may be so long for safety since concentrated acid is being subjected to high heat.]
When the alcohol to be tested and the standard solution have been subjected for one hour to the action of the acid at the temperature of 120°, the glasses are removed from the calcium chloride bath and allowed to cool, then compare with a colorimeter, using a thickness of 10 mm.
Since the intensity of the coloring is not proportional to the content of the higher alcohols, it will be necessary to make successive dilutions of the type or of the sample with pure alcohol at 66°7, in order to obtain neighboring colorations, or to use a correction curve. The official French method gives the following indications to serve for the establishment of such a curve:
Bonis (1) gives slightly different figures for the establishment of the curve:
(1) Ann. Falsif. I. 82, 1908.
For contents of higher alcohols exceeding 300 gr / hl of alcohol at 100°, the test liquid should be diluted with pure alcohol at 60°7.
The above figures were obtained by using, for the preparation of the standard liquor, commercial isobutyl alcohol purified by several successive fractional distillations, so as finally to have a product whose boiling point is situated between 106°8 and 107° C. According to Bonis, the main cause of the discrepancies observed between the results obtained by the different chemists lies in the differences of composition presented by the standard liquors, the product which is found in the commerce under the name of isobutyl alcohol, contains often, besides this alcohol, notable amounts of normal propyl alcohol, most probably some normal butyl alcohol and a small amount of amyl alcohol. The author advises therefore to remove impurities by fractional distillation in a ball column. He further recommends that each operator establish a correction curve himself, with the isobutyl alcohol solution he uses or at least that he verifies a few points of the curve he adopts.
[A ball column I think is just a column like the inner chamber of the birecitifer where the balls provide reflux. He may even specifically mean a birectifier! I don’t think they had vigreux columns back then.]
The official French method gives rise to serious criticism. First of all, it lacks sensitivity: from a content of higher alcohols corresponding to 0.100 gr of isobutyl alcohol per liter, the color developed is too weak to be easily compared to that of the standard liquor. The limit of sensitivity can however be retracted up to 0.050 gr, by distilling the spirits beforehand.
At the same time, it uses a standard liquor that does not correspond to the normal composition of the fusel oil of spirits. Since this is usually made up mostly of amyl alcohol, the use of isobutyl alcohol as a comparison type (1) may give results that are more than 50% below reality. Quantin (2) states that in the case of an ordinary rum from Martinique, it obtained a coefficient of alcohol of 152 g higher than the official French method, whereas the fractional distillation indicated a grade of 547 g. The error will be especially important when normal butyl alcohol is present in significant quantities, as in rums of Jamaica for example.
(1) The reasons given for the adoption of isobutyl alcohol as a type are that this alcohol is more sensitive than amyl alcohol to the sulfuric reagent and can be more easily obtained in the trade.
(2) C. R. 4th. Cong. Int. Chim. Appl. II, 266, 1900.
[This may limit the test to only narrowly comparing one rum to a previous generation of the same.]
On the other hand, the French official method would give consistent and comparable results. According to Bonis, by operating carefully in conditions always identical, it allows to never have differences greater than 5%. Since it is also easy to perform, this method is used in various countries outside France, particularly Italy, Belgium, etc. However, the typical liquor is sometimes modified: in Belgium, it is made up of 10% isobutyl alcohol and 90% amyl alcohol.
Komarowsky-Fellenberg method. — Reagents: Sulfuric acid (1 + 1). [1+1 may indicate the monohydrate, but not sure]
—N-silver nitrate solution (170 g per liter);
—Sodium salicylic aldehyde solution: dilute 1 cc. of pure aldehyde in pure alcohol at 95 ° (free of aldehydes), so as to have 100 cc. ;
—Standard solutions of alcohols greater than 2 and 3 p 1000 (3): distill at the time of use, amyl alcohol of pure fermentation and pure isobutyl alcohol, collecting the average fraction apart. Dilute to 250 cc, 1.6 cc of amyl alcohol and 0.4 cc of isobutyl alcohol from the middle portion with pure alcohol at 30° (stock solution). By diluting again to 1 liter on the one hand 61 cc and on the other hand 91.5 cc of the stock solution with alcohol at 30 °, we obtain the two standard solutions.
(3) Dans le cas du rhum Jamaique. Von Fellenberg conseille d’utiliser comme type une solution d’alcool butylique normal a 1 %.
