CHANGES IN SOME VOLATILE CONSTITUENTS OF BRANDY DURING AGING

Transcription

1 CHANGES IN SOME VOLATILE CONSTITUENTS OF BRANDY DURING AGING Masami Onishi, J. F. Guymon, and E. A. Crowell Respectively Assistant Specialist, Professor of Enology, and Staff Research Associate, Department of Viticulture and Enology, University of California, Davis, California Mr. Onishi is an Associate Research Chemist on leave from the Central Research Institute, Suntory Limited, Osaka, Japan. Financial assistance for this research from Suntory, Ltd. is gratefully acknowledged. Accepted for publication March 25, ABSTRACT Several kinds of experimental brandies, both unaged and aged in various types of U.S. and French oak barrels, were vacuum distilled, the components in the distillate were extracted with ether-pentane (2:1), and concentrated to small volume by removal of the solvent for gas chromatographic analyses with a Carbowax 20M column. Quantitative levels were estimated by comparing peak areas with those of an internal standard, 1-octanol. Acetate esters of isoamyl, n-hexyl, and fl-phenethyl alcohols decreased during aging in oak barrels. The ethyl esters of the fatty acids, caproic, caprylic and capric, increased during aging while ethyl laurate changed little or slightly decreased. Compounds derived mostly or entirely from oak, furfural, 5-methyl furfural, diethyl succinate and the cis and trans isomers of fl-methyl- 7-octalactone (oak lactones -a and -b), were more abundant in brandies aged in U.S. oak than in French oak and lesser amounts of oak-derived compounds were obtained in brandies from reused barrels. The flavor and aroma constituents of aged brandy are derived from each successive stage of the production process. Most chemical compounds thus far identified in brandy are either present in the grape raw material or, more importantly, are produced by yeast fermentation of grape sugar (14). The distillation process ~ the type of still and the operating conditions used~ not only concentrates the ethyl alcohol in the wine into the brandy distillate but also determines to a large extent the composition of volatile trace congeners that distill into the product (3). Also, new compounds not present in distilling wines, such as furfural, may be produced by heat during distillation and, though difficult to verify, esterification reactions apparently take place on the plates of continuous stills producing additional amounts of neutral esters (6). Aging of brandy distillates in oak barrels produces additional compositional changes. Both alcohol and water are lost as a result of diffusion through the oak stave and barrel head and subsequent evaporation into the atmosphere. The rates of diffusion of alcohol and water depend upon partial pressure differences across the diffusion barrier, 152 the oak stave, and the resistance of the oak stave to diffusion. Partialpressures are temperature-dependent, and the resistance to diffusion is a function of the path length of diffusing molecules, i.e. the stave thickness. The partial pressure difference of water across the stave depends upon the relative humidity of the atmosphere but that of alcohol does not since the vapor pressure of alcohol in the atmosphere is negligible. So the rates of losses by diffusion-evaporation depend on temperature, container size (surface to volume ratio), stave thickness and relative humidity as well as molecular size. Ethyl alcohol has a molecular weight about 2.5 times that of water so alcohol diffuses more slowly than water except under high humidity conditions. Accordingly the alcohol content of the spirit in the barrel increases during aging under low to moderate relative humidity but decreases under high humidity conditions. Compounds of comparatively high molecular weight, less permeable than alcohol or water, such as the fusel alcohols, are somewhat concentrated by aging (16). Aromatic aldehydes (5)and acids (7), sugars (1), tannin, and other chemicals are either extracted from oak or are produced by the action

