CHANGES IN WHISICY WHILE MATURING s
A. J. LIEBMANN -4ND BERNICE SCHERL Schenley Distillers Corporation, New York, N . Y . T h e maturation of whisky represents a complicated chemical process influenced by a number of contributing factors. The changes which are taking place in the course of maturation have been made the subject of several prior investigations which, however, were based on a number of samples which cannot be considered significant under modern statistical procedure. Four hundred sixtynine barrels of whisky, mostly bourbon and rye, were set aside periodically in lots of five during a period of 4 gears. Samples from these barrels w-ere withdrawn at regular intervals over a period of 8 years, and their significant characteristics were determined by analysis. The data so ohtained were transferred to punch cards and statisti-
T
HIS investigation of whisky while maturing represents a
continuation of the work reported and published in 1943 (6). Result,s then given were in the nature of a preliminary report for the first 4 years of a total observation period of 8 years, vhich has since been terminated. The complete results and conclusions are here presented. For the purpose of clarification of the intent of this study, it should be stated again that the quality of a beverage such as whisky is judged by the consumer on its properties affecting taste and odor. The development of a whisky is dependent on three fundamental production operations: fermentation, distillation, and maturing. This investigation is concerned solely with the last phase, maturing. The maturing stage begins immediately after distillation, as the new colorless whisky is drawn into charred oak barrels. The barrels are placed in bonded warehouses (constructed of wood, brick, or concrete) and usually permitted to remain for four or more years under variably controlled conditions of temperature and humidity. The mat,uring stage ends with the v,4thdrawal of the aged TThisky from the barrels. Three fundamental environmental factors are intimately related t o the development of a whisky during maturing-temperature, humidity, and ventilation. The barrel, serving primarily as a container, also act,s as a semipermeable membrane and permits the passage of alcohol and mater vapors from the interior of the barrel to the outside. This phenomenon is an integral phase of maturing. Under the conditions normally prevailing in storage warehouses-especially as to humidity and teinperaturethe barrel permits water vapor to escape a t a faster rat,e than alcohol vapor. Consequently, there is a gradual accumulation of alcohol a t the expense of water inside the barrel. Thus the proof of the contents rises with age. Many warehouses are constructed arid equipped to allow positive control of this phase of the maturing process. During this period, the maturing whisky undergoes definite and intended changes in aromatic and taste Characteristics. These changes are caused by three major types of reactions occurring continually in the barrel: Extraction of complex wood substances by liquid Oxidation of the orginal organic substances and of the extracted vood material Reaction between various organic substances present in the liquid t o form new products
534
cally examined. The means and standard deviations of eleven major constituents and characteristics of whisky were obtained for each age (period of storage) for all or most of the 469 barrels. The eleven characteristic properties of whisky used for an interpretation and definition of maturing were proof, total acidity, fixed acidity, esters, aldehydes, furfural, fusel oil, solids, color, tannins, and pH. The large samples permitted the calculation of dispersion limits. The average maturing characteristics of American whisky over an 8-year period and results of the chemical and statistical examination of the entire sample group are shown. The possibility is indicated for use of the data as a basis for specifications for w7hisky.
The development of quality, therefore, consists of specific chemical and physical changes in the properties of the liquid, some of which are relatively simple to determine. These properties, or Characteristics, are commonly used as a guide to and measure of quality. Experience and observation have shown that abnormalities arising in one or several of the physicochemical characteristics generally will result in abnormalities of the taste characteristics of the liquid. Despite the importance of the maturing development to an industry with an estimated yearly sales volume of about 5 billion dollars and with an excise revenue of more than 2.5 billion dollars to t,he Federal Government alone, only a small quantitv of pevtinent data have been made available by publication. The two previously published investigations originated in t,he laboratories of the .4lcohol Tax Unit of the Treasury Department (2, 9). The techniques of sampling, analysis, and data presentation are almost identical in both cases, although 28 w a r s separate the two investigations. The earlier work was based on 31 different barrels, all chosen from 3-month production in 1898. Most of these whiskies were dist,illed in types of stills now obsolete and stored in warehouses entirely different from those of today. The later work was based on 22 different barrels chosen from a 6-month production period of December 1929 to May 1930. Of the 22 barrels, three contained whiskies derived from such low-yield production as to indicate abnormal fermentation or distillation and thus result in atypical products. These highly skilled investigators worked under the handicap of being forced to rely on basic material and data furnished by producers not under their control, and on conditions not subject to their direct supervision. In both investigations, t,he average values for each characteristic were determined, and the minimum and maximum values (constituting the range) were used to indicate scatter. The exclusive use of the range as a measure of dispersion must be treated with caution sincc it' is subject t,o considerable individual variation. Crampton and Tolman as well as Valaer and Frazier were forced either t'o indicate next highest or lowest values or to delete extreme values which were considered abnormal (3, IO). The subjective deletion of dat,a, even when based on considerable experience, is a dangerous procedure. I n contrast' to these two studies the present work was based on a variety and number of samples which may be considered adequate For modern statistical calculations.
INDUSTRIAL AND ENGINEERING CHEMISTRY
March 1949
SAMPLES
Straight whisky is defined by the Treasury Department (8) as an alcoholic distillate from a fermented mash of grain distilled at not exceeding 160 proof and aged for not less than 24 months in charred new oak containers. Bourbon whisky is usually produced from a mash containing 60 to 88% corn; rye whisky generally comes from a mash containing 51% rye. These two types make u p the greater portion of American whisky. For the purpose of this investigation, there were set aside, with the consent of the Alcohol T a x Unit of the Treasury Department and with certain provisions for periodic withdrawal of samples, a total of 469 barrels of bourbon and rye whisky, produced during the years of 1937 t o 1942. Groups of four to six barrels were allotted biweekly during the whole year of 1937, and similar or gradually decreasing quantities were allotted during succeeding years. Table I shows the general distribution of these samples with respect t o mash type, distillation, characteristics, treatment, cooperage, warehouse, type, origin, and storage location. The 469 barrels constitute a comprehensive and representative cross section of the product obtained by the various methods used in manufacturing whisky. One-pint samples were withdrawn at the age of 0, 1 , 3 , 6 , and 12 months and every 6 months thereafter up to the age of 4, 6, or 8 years. The samples mere drawn by the plant chemists at the various distilleries and then forwarded t o the central research laboratories for immediate analysis. ANALYSES
rlnalyses were performed in most cases t o determine the following characteristics: proof, total acidity, fixed acidity, esters, aldehydes, furfural, fusel oil (up to 4 years), solids, color, tannins, and pH. The first eight characteristics were determined in accordance with methods described by the Association of Official Agricultural Chemists (A.O.A.C.) ( 1 ) . Proof. A National Bureau of Standards calibrated hydrometer, graduated in 0.2",, was used for indicating proof. Temperature corrections were made with a Bureau of Standards calibrated thermometer. Total Acidity. The method described by the A.O.A.C. was used without modification. The results are expressed as grams of acetic acid per 100 liters at 100 proof. Fixed Acidity. The A.O.A.C. method was used without modification. The results are expressed as grams of acetic acid per 100 liters at 100 proof. Esters. The distillate from a quantitative distillation through all-glass apparatus was saponified by permitting the sample t o stand 24 hours a t room temperature with excess sodium hydroxide. No blank was run. The excess sodium hydroxide was back-titrated with standard sulfuric acid. The results are expressed as grams of ethyl acetate per 100 liters at 100 proof.