Procedure. — Distill the amount of eau-de-vie corresponding to 20 cc of distillate grading 30°, i.e. 600 cc. Add to the distillate 0.05 cc of sulfuric acid (1 + 1), allow to react for 5 minutes, then neutralize with 30% caustic potash in the presence of phenolpthaleine. Then add 0.4 cc of silver nitrate and 0.2 cc of 30% caustic potash (more for aldehyde-rich spirits), heat for half an hour in a bain-marie with ascending refrigerant and distill finally with an open flame. Collect the distillate in a 20 cc volumetric flask and make up to the mark with water. The liquid thus contains 30% alcohol by volume (1).
(1) For high-strength rums, Von Fellenberg recommends the following procedure: Quickly distill 100 cc of spirits until 85 cc has been collected. Dilute the residue with 120 cc of water and distill it until the distillate reaches 200 cc. Add 2 cc of N. sulfuric acid to the distillate, let stand for 10 minutes, neutralize. Then add 4 cc of 30% sodium hydroxide solution and 7 cc of N. silver nitrate solution, heat with total reflux for half an hour and distill.
[It appears a goal here is to also saponify the esters to release any higher alcohols.]
Add 5 cc of distillate, placed in a 100 cc flask, 2.5 cc of the salicylic aldehyde solution and 2.5 cc of water, then slowly pour down the sides of the flask, suitably inclined, 20 cc concentrated sulfuric acid and mix carefully.
Operate at the same time and in the same way on one of the standard solutions containing 2 or 3 p. 1000 of higher alcohols. 45 minutes after the addition of the sulfuric acid, the mixtures being approximately cooled to room temperature, dilute with 50 cc. of sulfuric acid (1 + 1) and compare the color intensities with the colorimeter. Colorimetric examination should be done immediately after dilution. If the content of higher alcohols exceeds 4 p. 1000, repeat the assay after dilution of the distillate with pure alcohol at 30°. Express results in cc per liter of absolute alcohol.
The Komarowsky-Fellenberg method is used in Switzerland (officially), Germany and sometimes America.
It has also been the subject of various modifications as regards the preparation of the standard solutions and the operating mode.
Recently Fellenberg (2) proposed to replace salicylic aldehyde with para-oxybenzoic aldehyde, an isomer of the previous one, but much more stable and which, according to the author, would have the very interesting property of not giving a reaction with the aldehydes of the fatty series. Pure alcohol gives no appreciable hue with this reagent, while in the presence of higher alcohols, a purplish hue of redness becomes more and more red as the rate of these increases. The procedure adopted is also a little different: use of a less concentrated sulfuric acid (density = 1.77), reaction taking place in a boiling water bath for 5 h, in the presence of the 1.5% solution of para-oxybenzaldehyde in 30% alcohol.
(2) Mitt. Lebensm. Hyg. XX, 16, 1929.
Penniman, Smith and Lawshé (3) advocated the following process:
(3) Ind. Eng. Chem. (Anal. Ed.) IX, 91, 1937.
With 25 cc of the sample placed in a 500 cc round bottom flask, add 0.5 gr of silver sulphate and 1 cc of sulfuric acid (1 + 1), and bring the volume to 110 cc. Boil gently with total reflux for 15 minutes, then add 5 cc of sodium hydroxide (1 + 1) and heat again with total reflux for 30 minutes. Distill the liquid and collect 75 cc of distillate. The alcoholic measure of it has been reduced to one-third of the original sample.
Place 2 cc of distillate in a 125 cc flask. Add 20 cc of concentrated sulfuric acid, keeping the flask immersed in a cold water bath during the addition, then 2 cc of a solution of the reagent containing 10 mg of salicylic aldehyde (or dimethylaminobenzoic aldehyde) per cc of 95° alcohol and place again into cold water (4).
(4) If vanillin is used as a reagent, use 10 cc of sulfuric acid only and 2 cc of a solution of 17.5 mgr of vanillin in 1 cc of 95 ° alcohol.
[I’m not sure if that first 2 cc is a typo or not.]
Operate in the same way on 2 cc of standard solution. Place the 2 flasks simultaneously in a boiling water bath and, after 20 minutes, in a cold water bath. After cooling, add 25 cc sulfuric acid (1 + 1) and mix well, subjecting the flasks to a rotary motion. Compare colorimetric intensities.
[2 cc sounds really small, but it may not be a typo.]