2 of alcohol on the macrochemical structure of oak lignin (5). The porous oak barrel also permits cerrain oxidative reactions to occur, yielding some compounds which, in turn, affect the equilibrium composition of the brandy. Oxidative reactions almost certainly take place, but it is difficult to prove to what extent they occur. Reazin et al. (12) added a small amount of radioactive ethyl alcohol (1-~4C) to the bourbon high wine filled into a new 50-gal charred oak barrel and analyzed some of the components for radioactivity in periodic samples taken over 56 months of aging. Carbon labeling found in acetaldehyde, acetic acid and ethyl acetate indicated that ethanol is indeed oxidized by molecular oxygen to produce acetic acid via acetaldehyde which is then partially converted to the ester during barrel aging. Although several uncertainties remain, their work appears to be the best experimental evidence to date that these quite logical reactions in fact do occur. Their results, however, do not negate the fact that much, if not most, of the increase in acetic acid during aging is due to extraction of this and other acids from the barrel itself. Recent gas chromatographic/mass spectroscopic studies of various kinds of distilled spirits allow tabulation of a vast array of constituents, many of which were not previously known to be present (9,10,13). The very sensitive separation and identification techniques used for these studies yield a great deal of information but reveal little about changes that occur during aging. This report concerns a study of certain volatile esters and some other compounds in brandies of known origin and treatment and the changes in their concentrations that took place while in the oak barrel. MATERIALS AND METHODS Test samples: Brandies used for this study were produced using either a 26-plate, 12 in.-diameter continuous column or a 350-gal all-copper pot still by the Charente method of distillation ~ two successive simple distillations. Both pilot-scale experimental stills are operated at the University of California, Davis. Wine distillates after reduction with distilled water were filled into oak barrels for aging and a sample of the fill material was reserved in tightly-stoppered glass bottles for the same period. Sample extraction and concentration" Five hundred ml of brandy plus 100 ml distilled water were distilled at 40 C using a Buchler flash evaporator, model PTFE-1 GN. The vacuum distillate was collected from an ice water-cooled condenser until only~ about 30 ml of residue remained in the distilling flask. An additional 100 ml water were added to the residue and the evaporation continued until the residue was again reduced to 30 ml. Both the aged brandies and their unaged reference materials were distilled by exactly the same procedure. One ml of an internal standard solution con- VOLATILE CONSTITUENTS OF BRANDY--153 sisting of 415 mg l-octanol/100 ml 95 % ethanol was added, then the distillates were extracted five times with successive 100 ml aliquots of an ether-pentane mixture (2 + 1). Both extracting solvents were carefully purified by redistillation before use. The combined solvent layers were washed three times with 80 ml portions of water, dried with anhydrous MgSO~ and concentrated to 1 ml by distilling off the solvent using a Vigreaux column. Gas chromatography: A Hewlett-Packard 5711A instrument was used for all GC separations. The GC column contained 10% w/w Carbowax 20M plus 0.2% w/w Shell Ionox antioxidant coated on 80/100 mesh Chromasorb WAW, packed into 2.44 m x 3.2 mm (8 ft x V8 in) OD thin-walled stainless steel tubing. After 1 ~1 injections the column oven temperature was maintained at 60 C for four min, then programmed at 8 C/min to 210 C and held until all peaks were eluted. Detector and injector temperatures were 250 C and N~ carrier gas, hydrogen and air flows were, respectively, 30, 30 and 300 ml/min. Peak areas were calculated manually by multiplying peak height by width at half height for component peaks and for the internal standard. Areas were converted to concentration units by referring to the internal standard concentration. Component concentrations were not compensated for extraction recovery nor for detector response relative to 1- octanol. RESULTS AND DISCUSSION The grape