935
Aldehydes. The titrimetric method, btwed on a n iodometric reaction as described by the A.O.A.C., was used on the distillate obtained during ester determination. The results are expressed as grams of acetaldehyde per 100 liters at 100 proof. Furfural. The A.O.A.C. method was used without modification. The results are expressed as grams of furfural per 100 liters a t 100 proof. Fusel Oil. The Allen-Marquadt method as described by the A.O.A.C. was used. The carbon tetrachloride extractions and subsequent washings were done mechanically, and all distillations and reflux reactions were performed in all-glass apparatus. The results are expressed as grams of amyl alcohol per 100 liters at 100 proof. Solids. The solids were determined by evaporation and weighing the residue in a tared aluminum dish, as described by the A.O.A.C. The results are expressed as grams per 100 liters a t 100 proof. Color. The color of the sample was determined on a null-type photoelectric colorimeter, designed by the Schenley Laboratories. A 100-watt tungsten filament lamp was used as a source of light which passed through a matched pair of Corning daylight color glass filters before striking the photoelectric cells. The sample was reduced to 100 proof, brought to 25 O C., and placed in a cell which presented 1 inch of path for the passage of light. The instrument was calibrated to read directly in per cent transmission ( T ) based on water as 100%. The resuIts are expressed in terms of color density (d = log l / T ) . The fact t h a t all per cent transmissions, converted to decimal, give values which are less than one, results in negative logarithms for these, and thus a positive value is obtained for all densities. This method of measurement, of course, is only relative, but a relation to the Lovibond method or some other established scale can be established readily. Tannins. This determination is based on the Folin-Denis reagent which reacts specifically with compounds containing a n oxyphenyl bond. This method was refined to conditions of optimum sensitivity (7) and used for tannin determination in whisky. The results are expressed as grams of tannic acid per 100 liters at 100 proof. pH. The p H was determined on samples reduced t o 100 proof a t 25' C. with a glass-electrode electrometer (Coleman Model 3C). The application and limitations of this determination were described in a previous publication (6). The units for expressing all values in this study are grams per 100 liters calculated back t o a n alcoholic concentration of 100 proof (except color, pH, and proof itself). For color and pH, the standard 100 proof was obtained physically by reduction of the sample prior t o determination. All samples were analyzed within a reasonably short time (less than 4 weeks) after withdrawal, so t h a t changes i n glass were reduced to a minimum. Previous investigators had been forced by circumstances t o permit their samples t o be exposed to such influences-for instance, oxidation and the alkalinity of the
TABLE I. HISTORYOF SAMPLES Grain Formula TYPE No. Bourbon 60% corn 40% small grain 84 75% corn 25% small grain 43
Et% 88Vo corn
% ' 18
9
grain 151
32
12% small grain 112
24
' Z e r grains 79
17
R v r. -__
Distillation Type No. Singled 82 Doubled
387
__ 469
%
17
Treatment Type No. Untreated 255
54
Warehouse Type No* Rack (wood) 219
83
Oak ohip-treated
54
12
Concrete
Nnohar-treated
160
34
96 -100
- -_ 469
100
250
_-
469
% 47 53
--
100
Storage Location Louisville, Ky.
No.
%
128
27
Schenley, Pa. Lexington, K y .
114 64
24
91 72
19 16
I
Lawrenceburg, Ind. Frankfort, Ky.
14
-
--
469
100
INDUSTRIAL AND ENGINEERING CHEMISTRY
536
~~
TABLE 11. CHARACTERISTICS Age Yr.?rlon 0 I 3 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96
Proof X b c -
101.8 101.4 101.3 101.4 102.0 102.5 103.1 103,G 104.1 104.7 105.2 105.5 106.0 106.7 107.4 107 9 108.6 108.9 109.3
o
0.48 0.60 0.60 0.61 0.60 0.74 0.95 1.16 1.39 1.49 1.63 1.72 2.12 2.21 2.41 2.70 3.28 3.26 3.00
Total .kcids % u 5.9 20.4 32.2 42.5 53.4 08,l 61.8 64.1 65.8 67.8 69.2 69.7 70.2 72.0 71.6 74.4 76.2 79.4 81.9
3.8 7.; 7.4 7.4 7.5 7.9 8.1 8.5 8.9 8.9 8.8 8.6 9.0 9.5 10.1 11.0 10.2 11.2 10.0
Fixed Acids
-
X
0.8 3.7 5.3 6.6 8.3 9.0 9.2 9.3 9.3 9.4 9.4 9.4 9.5 9.5 9.5
1.5 1.9 2.1 2.0 2.3 2.6 2..8 2.6 2.4 2.7 2.5 2.6 2.5 2.5 2.5 2.5 2.7 2.6 2.0
9.6 9.7 9.7 9.7
2
U
16.7 6.1 5.7 17.2 5.8 18.5 6.1 21.8 26,8 6.7 7.2 31.1 7.9 35.5 9.1 38.9 9.2 41.8 44.7 9.6 47.6 10.6 9.6 48.0 5 1 . 9 11.2 55.6 11.0 57.6 10.8 61.2 14.6 62.0 16.1 64.4 12.4 64.8 17.4
~
U
1.4 2.1 2.8 3.3 4.1 4.8 5.5 5.8 6.0 6.0 6.1 6.1 6.2 6.3 6.5 7.0 7.0 7.0 7.0
0.7 0.9 1.1 1.1 1.3 1.4 1.7 1.7 1.7 1.8 1.9 1.2 1.6 1.3 1.6 1.6 1.5 1.4 1.3
~
AT
FurFnsel furnl__ O i l -
X
'
0.2 1.2 1.5 1.6 1.7 1.8 1.8 1.9 1.8 1.9 1.8 1.7 1.7 1.8 1.8 1.8 1.8 2.0 2.0
0.3 0.5 0.6 0.6 0.7 0.7 0.7 0.7 0.7 0.7 0.7 0.6 0.6 0.7 0.7 0.8 0.7 0.7 0.7
U
X
111 20 123 20 131 22 131 23 132 24 132 21 134 20 136 20 135 20 137 21 138 22 , . .
.., . ,. ... ... , , , , . ,
...
Vol. 41, No. 3
~
AMERICAN&rHISKIES
Aldehydes _____
Esters
u
OF
~~
.,
., .,
..
.. ,,
.,
.
,
VARIOUS -4GEs" -Solids U
8.7 44.1 66.6 87.7 111.1 127.6 137.5 147.7 152.7 157.7 165.9 166.0 173.0 174.2 181.5 186.0 198.6 198.9 209.6
Color uX
7.7 16.7 18.3 17.3 17.3 17.8 18.4 19.1 .19.2 20.3 20.5 19.9 22.3 21.3 21.0 24.4 28.1 23.2 21 . 0
(Dmsity) -
X
u
0.032 0.156 0,208 0.243 0.282 0.308 0.328 0.341 0,382 0.360 0.365 0.867 0.368 0.369 0.380 0.386 0.389 0 413 0.449
0.035 0.044 0.057 0.052 0.047 0.045 0.047 0.049 0,048 0,049 0.050 0,049 0.059 0.053 0,034 0.067 0.063 0.058 0.070
Tannins X u 0.7 12 21 28 35 39 42 44 47 48 48 49 49 49 49 50 50 50 53
3
7 I
7 7 8 7 8 8 8 9 8 10 10 8 12 10 11 13
a
All figures expressed as grams per 100 liters a t 100 proof, except proof (expressed as degrees proof), color (expressed as density), a n d p H .
b
S =
c
u =
-
-
pTT
X
u
4.92 0 . 2 2 4.62 0.14 4.46 0.00 4 38 0 . 0 9 4.38 0.09 4.29 0.08 4.29 0.09 4 . 2 8 0 09 4 27 0 10 4 26 0 09 4 26 0 09 4 26 0 0 9 4 26 0 09 4 26 0 10 4 24 0 0 9 4 24 0 0 9 4 23 0 09 4 22 0 0 9 4 20 0 13
ZX
- = average. -
4-p
n
TAB LE
Age Yr. Mon.