The authors advise using as a type a mixture of higher alcohols of the same kind and proportions as those forming the fusel oil of the spirits examined or, failing that, a mixture of isoamyl and isobutyl alcohols (4: 1). The standard solution must have approximately the same alcoholic strength as the prepared distillate, ie 15° in the case of an eau-de-vie originally having 45 or 50°. Since the color intensities are not proportional to the contents of higher alcohols, it is obviously necessary to refer to special tables, or better to make successive tests so as to obtain similar shades for the type and the sample examined.
[Colorimetric methods leave a lot to be desired. Modern Lovibond colorimeters also appear to be expensive. Likely no one will be using references from standards, but there may be value in making in house comparison during product development stages if GC-FID is still too expensive. Sulfuric acid monohydrate may be nothing more than pure sulfuric acid mixed 1:1 with water which likely can be formulated by density. Birectifier fraction 4, which holds 75% of the fusel oil, but remove a lot of the bias in methods and be useful to build upon. Many of the methods seem overly intricate only to get results of little confidence.]
Rose method. — It is based on the property of chloroform to dissolve the impurities of alcohol more easily than the ethyl alcohol itself, when it is diluted with water. The difference between the volumes of chloroform, after stirring with pure alcohol on the one hand and with impure alcohol on the other hand, is proportional to the content of the latter in impurities, if the test is carried out in the same conditions and alcohol on the same basis.
[Variations of this principle were used to better isolate the rose ketone damascenone in rum. I may be worth quoting at length so we can see the use varied uses of these solvents. Keep in mind, this does not directly apply, but we likely have more pure solvent options than they did back then:
The major constituents of “non-alcohol” spirits are either polar compounds such as propyl, isobutyl and isoamyl alcohols; or low polar compounds such as ethyl esters of higher fatty acids and some acetals. Based on the results obtained by WILLIAMS and TUCKNOTT (1973), we used pentane to avoid the extraction of a large quantity of higher alcohols. With this same solvent, on the other hand, the low polar compounds are very well extracted even when the alcoholic degree of the hydroalcoholic solution is high.
Thus, on a spirit at 75°GL, a single pentane treatment makes it possible to eliminate most of the ethyl esters of the higher fatty acids, and in particular the heavier ones which have little interest in the olfactory plane. Following this first operation, the dilution of the hydroalcoholic solution with 75°GL by four times its volume of water makes it possible to extract, still with pentane, the most interesting constituents of the “non-alcohol”.
So, down to 15%ABV. I’ve seen other people use 13%]
All impurities do not determine the same increase in volume at the same level. According to Sell, essential oils diminish the absorbency of chloroform; it is the same with methyl alcohol. Esters increase the volume of chloroform, but less than amyl alcohol. Aldehydes are poorly soluble, except for furfurol, which acts about as much as amyl alcohol. Finally, the various higher alcohols have different solubilities. The volume increases of chloroform produced by them, in solution in alcohol at 30° and at a temperature of 15° C, would be in the following ratios (Sell):
In order to apply the Röse test to the determination of higher alcohols, Stutzer and Reitmayr (1) recommended the elimination of other secondary products which affect the volume of chloroform, by pre-distilling the alcohol with a small quantity of soda liquor or potash: the aldehydes are thus resinified, the esters are saponified and the acids saturated. Von Fellenberg recommends that, in addition to silver nitrate treatment, it should be rid of essential oils and terpenes which, by diminishing the dissolving power of chloroform, lead to results which are too low. Unfortunately, methyl alcohol, which acts in the same direction (Fellenberg (2)), can not be eliminated.
(1) Rep. Anal. Chem. VI, 335, 1886.
(2) Mitt. Lebensm. Hyg. XX, 16, 1929.
[Fractioning by birectifier may remove a lot of these biases so we can focus on fraction 4.]
To carry out the exhaustion of the alcohol with chloroform, the Röse tube, modified by Herzfeld and Windisch, is used. The glass apparatus has a cylindrical lower tank with a capacity of 20 cc up to the first line of the graduation, connected by a graduated tube 18 cm long and with a capacity of 2.5 cc, divided into hundredths of a cc, to an upper piriform reservoir, having a capacity of about 200 cc, closed by an emery plug.
[I have seen pycnometers have have similar shapes. These days it would have to be custom made. Keep in mind one liquid layer floats above the other. You are hoping the immiscible layer of chloroform grows by only a very small degree.]
The Röse method, which is officially used in Germany and Italy (in the latter country concurrently with the Rocques method), if it gives relatively exact results, is very delicate.