varieties used for the wines distilled, the distillation method and the parameters of maturation in oak barrels are indicated in Table 1 for six groups of brandies. In each group the firstlisted is a sample of the brandy distillate reduced to barrel proof but stored in glass while the remaining samples of a given group represent different conditions of oak-barrel aging using the same fill. The first group of samples (I) consists of the unaged and two aged samples of 109 proof distillate produced at 139 proof in the pot still from wine made from a mixture of Folle blanche and French Colombard grapes. The brandy was aged for six years in a new, plain U.S. oak barrel (~211) and in a new 65-gal Limousin barrel (t212). During the aging period the proof increased from 109 to, respectively, 112 and 114. Group II represents a similar experiment but the distillate was produced at 166 proof in the 12-inch continuous column. Group III includes the control and six samples aged four years in oak, four of these being new or plain oak barrels of the same capacity but each a different type of wood (U.S. white oak, French Limousin, French Troncais, and French Gascony or Armagnac oak). Group IV includes the control and a continuous still brandy aged in a new charred U.S. oak barrel. In Group V, sample 17 (~277) represents the first reuse of the same Troncais oak barrel initially used

3 154--VOLATILE CONSTITUENTS OF BRANDY Table 1. Description of brandies used to study composition changes during aging. Group Sample Barrel Grape Still Proof Bar. Size Age Proof no. no. var. type distn, type a (gals) (yrs) (final) 1 Unaged Folle blanche Pot plus P Fr. Colombard L Unaged Chenin blanc C.S P L t13 7 Unaged Folle blanche C.S plus P St. Emilion L T G L R IV 14 Unaged Sauvignon C.S blanc C Unaged St. Emilion C.S T(R) VI 18 U n aged Peve rel I a C.S L L R L(R) a p, New Plain U.S.; L, New Limousin; T, New Troncais; G. New Gascony; C, New Charred U.S.; R, once reused U.S.; L(R), once reused Limousin; T(R), once reused Troncais. 5 Fill aged one year in Barrel 345, so total age 6.3 years. for sample 10 (~245), for which some possible significant comparisons will be noted later. In the samples of Group VI, the continuous still product from Peverella grapes was first filled into two new Limousin oak barrels. Sample 19 (Barrel 344) was aged for 6.3 years, but its companion, sample 20 (345), was emptied after only one year and its contents (less samples) transferred into a once reused U.S. barrel, sample 21 (366), for 5.3 years of additional aging. Barrel 345 was then refilled with the same, unaged distillate, which had been stored during the intervening year in a stainless steel drum, renumbered sample 22 (Barrel 367) and then aged for 5.3 years. This procedure was designed to study the use and reuse of Limousin oak barrels as compared with standard American oak. Most of the brandies listed in Table 1 are the same as were used and analytically characterized in a previous study of compounds extracted from oak during brandy aging (8). Sample ehromatograms are shown in Fig. 1 for 1 ~1 injections of concentrates of unaged brandy sample III-7 (top) and of the same brandy after four years of aging in a new U.S. oak barrel, sample III-8 (bottom). Comparable ehromatograms for all other samples studied are similar in appearance to these. The figure reveals that a number of compounds present in aged brandy are essentially products of oak extraction. These include furfural (peak 14), 5-methyl furfural (peak 17), diethyl succinate (peak 19), and the trans isomer of fl-methyl-~,-octalactone (peak 25), sometimes called "oak" or "whiskey lactone" (8). The cis diastereomer of this lactone is also extracted in lesser amounts from oak but is concealed on these chromatograms by fl-phenethyl alcohol (peak 24). An additional compound (peak 10), apparently an acetal, is present only in aged brandy samples and is particularly concentrated in brandies aged in new, charred oak barrels. The compounds given in Fig. 1 were identified using retention data, infrared spectroscopy and GC/MS. Typically total acids, largely acetic, and total esters, largely ethyl acetate, greatly increase in concentration during barrel aging. Total acids are initially quite low in concentration but increase rapidly during the first 12 months in the barrel and more gradually thereafter, while esters increase more uniformly throughout aging (2). The rise in total ester concentration as determined by chemical methods, typically is two to three-fold over four years, and is due almost entirely to ethyl acetate. Acetic acid, either extracted from oak or from air