Proof
= s t a n d a r d deviation.
111. CHARACTERISTICS Total Acids
Fixed Acids
Esters
OF
AMERICAN WHISKIES
Aldehydes
Furfural
101.8 5.9 0.8 16.7 1.4 17.2 2.1 3.7 20.4 101.4 5.3 18.5 2.8 101.3 3 32.2 6.6 42.5 21.8 3.3 101.4 6 26.8 4.1 102.0 8.3 53.4 12 31.1 9.0 102,5 18 58.1 5; 35.5 103.1 24 9.2 61.8 5.8 38.9 103.6 9.3 64.1 30 6.0 9.3 41.8 104.1 65.8 36 6.0 44.7 42 104,7 9.4 67.8 6.1 47.6 9.4 69.2 48 105.2 6.1 48.0 105.5 9.4 69.7 54 6.2 9.5 51.9 106.0 60 70.2 6.3 9.5 55.6 106.7 66 72.0 6.5 57.G 71.6 9.5 107.4 72 7.0 61 2 74.4 9.6 107.9 78 7.0 62 . 0 108.6 76.2 9.7 84 64.4 7.0 79.4 9.7 108.9 90 7.0 64.8 81.9 9.7 109.3 96 All figures represent average values a n d a r e expressed pressed as degrees proof), color (expressed a s density), a n d
0 1
Q
0.2 1.2 1.5 1.6 1.7 1.8 1.8 1.9 1.8 1.9 1.8 1.7 1.7 1.8 1.8 1.8 1.8 2.0 2.0
Fusel Oil
Solids
111 123 131 131 132 132 134 136 135 137 138
pH.
STATISTICAL TREATMENT O F DATA
The complexity, variety, and great volume of data necessitated the employment of mechanical means of handling. Punched cards and standard sorting and tabulating machines were utilized. The data mere recorded on master cards and transferred to specially designed punched cards, which facilitated the statistical analysis. The actual statistical treatment consisted of three sections. For the examination of the changes taking place in the whisky characteristics over the period of maturing, the arithmetic mean or average ( X )was obtained for each factor a t every age period. The variation about these averages was described by the computation of the standard deviations ( 4 , 11). The results are shown graphically for each of the characteristics (Figures 1 t o 11). The averages appear as a heavy line, and the scatter of values by light lines on either side. The latter represent the dispersion limits, determined by % . ? * Z r ~ Any barrel chosen a t random will exhibit characteristics lying within this region with a probability of 95% ( 4 , 11). Similar statistical calculations were made for groups identified by type, so that a tabulation was possible of the results for bourbon and rye whisky separately. Beyond this, the Characteristics were tested for corrclation with one another. Rloreover, the effect on the values of certain
AGES"
Color
Tannins
0.032 0.156 0.208 0,243 0.282 0,308 0,328 0.341 0.352 0.360 0.365 0.367 0.368 0.369 0,380 0,385 0.389 0.413 0,449 100 proof,
0.7 4.92 12 4.62 21 4 46 28 4.38 35 4.38 39 4.29 42 4.29 44 4.28 47 4.27 48 4.26 48 4.26 49 4.26 49 4.26 49 4.26 49 4.24 50 4,24 50 4.23 50 4.22 53 4.20 except proof (ex-
(Density)
8.7 44.1 66.6 87.7 111.1 127.6 137.5 147.7 152.7 157.7 165.9 166.0 ... 173.0 174.2 .. , . . 181.5 ,.. 186.0 , . . 198.6 ,.. 198.9 209.6 .,. 100 liters a t a s grams per
glass-with the result t h a t some of the characteristic properties underwent a significant change.
(x)
AT T'ARIOUS
pH
constituents of production factors such as type of whisky, warehouse, and cooperage a n d t r e a t me n t we re considered. CHANGES O F WHISKY CHARACTERISTICS
The tvvo primary statistical functions of the whisky, previously referred to, were calculated a t the various age periods for a number of barrels which differed a t various ages of the whiskies and averaged approximately 350 barrels. These data are given in Table 11. Some of the values for the first 4 years differ from those given for the same period in the previous publication ( 6 ) . The results given in the present study should be considered truer because they are based on a larger volume of data. For ready reference, Table I1 has been abbreviated and condensed into Table 111. Again all figures are evpressed as grams per hundred liters a t 100 proof (except proof and p H ) and rcprcsent the averages for the various ages. Proof. The initial or bonding proof is determined by the distillers and is set customarily a t 102 proof. The variation in proof of samples taken a t age zero is accounted for by the fact that some of the samples were taken before barreling and therefore beforc adjustment to 102 proof (Figure 1). A characteristic of change in proof is a n initial drop which continues to an age of about 3 months. This drop is caused by t h c residual moisture contained in the wood of the barrel. The initial proof is reached again at an age between 6 and 12 months. From this point on, there begins a linear increase, mhich is uniform, and equivalent to about 1.1' of proof per year. The average increase in proof of whisky over 4 years of maturing 1% as found to be 3.4 proof, and over 8 years 7.5'. Total Acidity. A rapid increase in total acidity takes placc during the early stages of maturing. Duiing the first 3 months, the average acid content is increased fivefold. The increase coiltinues a t a high rate to a n age of about 12 months, and then tho rate of inereax becomes lower and gradually assumes the typical asymptotic approach (Figure 2 ) to a maximum value of 81.9
-
INDUSTRIAL AND ENGINEERING CHEMISTRY
March 1949 114
-
L
8
-
80
i
+
A
.*0
3
94 980
100
e
pi
110
12
24
36
48
60
72
84
s%
60
20
0
96
12
r
r 24
36 48 Age, Months
60
72
84
96
-Figure 3. -Fixed Acidity Average;
36 48 Age, Months
60
72
84
96
-Average;
-
12
24
Figure 2. Total Acids -dispersion limit.
Age, Months
Figure 1. Proof Average; -disperaion limits
0
531
dispersion limits
grams per 100 liters observed after 8 years. The average increase over this period is 76 grams per 100 liters. This maximum value for the average total acidity is of con siderable interest, since it apparently represents the maximum amount of acid material which can be extracted from the barrel under normal storage conditions by an alcoholic solution, such as whisky, of 51 to 55% alcohol by volume. The facts are of considerable importance for the examination of the maturing of whiskies which have undergone special treatment before barreling, as reported later in this study. Fixed Acidity. Fixed acids develop along the same characteristic lines as total acids, but the sharp initial rise during the initial storage period is considerably reduced (Figure 3). The entire fixed acidity normally is due to extraction of material from the barrels. It has been suggested (6) t h a t some of this material may undergo degradation processes, thereby adding t o the volatile acid content, but this theory has not been confirmed experimentally. Another investigator (9) believes that some volatile acids are obtained from extraction processes, but does not furnish experimental proof. Undoubtedly, the extracted material undergoes oxidation processes, and is subject to other chemical reactions which may result in additions t o the volatile acid content. The aromatic nature of these constituents are of some importance as contributors to the quality of the final product. Under normal conditions, new whisky has a fixed acidity of zero. Table I1 shows at that age, a content of 0.8 gram (calculated in acetic acid) per 100 liters at 100 proof, and Figure 3 also indicates this level. The apparent discrepadcy is due to the fact that a part of the production was treated with toasted oak chips before barreling, thereby imparting these characteristics to the new whisky. The earlier publication shows the initial fixed acidity a t 2 grams per 100 liters. As previously pointed out, the
20
c 0
12
24
36 48 Age, Months
Figure 4. -Average;
60
72
84
96
Esters
-dispersion limits
earlier results were based on a sample size of 108 barrels, whereas the present work is based on examination of more than four times the size of the sample, or 469 barrels. The process of treating whisky was practiced t o a small extent during a short period immediately following repeal (1933), and only a small proportion of whisky stored during 1937 had been undergoing a treatment (with oak chips) before barreling. All the samples of this kind are part of the 108 barrels reported in the first article (6),and the 361 additional barrels included in the entire study are normal (untreated) whisky. This is reflected obviously in the newer averages, which therefore represent a large preponderance of normal whisky. The average value for fixed acidity a t age 4 years for the large sample size is 9.4 grams per 100 liters, and a t 8 years of age, 9.7 grams per 100 liters. This figure is not only lower than the one previously reported, but also is lower in comparison to figures found by ather investigators. This discrepancy is due t o several factors: First, the whiskies examined before and directly after prohibition came from barrels stored predominantly in old type rack warehouses, where the whisky was exposed, at least during certain periods, t o higher temperatures than in the present majority of concrete warehouses. In these modern structures temperatures can be regulated better and are generally held somewhat lower than in former years. The second influence is the choice and quality of barrels; these are more carefully selected in modern whisky production than previously. Finally, other investigators based their reports in many instances on single barrels, or at least such small numbers t h a t any individual barrel with high acidity might easily have influenced the given result. I n the present study, by the application of the indicated statistical procedure, the variation in results is accounted for and the averages are representative.