In order to have reliable figures, it is important to operate at a controlled temperature (+15° C) and with alcoholic liquids strictly controlled (30% by volume), the absorbency of chloroform being under close dependence on temperature and alcoholic strength. A difference of 1 degree of temperature, for example, corresponds to a variation of the chloroform layer of 0.1 cc, whereas for a content of 1 p. 1,000 of amyl alcohol (in pure alcohol at 30° and at a temperature of 15° C), the increase in volume of chloroform is only 0.2 cc. With regard to the alcoholic strength, an error in plus or minus 0°1 causes a difference of 0.0199% in the volume of the higher alcohols. Error tables can only be used to indicate approximately the quantity of water to be added, the definitive measure to be carefully checked, by taking the density using a Westphal balance or pycnometer. This density shall be 0.96557 at 15° C, with a difference which shall not be greater than 2 units of the fifth decimal place.
[This accuracy required basically makes this method not viable. What I don’t understand is why you cannot evaporate the chloroform.]
The chloroform used must be very pure (1) and dry. Despite the care taken in its conservation, this product is easily altered and its dissolving power is consequently modified. It will not be enough, to have the increase of volume due to the higher alcohols, to deduce from the increase in total volume that resulting from the dissolution of ethyl alcohol and which is at 15° C of about 1.64 cc per 100 cc of alcohol at 30%, because this figure may vary more or less. It will be important to make a blank test with pure ethyl alcohol at 30°, which will serve as a comparison. It will also be helpful, if one wants to have more precise results, to carry out two or more successive tests with the spirits to be examined, because one observes from one experiment to the other quite sensible variations in the volume of the chloroform.
(1) To purify the chloroform, it is stirred with concentrated sulfuric acid: the acid layer is separated and the chloroform is then washed several times with water in a decanting funnel; the last traces of acid are neutralized by stirring with a little dilute solution of carbonate of soda. The chloroform is dried, leaving it in contact with calcium chloride for one day and finally distilled.
All instruments used for the test (Röse tube, pipettes, etc.) must be thoroughly clean. They will be thoroughly washed with distilled water, then with alcohol and finally with ether; heat slightly on an alcohol lamp or in an oven, so as to drive off the ether vapors. From time to time, it will be necessary to clean them with a solution of dichromate in concentrated sulfuric acid.
Operating procedure. — To remove liquor impurities other than higher alcohols, 100 cc of the alcohol is usually treated with a few drops of a concentrated solution of caustic potash; then boiled for one hour with total reflux, then the bulk of the liquid is distilled and the distillate is brought back to the original 100 cc volume. However, it is preferable to use the silver nitrate treatment already described for the Komarowsky method.
The Röse test, with the modifications that have been made by Sell, Stutzer and Reitmayr, is practiced as follows:
We introduce in a Röse tube, using a dropping funnel with a long bushing which reaches the graduation 20 and taking care not to wet the walls of the tube, 20 cc of chloroform pre-reduced to 15 ° C. The apparatus was immersed 3/4 in a 15° water bath and allowed sufficient time to bring about the temperature equilibrium (about 5 minutes). If, after this time, the chloroform does not exactly reach the 20 cc mark, the volume is adjusted by adding or removing a little liquid with a long capillary tube.
At this point, 100 cc of alcohol to be tested, reduced to 30% and 15°, are poured into a tube, then 1 cc of sulfuric acid diluted to 1.286 density (to facilitate later the gathering of chloroform). The tube is closed and the equilibrium is allowed to settle again by immersing in the water bath for about 15-20 minutes. The apparatus is then removed from the bath, inverted 2 or 3 times to displace the chloroform from the lower reservoir, then, keeping the tube horizontally or slightly inclined, it is agitated (preferably with the aid of a mechanical stirrer) by a series of very sharp shaking (150 shaking for a period of one minute), so as to bring intimate contact alcohol and chloroform. To prevent the temperature from rising during agitation, it is preferable to stir under water or under a tap of water at 15° C. The apparatus is then straightened so as to bring the chloroform into the cylindrical reservoir and is re-immersed in the water bath; Chloroform deposition is facilitated by occasional rotation of the tube about its axis, and striking a few sharp blows against the walls. After one hour, the volume is read.
[One thing to keep in mind is chloroform is trichloromethane and has a specific gravity of 1.490 so it is much denser than water and forms the bottom layer. Dichloromethan is used for a lot of spirits analysis because it is extremely polar.]