4 VOLATILE CONSTITUENTS OF BRANDY~155 i' i i ~.6i ]! i i'' i27 1 ii I!2o 2,?,i Fig. 1. Gas chromatograms showing compounds in a concentrate of a brandy (Barrel 243) before (top) and after aging (bottom). 1 solvent, 2 ethyl formate, 3 ethyl acetate plus diethyl acetal, 4 ethyl alcohol, 5 n-propyl alcohol, 6 isobutyl alcohol, 7 isoamyl acetate, 8 isopentyl alcohols, 9 ethyl caproate, 10 unknown (acetal?), 11 n-hexyl acetate, 12 n-hexyl alcohol, 13 ethyl caprylate, 14 furfural, 15 benzaldehyde plus unknown acetal, 16 n- octyl alcohol [internal standard], 17 5-methyl furfural, 18 ethyl caprate, 19 diethyl succinate, 20 fl-phenethyl formate plus 1-decanol, 21 fl-phenethyl acetate, 22 ethyl laurate, 23 unknown, 24 fl-phenethyl alcohol plus oak lactone-a, 25 oak lactone-b, 26 caprylic acid, 27 ethyl myristate, 28 eugenol, 29 diethyl azelate, 30 capric acid, 31 ethyl palmitate. oxidation of alcohol, is esterified by the large excess of ethyl alcohol. The extent of concentration change depends on a number of factors including the initial composition ~ of the brandy distillate, which in turn depends on the composition of the distilling wine and the type and mode of operation of the still (3). The ester increase during aging depends largely on the activity of the barrel used, e.g. the amount of acetic acid available for extraction from the barrel and subsequent esterification with alcohol. Therefore, the Am. J0 Enol. Vitic., Vol. 28, No. 3, 1977 ester increase is always greater for brandies aged in new U.S. oak cooperage than in once or twice-reused barrels of the same type. Boruff and Rittschof (2) suggested that the increase is less with increasing barreling proof for American whiskies, i.e. the ester content referred to 100 proof is greater in whiskies barreled at 110 proof than in those at 127. Guymon and Crowell (4) found that ester increase during brandy aging is greatly affected by temperature of storage but less so by barrel proof. After five years brandies stored at 20 C contained approximately twice the concentration of total esters as those stored at 15 C. However, it is difficult to separate the influence of variations in ester contents due to analytical imprecision from variations in activity of individual barrels, supposedly the same. For the six groups of brandies described in Table 1, the levels of three acetate esters (not including ethyl acetate) and of four ethyl esters of fatty acids estimated from the gas chromatograms are given in Table 2. The changes in concentration as the result of aging are indicated by comparing the control (first sample in each group) with the aged samples of each group. All of the acetate esters of isoamyl, n-hexyl and fl-phenethyl alcohols decrease in concentration quite significantly during aging. In Group III brandies for example, the isoamyl acetate content of samples 8 through 13, all aged four years in a variety of U.S. and French oak barrels, is only 25-35~ of that contained in the unaged sample 7. Similar or greater decreases occur in the other two acetate esters. The most logical explanation for the loss of acetate esters is that ethyl alcohol is present in far greater concentration than the other alcohols and, acting as a nucleophilic reagent, displaces the other alcohol moiety in the following equilibrium reaction: H+ CH:~\ CHI'/CHCH.,COOCH:~ + \ C._,H~,OH ~- C._,H,~COOCH:~ + /CHCH._,CH._,OH CH:, CH:~ isoamyl ethyl ethyl isoamyl acetate alcohol acetate alcohol The mass action effect of the high ethyl alcohol content and the natural acidity of aged brandy, ph range about , would favor this acid-catalyzed transesterification reaction. Apparently acetates formed during aging by reaction of alcohols