INDUSTRIAL AND ENGINEERING CHEMISTRY
538 j
r
12
*r
0
12
24
36 48 Age. M o n t h s
Bigure 5. -Average;
--
200
Vd. 41. No. 3
60
72
84
96
Aldehydes
0 '
dispersion l i m i t s
0
12
24
36 48 Age, Months
Figure 6.
-Average;
-
60
72
84
96
Furfural dispersion limits
'i; 160
rk
-
d
-
3
5* 120 /
4 800
-
E
\
~4011 0
-
I
0
12
I
!
I
24 Age, Months
Figure 7 .
-Average:
--
I
I
J
48
36
Fusel Oil dispersion l i m i t s
r
200
-
c
0
12
24
36 48 Age, Montha
Figure 8.
-Average;
60
72
84
96
Solids
- dispersion limits
Esters. The average increase in ester content over the first 4 years of maturing is 30 grams (calculated as ethyl acetate) per 100 liters and 48.1 grams over 8 years (Figure 4). The increasc in csters with age is probably the most regular and consistent of all the characteristics and therefore may be regarded as the most reliable index for determination of age. An analysis of initial results of this study seemed to indicate that the ester content underwent a drop after barreling, and returned to the starting point only after about 3 months. In view of the otherwise regular and consistent progress of ester furmation with only a gradual lowering in the rate of iiicrease over the entire 8 years, this apparent phenomenon, to which attention was called previously, mas further investigated. On the basis of the larger sampling it became evident that the earlier observation was largely the result of chance occurrence and could hardly be considered in the light of coniplete results as significant enough to
wariant the assumption of an entirely new condition which moreover required the establishment of a specific theory. Aldehydes. The rate of increase in aldehydes is greatest during the first 3 years, and proceeds a t an aveiage for this period of 1.53 granis per 100 liters; the increase for the first year is 2.7 gramb (calculated as acetaldehyde) per 100 liters (Figure 5 ) . The rate of increase aftcr 3 years up to the age of 8 years is a t the average per year of 0.2 giam per 100 liters and is practically constant. In all probability, the actual increase of aldehydes goes on at a greater rate but is partlv offset as a result of evaporation. As the volume of the liquid in the barrel decreases, the evaporation of this highly volatile conqtituent become3 more pronounced The total aldehydes a t the end of 4 years is 6.1, and aftcr 8 years, 7.0 grams pry 100 liters. Both figures are somewhat lower than given in the previous study and also by other investigators. The general tendency in production of u hisky has been ton ard 5 lower content of the usual constituents, and what ir: usually termed a lighter type of whisky. Furfural. N e d y made whisky does not contain furfural. Thc content of this characteristic shown in Table I1 for age zero is due to thc inclusion in the averages of the portion of whisky which haq bren treated with toasted rhips before barreling. The greater proportion of this type of whisky in the first sampling (108 bairels) again accounts for a higher figure a t age zero and othei early ages, in cornparigon with the figures given in this Studv for the larger number (469 barrels) under investigation. Furfural is formed in the process of charring the barrcl, extracts of uncharred oak wood show only traces of furfural. Since practically all the furfural is extracted during the firot few months of storage, the increase after the age of 6 months at which point the average is 1.6 grams per 100 liters is small (Figurc
6). The average fuifural content at age of 4 years is 1.8 grams per 100 liters, and at age of 8 years, 2.0 grams per 100 liters. Fusel Oil. The fusel oil (higher alcohols) constitutes an important component of whisky so far as character and quality are concerned. I t s content is largely determined by the method of distilling used. The high evaporation point and chemiral sluggishness of this constituent are responsible for the fact that the original content a t age zero remains practically unchanged, although a slight loss has been found to occur in most cases. Since the absolute quantity of fusel oil remains practically constant, determinations wrre made in this study only up to 4 years of age (Figure 7 ) . These results have akeady been given in the previous report. They are nevertheless repeated here for the sake of completeness. The apparent increase in fusel oil during maturing shown in the figures is actually only an increase in concentration; the 1 0 s in volume is accounted for by the evaporation of ethyl alcohol. The average increase in fusel oil concentration after 4 years of maturing is 27 grams (calculated 88 amyl alcohol) per 100 liters at 100 proof.
INDUSTRIAL AND ENGINEERING CHEMISTRY
March 1949
539
I
0
12
24
Figure 10. 0
12
24
36 48 Age, Month8
Figure 9.
GO
72
84
96
-Average;
I
I
36 48 GO Age, Months
I
f
72
I
I
84
I
I
Yti
Tannins
-dispersion l i m i t s
Color (Density)
-Average; -dispersion limits Solids. The solid content in whisky is derived from the wood of the barrel by extraction, unless the whisky has been treated with toasted wood chips prior to barreling. Since all whisky barrels are manufactured from a standard grade of white oaks, the extracts are similar in character and impart to the final product 8 definite and authentic quality. Nearly half the solid contents reached over the 8-year maturing periodis acquired during the first 6 months; after that the amount 4 . 0 ~ 1 , 1 , , l l , , l l l l l l , of increase decreases sharply and approaches a uniform rate after the age of 4 years. 3.8 Because of these sharp changes the average increase for 8 years 0 12 24 36 48 GO 72 84 96 is almost meaningless. It is calculated a t 200.9 grams per 100 Age, Months liters, or 25.1 grams per year. The figure of 8.7 grams per 100 Figure 11. pH liters at age zero is ascribed to the fact that a small percentage of Average; -dispersion l i m i t s the whisky had been treated with toasted oak chips. The average increase for the first year was 102.4 grams per -D 100 liters; of this 79.0 was in the first 6 months and 23.4 in 6 the second 6 months (Figure 8). The average increase for & 0 'O the next years, up to the age of 4 years, was 18.2 grams 5! per 100 liters per year, and for the final 4 years, up to age c m * / of 8 years, it was 11.0 grams per 100 liters per year. The highest point reached was 209.6 grams per 100 liters at 100 proof. Color. Similar to solids, color is entirely due t o substances extracted from the charred oak barrel. Newly distilled whisky is colorless, and the treatment of a porA . tion of the whisky before barreling is responsible for the figure 0.032 at age zero. The color of the aging whisky d changes gradually from colorless to light and deep yellow, m" then to amber, and finally to reddish brown (Figure 9). 6 30 J , I I I I I I I Since this development is due to progressive extraction 60 70 80 90 from wood entirely, the development curve and characTotal Acids as G . Acetic Acid/lOO L. at 100 Proof teristics are similar to those of total acids and solids. The Figure 12. Total acids us. Esters rate of increase declines rapidly after the first six months of maturing. Tannins. Wood extractives are solely responsible for the wood extractives-namely, total acidity, solids, tannins, and tannin content of whisky, the same as for solids and color. color, except that its graph appears ~ t sa mirror image of these Therefore, the initial rapid increase is again followed by a sharp (Figure 11). This is due to the definition of pH, which condecrease in rate, although the break occurs somewhat later than cerns the reciprocal of the hydrogen-ion concentration. in the case of solids. Observations during the first 4 years of The initial rapid decrease is succeeded by a sharp drop in the maturing indicated a possibility of a steady linear increase, but rate of decrease after about 6 months, followed by the typical the results of the succeeding 4 years showed the characteristic asymptotic approach after the age of 1 year. decrease in rate and asymptotic approach as for the other The minimum average value after 8 years of maturing is 4.20. characteristics. The value shown a t age zero is again ascribable to The average decreases of p H during 4 years is 0.68 and during 8 the presence of a small quantity of treated whisky. years, 0.72. The average increase in tannin content during the first 4 years of maturing is 47.3 grams (calculated as tannic acid) per 100 liters, MATHEMATICAL REPRESENTATION and for 8 years, 52.3 grams per 100 liters a t 100 proof (Figure 10). I n the previous publication of results observed over the first 4pH. The curve representing changes of pH is similar in year period, the hypothesis was established t h a t three of t h e characteristics to those describing properties derived from
1
-
. .