Simultaneously and exactly the same is done for the control, using instead of spirits pure ethyl alcohol at 30 °.
The difference between the increase in volume of the chloroform in the tube containing the alcohol to be examined and the increase in volume in the tube containing the pure alcohol corresponds to the quantity of the higher alcohols. According to Sell, a volume increase of 0.01 cc represents a proportion of amyl alcohol of 0.006631% by volume. Calculations will be avoided, using the Stutzer and Reitmayr table, which gives the volume proportion of amyl alcohol contained in 100 cc of alcohol reduced to 30 °, according to the increase in the volume of chloroform.
The results are generally expressed in amyl alcohol. However, if the alcoholic liquid to be examined contained mainly normal butyl alcohol or isobutyl alcohol, it would be preferable to evaluate the latter, by making use of their respective coefficients of chloroform increase indicated by Sell, it is ie by multiplying the results expressed in amyl alcohol by 1:0.57 and 1:0.50.
Allen-Marquardt method. — This method, developed by Dupré, Marquardt and Allen, is based on the extraction of higher alcohols by means of carbon tetrachloride, followed by their oxidation to transform them into fatty acids, which are titrated with an alkaline solution.
[This seems to be the most important method. Hopefully we can understand it!]
Although it is used officially in Anglo-Saxon countries, the Allen-Marquardt method is not very precise. According to Valaer (1), it gives only 80 to 90% of the amyl alcohol actually present, and figures far too low for butyl and propyl alcohols. Isopropyl alcohol is converted by oxidation and escapes as a result of the assay. As for the secondary isobutyl alcohol, if it is oxidized in the state of acetone, it does not enter either, whereas if the oxidation is more thorough, it provides 2 molecules of acid acetic acid per molecule of alcohol and consequently too high figures.
(1) J. Ass. Off Agr. Chem. XIX, 187, 1936.
[Shit!, I thought it was precise!]
Penniman, Smith and Lawshé (2), analyzing mixtures of known proportions of various higher alcohols with ethyl alcohol at 50° by the official US method, have reached the following conclusions. The significance of the error affecting the results of the analysis varies with the concentration and nature of the alcohols. It is greater when the content of higher alcohols is high and for the propyl and isopropyl alcohols. The method can be considered as giving on average 60% of the actual fusel oil, but in some cases the proportion can fall to 40%.
(2) Ind. Eng. Chem. (Anal. Ed.) IX 91, 1937.
The consistency of the Allen-Marquardt method is not great either. In a series of “tests” carried out in 1909, under the auspices of A.O.A.C. by various laboratories in the United States, the results found represented, according to the chemists, 42 to 145% of the amyl alcohol actually present (Tollman (3).
(3) U.S. Dept Agr. Bur. Chem. Bull. 122. 28. 1909.
[I did not know this and its a cool detail. In the olden days you could be a referee for new analysis techniques. Harris Eastman Sawyer, the New England rum distiller at Felton’s & Sons was one of them. He leveraged that connection to get the distillery incredibly favorable treatment as prohibition loomed. Felton’s was able to produce rum and only denature it at the last moment with nicotene to sell to the tobacco industry. This helped them amass legendary old reserves during prohibition. Many of those rums are featured in Kervegants texts from analysis conducted by Peter Valaer and published in 1937.]
Since it was perfected by Allen in 1891, this method has been the subject of various modifications. We give below the “modus operandi” currently in use in the United States.
US Official Method — Reagents — (a) Purified carbon tetrachloride. Mix, in a funnel, crude tetrachloride with 1/10 of its volume of sulfuric acid; shake vigorously at frequent intervals and allow to stand overnight. Eliminate the acids and impurities by washing with ordinary water, add an excess of sodium hydroxide and distill the tetrachloride.
[Carbon tetrachloride is quite volatile and boils at 76.72°C. It is also toxic. Its use seems to be superseded by Tetrachloroethylene which is essentially dry cleaning fluid, but I’m not sure if that applies to spirits analysis.]
b) Oxidizing solution. Dissolve 100 gr. of K dichromate in 900 cc of water and add 100 cc of sulfuric acid.
[Potassium dichromate is another dangerous chemical. Very cool red color.]
Operating procedure – To 50 cc of the spirit to be analyzed, add 50 cc of water and 20 cc of 0.5 N sodium hydroxide solution. Saponify the mixture, boiling with total reflux for 1 hour. Distill 90 cc, add 25 cc of water and continue distillation until 25 cc of additional distillate is obtained.