with acetic acid are partially lost through transesterification, resulting in a net overall loss. The three acetate esters included in Table 2 are generally described as having fresh, fruity odors and their decrease during aging might be considered detrimental to aroma intensity. Be that as it may, the net effect of aging is to enhance generally the odor appeal or bouquet even if the aromas associated with fruitiness are partially destroyed during aging or subdued by the odors derived from oak extractives. Also shown in Table 2 are analytical values for ethyl esters of the fatty acids, caproic (C,;), caprylic (Cs), capric (C,o) and lauric (C~). These compounds are the most abundant of the fatty acid ethyl

5 156--VOLATILE CONSTITUENTS OF BRANDY Table 2. Changes in concentration of various esters in brandy by aging in oak. Group Sample Acetate esters* no. lsoamyl Hexyl /~-Phenethyl C 6 Fatty acid ethyl esters* 08 C10 C12 1 (U) (P) (L) 1.8 tr (U) (P) (L) (U) (P) (L) 3.2 tr (T) 3.0 tr (G) 3.2 tr (L) 4.6 tr (R) IV 14 (U) (C) VI *mg/i at 100 proof. 16 (U) (T-R) (U) (L) (L) (R) (L-R) esters present in distilled spirits (15). The data indicare an increase for the first three of these esters but usually a small loss for ethyl laurate, particularly for those samples initially containing relatively higher concentrations. In all cases ethyl caprylate increases considerably more than either caproate or caprate. The next higher esters, ethyl myristate ~/nd palmitate, not shown in Table 2, appear to drop in concentration even more than laurate. The peak areas of these two esters could not be estimated from the chromatograms due to interference from free caprylic and capric acids which elute just before myristate and palmitate respectively on Carbowax 20M columns. Because of possible losses by hydrolysis of some compounds, particularly the oak lactone, the decision was made not to remove free acids from the brandy samples by alkaline washing of the solvent extracts before concentration as was a part of our previous procedure (8). A satisfactory explanation for these patterns of change of fatty acid esters during aging is not readily apparent but is likely related to the relative concentrations of the free acids and their respective ethyl esters present before aging. Probably the free acids, Cs through C~o, initially exceed the equilibrium levels of their esters, so during aging some of the free acids are esterified with the abundant ethyl alcohol. All samples were analyzed by gas chromatography im- mediately after the concentration step so it is unlikely that the concentration changes occur artificially in the 500"1 concentrate rather than during the relatively long-term aging. The data of Table 2 show further that the type of oak barrel used has little effect on the changes occurring in either class of esters. Table 3 shows the estimated concentrations of compounds largely extracted from oak during aging. The first two compounds, furfural and 5-methyl furfural, are products of acid dehydration and cyclization of, respectively, pentoses and rhamnose to form the appropriately substituted furan ring compound. Furfural may be formed during distillation, particularly pot distillation, but is mostly derived from extraction of oak as is 5-methyl furfural. The concentrations found in brandies aged in American oak are always appreciably higher than in those aged in French oak barrels. The values given in Table 3 for furfural are lower than the concentrations which have been reported for this compound by colorimetric or spectrophotometric methods of analysis and by gas chromatography. Concentrations times higher than those given in the table are customary for spirits aged in new U.S. oak barrels (2,8). However, the concentrations of furfurals found in this study are comparable to those reported by Suomalainen and Nyk~nen (15) in a variety of