. .:
..
INDUSTRIAL AND ENGINEERING CHEMISTRY
540
TABLEITi.
hfATHEYATICAL RCPREsESTATION O F h l A T U R I N G CHARACTERISTICS
Total Acids
Esters y =
0.109 +to.olzf + 5.9 obsd. y oalcd. Difference
=
t
y
___-
1.353 A k.O(J53rt obad. y calcd. 16.7 16.7 17.2 17.4 18.9 18.5 21.8 21 .0 26.8 25.2 31.1 29.1 35,s 32.9 36.o 38.9 41.8 40.0 43.3 44.7 46.5 47.6 48.0 49.6 51.9 52,6 5,S.G 55,4 57.6 58.2 61.2 60.8 62.0 63.4 66.8 R4.4 68.2 64.8
t
0 6.3 6.1 4.6 2.2 -0.3 -1.1 -1.9 -2.5 -2 2 -2.2 -2.8 -3 2 -2.5 -3.2 -1 0 0 3 3.1 5 2
y
0 1 3
e
12 18 24 30 36 42 48 54 GO 66 72 78 84 90
96
t
?I 3 fi
li 18
24 30 36 42
4x . .
54 60 66 72 78 84 90 96
gohsd. 8.7 44,I 66.6 87.7 111.1 127,6 137,s 147.7 152.7 157 .7 165.9 166 .O 173.0 174.2 181.5 186.0 198.6 198.9 209.6
ycalcd. 8.7 23.7 48.2 75.3 110.1 131.5 146.0 l56,4 164.3 170.5 175.4 179.5 182.9 185.8 188,3 190.4 192.3 194.0 lYj.5
Aldehvde,
'
8'7 Difference
t
n
0
1
20.4 18.4 12.4 1.0 - 3.9 - 8.5 - 8.7 -11.6 -12.8 - 9.5 -13.5 - 9.9 -11.6 - G.8 - 4.4 6.3 4 9 14.1
Z
6 12 18 24 30 36 42 4x 54 60 66 72 78 84 90 96 ~~
2
18.717 2.483t yobsd. goaled.
0,023 -0.156 0.205 0.243 0.282 0.308 0.328 0,341 0,382 0.360 0.36,5 0.367 0.368 0.369 0.380 0,385 0.389 0.413 0.449
0.032 0.079 0,147 0.211 0.279 0.316 0.339 0.354 0,365 0 , 373 0.380 0,388 0.390 0,393 0.397 0.399 0.402 0.404 0.403
t
y obsd.
-6.2 -0.4 0.8 1.6 2.0 2.6 2.4 1.8 1 4 1 1 -1.6 -0.7 t0.2 -0.6 -0.4 -1.4 -1.4 -3 4
0 I 3 6 12 18 24 30 36 42 48 54 GO 66 72 78 84 90 96
1.4 2.1 2.8 3.3 4.1
0'032 Difference
0 -0.077 0.058 0.032 0.003 -0.008 -0.011 -0,013 -0,013 -0,013 -0,015 -0,018 -0,022 -0.024 -0.017 -0,014 -0,013 0.009 0 044
whisky characteristics-namel~r, total acidity, solids, and color-follow a hyperbolic form. The extension of this investigation not only confirmed the propiiety of the use of this equation type for these three characteristics but indicated that the same form is applicable also to most of the other characteriqtics. As already indicated in the first and partial publication of the results of this investigation, the development of most of the whisky constituents (the majority of which are customarily referred to as congenerics) can be represented by the following hyperbolic formula : Y =
t
+ yo
where t = age in months; y = valuc of characterintic; yo = value of characteristic a t t = a; and a, b = constants. The values of a and b can be found readily since t / y - yo is a linear yo = a bt. function of t-namely, t / y The method of least squares (11)was used in determining a and
-
y =
+ "" Difforenoe
Color
Solids = 0 . 0 6 1 ~+t0.00471t +
Vol. 41, No. 3
+
b.
Table IV shows the close agreement between values calculated from these equations and those determined by actual chemical analyses (observed values). CORRELATIOV O F WHISKY CHARACTERISTICS
The importance of the correlation between the characteristic constitucnts has been indicated by some of the previous investigators, especially Crampton arid Tolnian (2). The technical and statistical data available to them were probably insufficient to permit the establishment of a niathcmatical interdependence. The relationship, as it was found to exist, is undoubtedly of considerable practical importance. In the manufacture of the Lvhisky, it will be possible to influence a given factor, which is important for final quality but difficult to control during production by the manipulation of one of the closely related factors
2.104 :O.l61t g calod.
::: 5 ,8
6.0 6.0 6 1 6.1 6.2 6.3 6.5 7 9 7.0 7.0 7.0
1.4 1 8 2 6 3.4 4.4 5.0
5.4
5.7 6.0 6.1 6.3 6.4 6 5 6.6 6.7 (1.7 8.8 6.8 6.9
1' urfiii al y =
+
Difference
0
nR
02 -0 1 -0 3 -0 2 -0.1 0 1
0
-0.1 -0.2 -0.3 -0.3 -0.3 -0.2 0.3 0.2 0.2 0.1
Tannins t Y = _____ 0.125 0.0178t + O" t yobad. ycaicd. Difference n 0 0.7 0 7 4 1 12 8 18 3 21 fi 28 L 27 12 -1 36 35 -2 18 41 39 -2 42 24 44 -2 46 44 30 -1 36 47 48 -1 48 42 49 -2 48 48 50 -1 49 54 50 -2 49 51 60 -2 49 66 51 5 2 -3 49 72 -2 50 52 78 -2 50 52 84 -3 90 50 53 53 53 0 96
t
n
0.2 1.2
1 3 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96
1.5 1.6
1 7 1.8 1.8 1.9 1.8
1.9 1.8 1.7 1.7 1.8 1.8 1.8 1.8 2.0 2 .O
0.577s I/
ciilcl.
0.2 0,s 1.2 1 ..5 1.7 1.8 1. 8
1.8 1.8 1.9 1. 9 1.9 1. 9 1 9 1 9 1.9 1.9 1.9 1. 9
O" Differonce 0 0 4 0 3 0 1 0 0
0 0 1 0 0 -0 1 -0 2 -0 2 -0 1 -0 1 -0 1 -0 1 1 0 1 0
PH
+
3
1.250
g obsd.