If the aldehyde content is greater than 15 p. 100,000, add 0.5 gr of m-phenylenediamine hydrochloride to the distillate, boil with total reflux for 1 hour, distill 100 cc, add 25 cc of water and continue distillation until 25 cc of additional distillate is obtained.
Saturate the distillate with finely ground NaCl by adding a saturated solution of NaCl until the specific gravity is 1.10. Perform on this solution 4 successive extractions with pure carbon tetrachloride, using respectively for each treatment 40, 30, 20 and 10 cc of solvent. Wash the tetrachloride 3 times with 50 cc of a saturated solution of Na chloride and 2 times with a saturated solution of Na sulphates.
[I suspect you use a separatory fennel for the carbon tetrachloride extractions. When you wash with Sodium chloride or sulphate those are dissolved in water so I imagine you will again be using the separatory funnel.]
Transfer the tetrachloride to a flask containing 50 cc of the oxidizing solution and boil for 8 hours under total reflux. Add 100 cc of water and distill until only 50 cc remains. Add 50 cc of water again and continue distillation until the residue is reduced to 35-50 cc. Take care that the oxidizing mixture is not deposited and burns on the walls of the flask. The distillate must be colorless like water; if it is colored, it must be rejected and restart the operation. Titrate the distillate with 0.1 N sodium hydroxide using phenolphtaleine as an indicator.
[Why would it not be clear? and 8 hours?!]
Make a control test on 100 cc of C-tetrachloride, starting at the stage immediately after extraction and just before washing with chlorine and Na sulphate solutions.
The results are expressed in amyl alcohol, 1 cc of 0.1 N NaOH = 0.0088 gr. of amyl alcohol.
Value of the different methods.
From what has just been said, it follows that the methods used for the determination of higher alcohols lack accuracy. Nor do they provide concordant figures.
Schidrowitz (1), for example, obtained the following results, applying to the analysis of whiskey samples the 3 official methods Allen-Marquardt, Röse and Rocques:
(1) J. Soc. Chem. Ind. XXI, 814, 1902.
There is no proportionality between the figures above (2).
(2) It should be pointed out, however, that the methods used at that time were quite different from the current methods with the same name, except for the French method.
On the other hand, the operating methods of the official methods, except for the French method, which has remained the same for more than half a century, have been the subject of the various modifications, which have had a significant influence on the order of magnitude of the results obtained.
As a result, the indications found in the books and technical articles relating to the content of spirits in higher alcohols have only a poor comparative value.
The volumetric methods are long and difficult to apply. The one from Röse makes it possible to obtain relatively exact figures, but with fairly pronounced differences from one analysis to another. The results provided by the Allen-Marquardt method are substantially inferior to reality.
[I wonder if easier vacuum distillation allows any improvements on the Röse method.]
The colorimetric methods are easier to run than the volumetric ones, but the higher levels of alcohol found can be much more remote from reality, if a standard liquor is used departs significantly from the composition of the fusel oil of the spirits examined. From this point of view, the French official method appears particularly flawed, especially for the analysis of grand arôme rums. That of Komarowsky-Fellenberg, by the use of p-dimethylaminobenzoic aldehyde and a suitable standard, could on the other hand give results at least as accurate as the Röse method. This would be the most advisable for everyday use.
It has been recognized for some time that methyl alcohol exists in all natural spirits, in varying proportions, but generally low. The evaluation of this body, which was formerly carried out only if one suspected the fraudulent use of denatured alcohol, tends today to become a more common practice, because it provides interesting indications on the nature and the origin of spirits.
There are many colorimetric methods for the detection and determination of methyl alcohol. They are based on the transformation of the latter into methyldehyde, or methanal, by moderate oxidation, by means of potassium permanganate in a sulfuric medium (Ferrière and Cuniasse, Fendler and Mannich) or phosphoric acid (Denigès), or of the sulfochromic mixture (Trillat ); then on the characterization of this aldehyde using various reagents: fuschine bisulfite (Denigès), gallic acid (Barbet and Jandrier, Ferrière and Cuniasse), dimethylaniline (Trillat), phenylhydrazine hydrochloride (Schryver, Bertrand), morphine hydrochloride (Fendler and Mannich), etc.