6 VOLATILE CONSTITUENTS OF BRANDY--157 whiskies by gas chromatography of concentrated ether-pentane extracts. The reason for this discrepancy was recently determined to be that etherpentane is a relatively poor extractant for furfural as compared to methylene chloride. In a solvent comparison experiment, 300 ml of a 40 volume % alcohol solution containing approximately 5 mg furfural and 2.5 mg l-octanol as internal standard was extracted three times with 50 ml of ether-pentane, washed and concentrated for gas chromatography as described above. Another 300 ml of the test solution was extracted with methylene chloride in a parallel manner. Gas chromatography of the concentrated extracts indicated that only about 17 % of the furfural was recovered by the ether-pentane solvent whereas about 77% was recovered by methylene chloride solvent. Presumably 5-methyl furfural exhibits similar solvent partition properties. Diethyl succinate increased significantly as a result of aging for all of the samples, more so in American than in French oak. This compound is also present in small amounts in the unaged distillate, contrary to a previous statement (7). Since succinic acid is a normal component of wine, some would be esterified with ethanol either before or during the distillation process. The volatility of diethyl succinate in dilute alcohol-water solutions is not known. However, it boils at C, and is soluble in alcohol but insoluble in water. Therefore it should be weakly volatile in dilute alcohol solutions, comparable to fatty acid esters, so some diethyl succinate should distill over into the brandy distillate. However, it appears that the more important source of diethyl succinate is a reaction between ethyl alcohol and succinic acid of wood or possibly decomposition of wood constituents during aging. Small amounts of diethyl azelate [C~H~4 (C00C2H5) 2] were also found in aged brandy (peak 29, Fig. 1), presumably formed in a fashion parallel with diethyl succinate. The cis and trans isomers of fl-methyl-~/-octalactone (oak lactones -a and -b) are products of oak extraction (11) and are much more concentrated in brandies aged in U.S. oak (8). As noted before, oak lactone-a, the cis isomer, has the same retention time as fi-phenethyl alcohol on Carbowax 20M but its presence can be readily demonstrated in neutral alcohol-water extracts of oak. The values given in the table for combined fi-phenethyl alcohol and oak lactone-a show an increase during barrel aging. Obviously, some of the increase is due to fl-phenethyl alcohol from hydrolysis of its acetate. The size of peak 24 (fl-phenethyl alcohol + oak lactone-a) was larger for sample 10, the Troncais oak barrel, than for the other aged brandies in Group III. This suggests that a somewhat greater amount of oak lac- Table 3. Minor brandy compounds produced primarily by aging.* Group Sample 5-Methyl Diethyl no. Furfural furfural succinate /~-Phenethyl alc. (+ Oak lactone-a) Oak lactone-b 1 (U) 0.13 tr (P) (L) (U) m 5 (P) (L) tr (U) tr. tr (P) (L) (T) (G) (L) (R) IV 14 (U) 15 (C) (U) (T-R) tr. ~ VI 18 (U) (L) (L) (R) (L-R) tr. tr * mg/i at 100 proof.

7 158--VOLATILE CONSTITUENTS OF BRANDY tone-a was extracted from the Troncais oak since the fl-phenethyl alcohol should have been about the same for all aged samples of this group. Brandy sample 17 of Group V was aged nine years in the same Troncais oak barrel previously used for sample 10. Both samples reflect a relatively high value for combined fi-phenethyl alcohol and oak lactone-a. A study made in this laboratory of neutral alcohol extracts of several authentic oak chip samples agreed with previously reported results for U.S. and Limousin oaks (8); U.S. oak contains relatively large amounts of oak lactone-b but little of lactone-a and Limousin contains only traces of either isomer. One sample of Troncais oak chips contained 10 times more oak lactone-a than -b. On the other hand, oak chip samples taken from two different Troncais staves contained the reverse ~ nearly 10 times more -b than -a. The significance, if any, of these seemingly anomalous results for Troncais oak is not clear. As indicated earlier, Group VI includes a continuous still unaged brandy and four unaged samples involving two new Limousin barrels and one reused U.S. oak barrel. The changes in esters during aging as shown in Table 2 and of oak-derived components as seen in Table 3 are consistent with the general effects of aging in oak as