Y
=
-
t
y0bs-L.
0 1 3 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 90
4.92 4.62 4.46 4.38 4.38 4.29 4 29 4.28 4.27 4.26 4 26 4.26 4.26 4.26 4.24 4.24 4.23 4.22 4.20
4.9072-f;3x i4.92 7jc~Ic:l. 4.92 4.76 4.59 4.46 4.35 4.31 4.28 4.26 4.25 4 24 4.23 4.23 4.22 4.22 4.22 4.21 4.21 4.21 4.21
Diffcrence 0 -0 1 4 -0.13
-0.0'3 0.03 -0.02 0.01 0.02 0.02 0.02 0.03 0.03 0.04 0.04 0.02 0.03 0.02 0.01 -0.01
TTliich can be changed readily. The anal for rapid tests or quality control by examination of only one of the pairs of closely related constituents. Finally, the existcncc of close relation between certain characterist,ics in normally matured whisky will make it possible in some instances to dctcct artificial manipulation and falsification. Crampton and Tolman place great importance in this application, believing it of great value in the determination of the adulteration of commercial whiskies. The findings presented here confirm this possibility but, caution should be exercised especially where conclusions arc inado concerning the age of the investigated product based solely on the relation bet\yeen the various constituents. For the purpose of determining relations between values of the characteristics, scatter diagrams ( 1 1 ) ryere first set up for all possible pairs of congenerics and these actually indicated linearity of relations. One of these scatter diagrams, shoiving the rclalion of total acids and esters, is shown in Figure 12. As a consequence, the correlation coefficients were calculai ed. (Statistically, if the possible relation between tivo variablw can be assumed linear, the correlation coefficient, r , d l determine the amount of such relation which exists. The value of P will lic between 1 and +1. The value r = +1 indicates perfect direct; relation, r = - 1, perfect inverse relation.) I t was found that proof and aldehydes are two independent variables. I n this connection, it should t)c noted that in the early article by Crampton and Tolnian (a), the stalcmenl is made that changes in proof and volume are important in thcir bearing on the changes taking place in substanccs contained in the whisky. Such a relation viould have been difficult to undcrstand from the purely chemical viewpoint. The authors' results concerning correlatious of proof with constituent substances indicate no such interdependence. Proof appears t o have no bearing on changes in the various congenerics.
-
INDUSTRIAL AND ENGINEERING CHEMISTRY
March 1949
541
TABLE V. COMPARISON BETWEEN CHANGES IN CHARACTERISTICS OF UNTREATED AND CHIP-TREATED AMERICANWHISKY Age. Months
Proof
Xlb
Total
Acids x2c _~ zl z2
0 102.0 101.6 3.4 5.9 1 23.1 25.3 101.4 101.0 3 8 . 1 37.6 3 101 I 5 101 . o 6 102.0 1 0 1 . 1 49.4 50.3 21 59.0 62.7 102.8 102.6 103.6 1 0 3 . 3 18 65.6 64.5 24 69.2 6 8 . 2 104.2 104.6 7 4 . 2 70.2 30 1 0 4 . 8 104.6 36 73.4 7 7 . 8 105.5 105.2 42 77.5 74.7 106.0 1 0 5 . 9 80.7 79.4 48 106.6 106.4 a Color in per cent transmission. = Untreated. C 8% = Chip-treated.
TABLEVI. Proof
Xlh
-iicids X Z c X1 22 -
5:5
4:4 7.5 8.8 1 1 . 8 12.9 14.6 12.6 12.6 13.1 13.4 1 3 . 8 12.6 14.7 14.3 14.1 14.0 14.4
8.0 10.0
Esters
El
17.1 16.4 18.8 22.0 26.8 34.0 37.2 43.1 47.8 48.9 50.6
hydes Alde-
xz
17.0 16.5 19.3 23.8 31.3 35.2 38.4 42.6 46.0 50.2 51.9
fural Fur-
Colors
Solids
21
3 2
x i
2 2
0.8 1.6 2.0 3.0 3.6 4.9 5.9 7.3 6.8 7.2 4.8
1.0 2.2 2.6 2.4 3.4 4.4 5.1 6.2 7.3 6.7 5.5
1.9 2.0 2.4 2.6 2.8 2.7 2.6 2.9 2.6 2.9
0.0
0.0
1.3 1.8 2.0 2.2 2.3 2.2 2.4 2.6 2.8 2.6
xi
2 1
;i.2
3.0 18.6 49.3 80.1 8 6 . 2 101.3 116.3 133.0 143.7 163.0 162.0 1 8 0 , s 173.8 1 8 8 . 1 184.7 202.8 193.4 203.3 197.7 2 1 1 . 3 214.0 218.9
99.2 65.9 55.0 50.0 45.8 41.9 40.0 37.3 36.8 35.9 34.9
Tannins
-X2
xl
72.4 56.4 48.0 43.1 40.1 37.8 36.6 35.5 33.9 33.0 31.4
16 28 37 43 44 51 54 54 69 68
0
PH x1 x2
2 2
0
26 34 45 53 50 55 60 61 63 72
4.77 4.52 4.42 4.34 4.21 4.23 4.20
4.80 4.53 4.41 4.35 4.28 4.22 4.22
4 . 2 90 4.16 4.15
4.22 4.19 4.12
BETWEEN CHAKGES IN CHARACTERISTICS OF BOURBON AND RYE AMERICAN WHISKIES COMPARISON
Total
Age, hlonths
Fixed
Acids zl x2
Fixed
Alde-
Fur-
Fusel
Color5 hydes fural Oil -E x t r a c t _XlAcids-X z - XiEstersxz ~~21 2 s xi 22 2 1 2 2 XI 2 2 2 1 3
1 3 6 12 18 24 30 36 42 48
100.4 100.9 1 0 0 . 5 100.6 100.8 100.9 101.7 1 0 2 , l 101.9 1 0 2 . 3 102.9 1 0 3 . 1 1 0 3 . 1 103 5 1 0 3 . 9 104.1 106.4 1 0 5 . 5 105.4 1 0 5 , 7 BOd 1 0 7 . 3 106.4 84d 1 0 5 . 2 104.2 a Color in per cent transmission. b 21= Bourbon. c X p = Rye. d Results for ages 60 and 84 months are based on only p a r t of the original 24 barrels.