According to Flanzy, the colorimetric methods are all more or less defective, because the transformation of methyl alcohol into methanal under the effect of moderate oxidation would never take place with a total yield (it would generally be between 4 and 10%) or at least constant. In most of the oxidation techniques used, the presence of methanal would also not be characteristic of the initial presence of methyl alcohol, because, at temperatures close to 100°, there is often formation of methanal from ethyl alcohol.
The author advocates a rather complicated method based on:
a) the transformation of the primary alcohols into iodides, using hydriodic acid, and their subsequent regeneration, under the effect of an aqueous solution of silver acetate: the iodination index P is thus determined;
[This is the most over my head things in the chapter I’ve read so far.]
b) the cold oxidation by the sulfochromic mixture of regenerated primary alcohols, with determination of oxygen: the oxidation index p is obtained. This oxidation, carried out under specific conditions, differentiates methyl alcohol from its counterparts, the oxidation of the former being complete and giving as the final term of carbon dioxide, while that of ethyl alcohol and other alcohols results in corresponding acids.
Each formula of iodination and oxidation allows the establishment of an algebraic relation. We thus have a system of 2 equations with 2 unknowns, from which we can draw the quantity of methyl alcohol existing in the mixture.
According to the author, this method would be accurate and sensible: it would allow, when the ratio of methyl alcohol: ethyl alcohol reaches 1/1000 to 1/2000 and by the use of rigorously pure reagents, to dose 1/2000 methyl alcohol, with an error of only 5 to 10%. However, Flanzy’s method is long and delicate; it can not therefore be easily used for routine analyzes.
The method most commonly employed is that elaborated by Denigès (1) in 1910, and since then variously modified.
(1) C.R. CL 632, 1910.
It rests, as we have already pointed out, on the transformation of methyl alcohol, by moderate oxidation by means of potassium permanganate, into methyl aldehyde. This body possesses the property, to the exclusion of other aldehydes, of staining in strongly acidic medium. Fuschine bisulfite (Schiff reagent) stains violet.
We give here the method of Denigès, modified by Georgia and Morales on the one hand and by Von Fellenberg on the other hand.
Georgia-Morales Process – Reagents: a) Manganese solution: dissolve 3 gr. K permanganate and 15 cc of syrupy phosphoric acid (85%) in 100 cc water;
b) Oxalic-sulfuric solution: dissolve 5 gr. oxalic acid in 100 cc of sulfuric acid (1 + 1);
c) Rosaniline bisulfite solution: Dissolve 0.2 gr of Kalminbaum rosaniline hydrochloride in approximately 120 cc of hot water. Cool and add 2 g of sodium bisulphite previously dissolved in 20 cc of water, then 2 cc of HCl. Bring the solution to 200 cc and keep in a cool place in emery-clogged yellow bottles.
Operating procedure — Introduce 25 cc of the spirits to be analyzed in a 250 cc Erlenmeyer flask, which is joined to a fractionation column. Collect 8.5 cc of liquid, in 7 fractions of about 1.2 cc each. Allow the column to reflux for 30 minutes before collecting the first fraction and 15 minutes between each of the following fractions. This gives a distillate containing about 94% alcohol by volume. Then dilute the distillate to 22% alcohol.
[This seems a little bit like birectifier territory, but I have never heard of anything like it. Those are very small volumes and put this squarely in micro distillation territory.]
Place 0.25 cc of diluted distillate in a 15 cc Nessler tube containing 4.75 cc of water. Add 2 cc of the manganese solution and let stand for 10 minutes, shaking occasionally without spilling the tube. To remove the excess permanganate, add 2 cc of oxalic-sulfuric solution, then, when the liquid has become discolored, 5 cc of rosaniline bisulfite; mix well by inverting the tube 3 times: stop and let rest for 1 hour. Compare the resulting staining with those of standard solutions containing known proportions of methyl alcohol (0.02, 0.04, 0.06, 0.08, methyl alcohol by volume %) in pure alcohol at 22° and treated as samples to be examined.
When the methyl alcohol content exceeds 0.15%, the developed color is too intense to allow precise colorimetric discrimination (Valaer). In this case, dilute the sample with 22 ° alcohol.
This method is used in the United States.