already discussed. However, the levels of some constituents found in samples are inconsistent with the aging treatment used. Particularly since sample 20 (one year in new Limousin) was the fill for sample 21 (5.3 additional years in a once used U.S. barrel), one would expect the changes by aging would in all cases be greater in sample 21 than in sample 20. This was true for the C,~-Cs fatty acid esters, diethyl succinate, oak ]actone-b and the fi-phenethyl alcohol + lactone-a components, but did not hold for other constituents. Furthermore, one would expect the aging effects in sample 19 (6.3 years in new Limousin) to be greater than in sample 22 (5.3 years in reused Limousin). This was true as to the relative increases during aging in C,-Cs esters, the two furfurals and diethyl succinate and the relative decreases in the acetate esters. However, the apparently greater levels of the two oak lactones in the brandy aged in the reused Limousin barrel suggest variations in oak composition between the two supposedly identical containers. Of course, the differences in concentrations of these minor constituents in the four aged samples are small and perhaps some inconsistencies are due to analytical errors. In summary, it can be concluded that while the total amount of esters increase during aging in oak, most of the increase is due to ethyl acetate and to a lesser extent the ethyl esters of caproic, caprylic and capric acids. However, the acetate esters of higher alcohols, especially of isoamyl, decrease, presumably by transesterification of the alcohol moiety with the much more abundant ethyl alcohol. Somewhat larger amounts of volatile oak extractives, including furrural, 5-methyl furfural, diethyl succinate, and the two oak lactones (cis and trans fl-methyl-7-octalactone) are obtained in brandies aged in U.S. oak than in French oak. LITERATURE CITED 1. Belchior, A. P., and L. C. Carneiro. Identification de substances extraites du bois neuf de ch6ne du Limousin par deseaux-de-vie de vin. Connaissance de la Vigne et du Vin 6: (1972). 2. Boruff, C. S., and L. A. Rittschof. Effects of barreling proof on the aging of American whiskeys. J. Agric. Food Chem. 7:630-3 (1959). 3. Guymon, J. F. Chemical aspects of distilling wine into brandy. Chemistry of Winemaking. Chapter 11, pp (A. D. Webb, ed.) Adv. in Chemistry Series 137, Am. Chem. Soc., Washington, D.C. (1974). 4. Guymon, J. F., and E. A. Crowell. Unpublished data. 5. Guymon, J. F., and E. A. Crowell. Separation of vanillin, syringaldehyde, and other aromatic compounds in the extracts of French and American oak woods by brandy and aqueous alcohol solutions. Qual. Plant. Mater. Veg. 16: (1968). 6. Guymon, J. F., and E. A. Crowell. Gas chromatographic determination of ethyl esters of fatty acids in brandy or wine distillates. Am. J. Enol. Vitic. 20:76-85 (1969). 7. Guymon, J. F., and E. A. Crowell. Brandy aging. Some comparisons of American and French oak cooperage. Wines and Vines 51:23-5 (1970). 8. Guymon, J. F., and E. A. Crowell. GC-separated brandy components derived from French and American oaks. Am. J. Enol. Vitic. 23: (1972). 9. Kahn, J. H., P. A. Shipley, E. G. LaRoe, and H. H. Conner. Whiskey composition: Identification of additional components by gas chromatography-mass spectrometry. J. Food Sci. 34: (1969). 10. Liebich, H. M., W. A. Koenig, and E. Bayer. Analysis of the flavor of rum by gas liquid chromatography and mass spectrometry. J. Chromatogr. Sci. 8: (1970). 11. Masuda, M., and K. Nishimura. Branched nonalactones from some Quercus species. Phytochemistry 10: (1971). 12. Reazin, G. H., S. Baldwin, H. S. Scales, H. W. Washington, and A. A. Andreasen. Determination of the congeners produced from ethanol during whiskey maturation. J. Assoc. Off. Anal. Chem. 59:770-6 (1976). 13. Schafer, J., and R. Timmer. Flavor components of cognac. J. Food Sci. 35:10-12 (1970). 14. Suomalainen, H., and L. Nyk&nen. The aroma components produced by yeast in nitrogen-free sugar solution. J. Inst. Brew. 72: (1966). 15. Suomalainen, H., and L. Nyk~.nen. Composition of whisky flavour. Process Biochem. 5:13-18 (1970). 16. Valear, Peter. Brandy. Ind. Eng. Chem. 31: (1939).