Characteristics which were found to be highly correlated were: total acids and esters; solids and tannins; solids and color; and tannins and color. Relation was also established between total acids and solids, total acids and tannins, and esters and solids. SPECIAL TREATMENT OF WHISKY
Attempts have been made from time to time to alter or influence the natural maturing process of whisky by the application of various methods of treatment either before or shortly after barreling. The purpose of this procedure is to impart to the newly made product, within a short time, some of the important characteristics which are normally acquired through the extended aging process. The practice of so-called quick-aging was of some importance during the period immediately following the prohibition period, which brought about an almost complete exhaustion of aged whisky. The special treatments then applied have since been com; pletely abandoned by practically all distillers. The effect of treatment was a t one time of considerable importance and is of general interest. The treatment can be applied to the newly made product by means of percolation over toasted white oak chips before barreling, or by heating the newly barreled product shortly after barreling either with or without the addition of toasted chips. For this study, comparisons were made of a group of seven barrels of whisky which had been treated with toasted oak chips before barreling and eight barrels of the same whisky not so treated. All fifteen barrels were of the same manufacture, and stored under identical conditions. Some smaller groups of other whisky in different distilleries and warehouses were tested similarly, but because the results were largely confirmatory, only data obtained from the first and largest group are here reported (Table V). Chemical analyses were made for samples drawn at regular intervals from age zero to 4 years. The values in the table
56.6 122 119 84.5 122 125 119 124 1 0 0 . 8 123 122 122.6 120 135 1 4 0 . 4 119 131 147.8 132 130 1 6 1 . 6 137 135 1 6 7 . 8 136 134 1 6 8 . 5 135 137 1 8 2 . 9 . . . 205.7 . . . 230.2
58.4 85.9 105.0 123.1 137.6 147.6 160.8 168.8 173.4 186.2 211.7 225.0
Tannins 2
61.9 59.9 54.2 51.7 5 0 . 5 48.6 4 8 . 3 47.9 46 2 45.0 44.4 43.6 42.5 41.8 40.2 39.5 4 0 . 2 38.4 37.3 37.5 37.2 36.7 3 6 . 8 37.9
XI XZ
14 29 35 39 41 42 45 48 47 59 55 61
17 30 39 36 42 41 47 49 50 59 51 62
PH
-
x1 4.60 4.44 4.37 4.30 4.24 4.30 4.30 4.30 4.28 4.19 4.22 4.19
2 2
4.43 4.47 4.41 4.30 4.26 4.32 4.30 4.36 4.31 4.20 4.15 4.21
represent the average of the untreated and treated members of the group. Statistical analysis of results for treated and untreated samples showed no significant difference (4) between values for proof, aldehydes, esters, furfural, and pH. No significant difference is shown in the ester values throughout the aging period-that is, the esters for the treated whisky continue to increase a t the same rate as for the nontreated. The development of the esters, which contribute to the characteristic aroma of whisky, thus is not stunted in any way by the addition of chips. On the other hand, a significant difference, shown for color, solids, total acids, and tannins, disappears after the 3-year period. Total acids show no significant difference as early as after the first month. The results for tannins are all higher for the chip-treated whisky, but not significantly so after 12 months. Therefore, any contention that the tannins in chiptreated whisky accumulate abnormally as the aging continues is disproved. Similarly, the solids content is higher for chiptreated whisky, but no significant difference is shown beyond 18 months. Such differences as are shown after 18 months of age are attributed to sampling fluctuations. The values for the chip-treated whisky approach those for the nontreated by the thirty-sixth month of aging, so that no excess of tannins or solids occurs which would militate against the taste. For color, which is reported in per cent transmission (transparency), all values for chip-treated whisky were lower (color more intense) but after the 3-year period the significance of the difference is doubtful. Thus, while the addition of chips intensifies the color of the whisky a t the start, the ultimate results are products which are alike with respect to color whether with or without chip-treatment. The coloring of the chip-treated whisky occurs in advance of normal aging. At first its depth is 1 month ahead of time. At 1 month, it is equivalent to that a t 3 months for normally aging whisky. Beyond that, the color is 6 months ahead up to 36 months where an asymptotic approach to a common value begins in both cases.
INDUSTRIAL AND ENGINEERING CHEMISTRY
542
TABLE 1'11. Age, Months
COMPARISON BETWEEK -
Proof
XI= 0 1 3
101.3 101 . 0 101.1 fi 101.3 102.4 I2 18 103.2 104.3 24 105.1 30 105.9 36 107 .O 42 107.7 48 Color in per cent b XI = Rack. 2 X, = Concrete.
X?b
Total Acids
AYui
8.4 101.3 101 . o 28.0 100.9 37 6 48 7 101 .o 59 2 101.7 64 1 101.7 1 0 2 . 4 67 . 0 102.6 71.6 73.2 103.0 75.6 103.7 104.1 76.7 transmission.
.T? 6.7 23.1 32.7 44.2 57 . 3 64.0 $17.4 12.1 74.7 75.9 7g .0
CH.4KQEs I S
CHARACTERISTICS
Fixed Acids
-
XI
l e
4:3 5.9 8.4 10.2 10.8 10.9 11.9 11.5 12.9 14.4
4'8
5 5 8.6
10.5 11.3 10.7 11.7 12.3 14.1 14.3
OF
Esters
21
xz
XI
E2
22.6 21.8 21.5 24.7 31.1 36.7 39.7 42.8 46.3 61.6 51.4
21.7 22.6 20.3 23.7 28.3 33.9 38.4 42.9 44.6 49.4 49.2
1.3 2.0 3.1 3.3 4.0 5.1 5.9 7.0 7.7 7.6 9.1
1.9 2.6 2.9 3.6 4.3 5.2 6.8 6.9
As it has been shown already that some of the characteristics are closely related to those which can be controlled by means of chip treatment, this procedure would seem to offer a tool for influencing other characteristics which are considerrd of importance to the quality of the final product. COMPARISON OF TYPES
The results reported for this entire study were carried out preponderantly with bourbon whisky. The two main types of whisky, bourbon and rye, do not show significant differences in their constituents which can be ascribed to the type of grain predominantly used in the mash. Other factors in the distilling operation are of greater influence on the constituents than the t,ypes and proportions of original grain. -is a rule, the difference in the proportion of grains used for the various types is not a8 great under modern practice as before prohibition, but even in cases of ext.reme uses of one grain or another (corn or rye), chemical analysis cannot find the difference p-hich organoleptic exaininat,ion (by taste and smell 1 is able to detect. The differences in characteristics pointed out by other investigators are chiefly due to production proccdures, especially distillation, and maturing conditions. Crampton and Tohnan (W) stat,e in their conclusions that rye whiskies show a higher content in solids, acids, esters, etc., than do bourbon whiskies, hut t,his is explained by the fact that heated warehouses are almost universally used for maturing of rye whiskies, and unheated warehouses for the maturing of bourbon xhiskies. The conditions referred to by these investigators have long since changed materially. All factors important, for maturing are carefully controlled in modern warehouses, and consequently the differences referred to in the above quotation are no longer encountered. The results of such studies made in this investigation as concern differences in type are entirely consistent in this respect, The comparisons made were under identical conditions as to distilling, treatment, eooperagc, and location; 24 barrels equally distributed between bourbon and rye were observed. These barrels were divided for the two types of whisky into groups, homogeneous a.s to distillery, treatment, source of cooperage, and warehouse construction. The results of this comparison are given in Table V1. EFFECT OF WAREHOUSE CONSTRUCTIOR
To test t,he possible effect of different warehouse construet,ions on the maturing whipky, samples were examined from barrels of whisky homogeneous with respect to all other factors. The figures in Table VI1 represent average values obt,ained from 50 barrels. These barrels were divided, for the two types of warehouses-concrete and rack (wood)--into groups homogeneous as t o type of whisky, treatment, cooperage, and dist'il-
AWX=~ICAK m H I S K Y IK RACKAIVD Furiura1
Aldehydes
8.1
9.3 9.8
Vol. 41, No. 3
ZI
Solids X l
0 0.3 1.2 1.6 1.6 1.7 1.6 1.8 1.7 2 . 0 1.8 2 . 1 2.0 2 . 3 2.1 2.3 2.2 2.3 2.2 2.8 2.2 2.4
til 7.6 .io. 0 74.9 96.6 122.0 135.6 153.8 164.4 163.2 174.4 182.3
-
X Y 8.2 43.3 66.3 87.8 119.2 133.7 153.0 165.2 172.9 176.9 186.3
Color'
.-
xi
96.2 68.0 59.9 54.6 50.4 47.4 43.9 42.5 40.9 40.3 38 7
CONCRETE WAREHOUSES
-
Tannins
x2
3 1
92.9 68.1 61.2 56,l 50 . 0 46.3 43.2 41.6 39.7 39.3 38.1
0 13 24 29 36 37 42 46 48 53
58
27% 1.5 1R 23 28 36 40 44 46 50 56 60
-
nH
.-
x1
X?