Fellenberg procedure. — Von Fellenberg (2) has observed that the intensity of coloration developed depends on many factors: concentration of ethyl alcohol, composition of fuschine bisulfite, quantities of permanganate and sulfuric acid employed, duration of the reaction, etc. For accurate results it is important that these factors are closely controlled. On the other hand, it is necessary to distill the spirit beforehand because the oxidation of certain elements of the extract (glycerine for example) with permanganate can give rise to methanal. On the other hand, it is not necessary to get rid of volatile acids, aldehydes or terpenes by saponification in the presence of silver oxide. Isobutyl and isoamyl alcohols influence the reaction somewhat, but they do not occur in relatively high proportions in natural spirits to constitute a disorder.
(2) C.R.S. Cong. Int. Ind. Agr. I, 184, 1937
Von Fellenberg, following his observations, proposed the following method:
Reagents. — 95 ° ethyl alcohol free from methyl alcohol: subject extra-fine alcohol to fractional distillation by means of a strong fractionating column and retain only the last fractions free of methyl alcohol; [These last fraction may only be the last half of the distillation run.]
—Ethyl alcohol at 6°67 free of methyl alcohol;
—Concentrated sulfuric acid, very pure;
—Solution alcohol-sulfuric acid: Dissolve 21 cc of 95 ° alcohol (free of methyl alcohol) in about 50 cc of water, add 20 cc of concentrated pure sulfuric acid, cool to room temperature and bring to 100 cc;
—Sulphuric acid diluted: dilute with 100 cc, with water, 20 cc of pure concentrated acid;
—Solution of potassium permanganate: 5 gr. in 100 cc;
—Oxalic acid solution: 8 gr. in 100 cc;
—Fuschine bisulfite: dissolve 1 gr. crystallized fuschine from Merck in 200-300 cc of hot water, cool to room temperature, add 12.5 g of crystallized sodium sulphite, previously dissolved in 200 cc of water and 100 cc of N. HCl, then extend to 1 liter . The solution can be used after a few hours (after discoloration); kept in the dark, it is kept for several months.
—Methyl alcohol solution: dilute to 1 liter, 10 cc Merck pure methanol, free of acetone and extend the liquid to 10 times its volume: 1 cc of the solution contains 1 mmc of methyl alcohol.
Operating procedure. — Distill 100 cc of the previously diluted spirits to be analyzed until 2/3 of the liquid has been collected. Bring the distillate to 100 cc and measure the alcoholic strength. Then dilute 100 cc to 667 cc: a, where a is the alcoholic strength. In the case where the spirits are devoid of extract, the distillation is not necessary: the dilution can be carried out immediately to 6°67.
[He has a really strange way of asking you to dilute a liquid with an unknown variable strength.]
Introduce 3 cc of the above distillate into an 18 mm ID test tube with 1 cc of dilute sulfuric acid and 1 cc of manganic solution. Shake and let stand exactly 2 minutes. At the same time, prepare several types of comparison with methyl alcohol contents that are higher and lower than the sample that is being tested. To do this, place the appropriate amounts of methyl solution in test tubes, dilute with water to 3 cc, add 1 cc of alcohol-sulfuric acid solution, and proceed to the oxidation with the permanganate as before.
After 2 minutes, pour 1 cc of oxalic acid into each tube and shake: the color becomes light brown or disappears completely. Then add 1 cc of concentrated sulfuric acid, shake, immediately add 5 cc of fuschine bisulfite and stir well again.
In a few minutes, there is a coloring from blue to purple. It can be seen then if the types of comparison were well chosen, In case the coloring of the distillate would be too pronounced (more than 3 mg of methyl alcohol) it would be necessary to prepare a new sample, by taking a quantity of inner distillate at 3 cc and diluting it to 3 cc with 6°67 alcohol.
After a rest of exactly 1 hour, examine with the colorimeter (preferablya Lange electric colorimeter). The quantity of methyl alcohol in mmc contained in 0.2 cc of absolute alcohol is obtained. By dividing by 2 the figure obtained, we have methyl alcohol in volume % of absolute alcohol.
When methyl alcohol exists only in very small quantities, it is preferable to operate at an alcohol concentration of 0.07 cc. For the preparation of the alcohol-sulfuric acid solution, use 7.37 cc of 95° alcohol and reduce the strength of the diluted alcohol solution and the distillate to 2°33. The volume of methyl alcohol % of the absolute alcohol is obtained by multiplying the figure found by 1.43.
[I have a feeling no one is going to perform this, but it is interesting to see if we can understand their chemistry and assess the time they invested. Grappa producers will be the most interested in methanol. Unless you have in house chromatography, a wet chemistry test for methanol will never be cheaper than contracting the evaluation to a modern lab.]