4.70 4.53 4.43 4.33 4.2R 4.29 4.28 4.28 4.27 4.18 4.22
4.68 4.54 4.43 4.37 4.25 4.25 4.24 4.25 4.21 4.16 4.15
lery. KO significant differences in the characteristics could be found. The only exceptions were the values for proof which were higher for barrels in rack warehouses than for corresponding members of the group in concrete warehouses. This difference can be ascribed t o higher temperatures which prevail at least in the upper sections of t,hc rack warehouses. The absence of a decided difference betm-een the two types of warehouses can be ascribed to the fact that both types were of relatively modern construction and temperature cont,rolled as far as heating and ventilation permitted.
The effect of cooperage as studied in two distilleries whcre it was possible to set aside homogeneous groups differing only as to the source of cooperage. Conditions made it, difficult to establish corresponding groups of equal size. In one instance the sample sizes were t\To and four, and in the other, two arid eight. Consequently, the significance of the results should be judged accordingly. In dist,illery I, the resulbs from manufacturers MI and M 4 showed littlc difference. This is not surprising, as all barrels were ordered to specifications and were accepted only after close inspection. Nevertheless, the proof in !MI barrels was always higher, indicating great,er permeability for water vapors, arid in the same barrels, higher values appeared for furfural, color, and PH. In distillery 11, barrels from manufacturers M , and M4 showed only slightly different results. Higher valups appeared COILsistently for barrels from $13 for total acids, esters, solids, tannins, and color. Obviously, differences cither in qualities of the ;\ood or degree of charring or both were responsihlr. SUMMARY
systematic study has been made on a representative sample (469 barrels) of the changes occurring in American whisky while maturing. The changes in American whisky while maturing over a period of 8 years arc given in a table of eleven Characteristics. Most of the important characteristics of whisky undergo changes due t o or influenced by Substances extracted from the barrel. The extraction of these substances is greatest at, the beginning of t,he storage period and decreases a t a gradually diminishing rate. These changes are mathematically expressed by a hyperbolic equation representative of the asymptotic approach to a maximum value. .4 close correlation has been shown between several characteristics, especially between total acidity and esters. The choice of grains has a relatively small influence on the chemical composition of whisky. Changes while maturing in different, types of whisky (bourbon and rye) are similar. ;Z
'
March 1949
INDUSTRIAL AND ENGINEERING CHEMISTRY ACKNOWLEDGMENT
The authors wish to acknowledge the assistance of Jerome Polatnick in directing and supervising the great number of analytical determinations necessary for this investigation. The authors are grateful also t o the plant managers and chemists a t the various distilleries for setting aside and sampling the experimental barrels. Acknowledgment also is made to M. Rosenblatt, the coauthor of the first section of this study, for his part in organizing and initiating the entire work.
543
(3) Ibid., 122-3, Table XIVa. (4) Croxton, F. E., and Cowden, D. J., “Applied General Statistics,” pp. 286-7, New York, Prentice-Hall, Inc., 1940. (5) Liebmann, A. J., and Rosenblatt, M., IND.ENG.CHEM.,35, 994 (1943).
(8) Liebmann, A. J., and Rosenblatt, M., J. Assoc. O f l . Agr.
Chemists, 25, 183 (1942). (7) Rosenblatt, M., and Peluso, J. V., Ibid., 24, 170 (1941). (5) U. S. Treasury Dept., Federal Alcohol Administration, Regulation No. 6 , Article 11, Sect. 21 (b) (March 1, 1939). (9) Valaer, Peter, and Frazier, W. H., IND.ENQ.CHEM.,28, 92 (1936).
LITERATURE CITED
Assoc. Official Agr. Chem., Official and Tentative Methods of Analysis, 4th and 6th ed., Sect. XVl (1940). (2) Crampton, C. A., and Tolman, L. M., J. Am. Chem. S o t , 30,97
(1)
(1908)
(10) Ibid., 105, Table XIV. (11) Yule, G. U., and Kendall, M. G., “Introduction to Theory of
Statistics,” p. 386, London, G. Griffin, 1937.
R m m v R f ~December 18, 1947.
Thermal Properties of GR-S Latices F. W. DITTMAN’ AND
c. c. WINDING
Cornel1 University, Ithaca, N . Y . Methoas were devised and apparatus built to measure direct the heat capacity and thermal conductivity of synthetic latex under pressure, vented and stripped. Samples were obtained from pilot plant reactors a t various degrees of completion of the butadiene-styrene copolymerization reaction. Heat capacity of unvented samples increases moderately with temperature and decreases sharply with conversion as the reaction proceeds to completion; most of the decrease occurs between 40 and 50% conversion. Heat capacity increases slightly on venting because of loss of butadiene. All values lie between 0.75
and 0.91 cal. per gram, O C. Heat capacity of stripped latex is constant a t 0.837 * 0.005 cal. per gram, O C. between 8” and 60” C. Thermal conductivity of latex under pressure shows a slight increase with temperature and a marked increase with per cent conversion. Values ranged from 0.00061 to 0.00099 gram-cal. per second cm., O C. Thermal conductivity could not be measured on samples below 50% conversion because of the existence of two liquid phases. Thermal conductivity of stripped latex ranged from 0.00095 to 0.00105 gram-cal. per second cm., O C., increasing slightly with temperature.
T
enough to be shaken thoroughly by hand before withdrawing portion for determination of thermal properties.
H E rapid development of the synthetic rubber program required many estimations and approximations of fundamental data to complete the design and construction of synthetic rubber plants. Concurrently, research programs were started to obtain actual experimental data to facilitate the operation of existing plants and t o provide information on which to base further development work. As a part of this general program, experimental apparatus and techniques were developed to obtain the heat capacity and thermal conductivity of latices formed by the copolymerization of butadiene and styrene, Experimental work was confined to the latices obtained in various stages of the production of standard GR-S rubber. Samples of latices were obtained direct from pilot plant polymerization kettles at various stages of conversion, These samples were kept under pressure and all contained excess butadiene and styrene. They are referred to as unvented latices. Vented latices were obtained from this material by allowing the excess butadiene to flash off at atmospheric pressure. The removal of the excess styrene by steam stripping produced stripped latex. The composition of each of the above types of samples is best specified by per cent conversion which is a direct function of the solids (copolymer) content of the samples as measured in the laboratory Samples were drained direct from the reactors into bombs containing a small quantity of inhibitor The bombs were small 1
Present address. Koppers Company. Inc.. Pittsburgh. Pa.
B
HEAT CAPACITY
Apparatus. The technique of accurate calorimetry has been well standardized and described a t length in the literature. Standard works on physical measurements, such as those of Drucker ( 7 )and White ( 2 4 ) were referred t o in detail during the process of designing and operating the present calorimeter. The usual liquid calorimeter consists of an enclosed sample chamber with stirrer, heating element, and thermometer of the desired degree of accuracy, surrounded by an air gap within a thermostatically controlled water bath. Such equipment had to be modified in order to mmsure the heat capacity of stripped, vented, and unvented latices. The unvented samples containing butadiene had to be kept under pressures ranging up to 75 pounds per square inch and totally enclosed a t all times to prevent loss of butadiene. All latices also deposited a thin film of rubber on the walls which had l o be easily removable. The a p aratus finally adopted consisted essentially of two cylindrical ciambers cross-connected at top and bottom and surrounded by cotton-wool insulation. The entire assembly was placed in a large container which was fitted with a circular wooden lid. A Beckman thermometer, agitator shaft, and heating element passed through openings in the lid into the circulating liquid in the inner chambers. Oil was used to submerge a pressure-tight bomb containing unvented samples, but
