Charring involves exposing the cask to controlled heat to establish a thermal gradient through the wood for a specified time. In the Scotch whisky industry it has been used mainly to regenerate exhausted casks for further use, but is carried out routinely in the United States, where all Bourbon whiskey must be matured in new charred oak. This is partially responsible for the smooth, sweet, vanilla flavour characteristic of these whiskies. After only one maturation these casks are used for subsequent maturations of other whiskies, including Scotch.
Many authors have studied the chemical changes brought about by charring, the majority of whom have found increased levels of the aromatic aldehydes vanillin and syringaldehyde (Reazin, 1983; Philp, 1989; Conner et al., 1990) and increased levels of other aldehydes and esters (Nishimura et al., 1983). All these authors studied the effects of charring using oak wood chips with storage times of up to one month in alcohol, with the exception of Philp who used full-sized American and Spanish casks with a maturation time of 2 years. The effectiveness of cask charring has been found to vary with the temperature used, higher temperatures resulting in greater release of aromatic aldehydes into the maturing spirit (Nishimura & Matsuyama, 1989). Increased removal of sulphur compounds by charred wood has also been reported (Philp, 1989; Perry, 1986).
Thus, the effect of charring can be described in terms of (1) the production of a layer of highly active adsorbent which effectively removes many 'undesirable' flavour congeners more rapidly and catalyses other changes; (2) the thermal gradient produced beneath this layer produces a controlled release of flavour congeners from the cask lignin; and (3) the total polyphenolic content of the extract is increased due to disruption of the wood surface and an increase in effective surface area with the potential production of flavour congeners by the oxidation of the vicinal polyhydric phenols (Philp, 1989).
There have been few studies of the changes in sensory properties caused by cask charring and no descriptive sensory analyses of such changes have been reported. The work described here was undertaken to investigate the effect of cask charring and subsequent maturation on a new distillate, using both chemical and descriptive sensory methods, and to relate the changes in sensory characteristics to the chemical changes taking place.
As reference materials (>95%) 2-methyl-butan-1-ol, 3-methyl-butan-1-ol, 1-amyl alcohol, 1-hexanol, furfural, 5-methyl furfural, butanoic acid, 2-methyl butanoic acid, 3-methyl butanoic acid, 1-octanol, ethyl hexanoate, ethyl octanoate, ethyl decanoate, vanillin, vanillic acid, syringaldehyde, syringic acid, 2,3-dimethyl phenol, 2-naphthol, 4-coumaric acid and ferulic acid (Aldrich Chemical Company Ltd, Gillingham SP8 4JL, UK), and ethyl nonancate, ethyl dodecancate, ethyl tetradecanoate, ethyl hexadecanoate, 1-decanol, 1- dodecanol, 1-tetradecanol, 1-hexadecanol, ellagic acid, octanoic acid, decanoic acid, dodecanoic acid, tetradecanoic acid, hexadecanoic acid, 2-phenylethanol, phenol and garlic acid (Sigma Chemical Company Ltd, Poole BH17 7NH, UK) were used; others were hexanoic acid (GPR; BDH Ltd, Poole BH12 4NN, UK) and dichloromethane and water (HPLC; Rathburn Chemicals Ltd, Walkerburn EH43 6AU, UK).
Dichloromethane (2 ml) was added to spirit (10 ml), diluted to 23% v/v ethanol with water, shaken overnight at approximately 20°C and the dichloromethane separated and analysed immediately.
Gas chromatography (GC) (Carlo Erba 5300 equipped with cold on-column injector and flame ionisation detector) was on a 25mx0.32mm Carbowax BP20 column (SGE UK Ltd, Milton Keynes MK11 3LA, UK) with a 2mx0.53mm deactivated fused silica pre-column and helium at 1.8ml min-1 as carrier gas. Samples were analysed in duplicate with an initial temperature of 60°C, increasing after 3 min to 230°C at 4°C min-1. Peak areas were measured with a computing integrator and standardized with 2,3-dimethylphenol (10 ng internal standard on column). Variation between replicates was ±5% standardized peak area or better. Components of selected samples were analysed by gas chromatography-mass spectrometry (Finnegan-MAT ITS40), using the same column and conditions. Comparison of mass spectra and retention times with authentic compounds identified 30 compounds, and a further 18 were tentatively identified by mass spectra and are marked (T) in Table 1.
Spirit samples were analysed for total phenolic content using Folin-Denis reagent and for absorbance at 520 nm with glass stored distillate as reference. Absorbance at this wavelength was found by MacDougall (1989) to provide the best correlation with lightness (CIELAB L*), a measure of intensity of colour. Galloyl esters were determined by reaction of sample (2 ml) with acetone (0.5 ml) and 1M sodium hydroxide (0.5 ml) with absorbance measured after 2min at 485 nm against a blank of sample (2 ml), water (0.5 ml) and 1M sodium hydroxide (0.5 ml), with garlic acid as calibration standard (D. Pert, personal communication).
Gallic, vanillic, syringic, coumaric, ferulic and ellagic acids, vanillin, syringaldehyde, coniferaldehyde, sinapaldehyde and total ellagitannins (Moutounet et al., 1989) were analysed by high performance liquid chromatography (HPLC) using a Kratos 400 solvent pump, Kratos 430 low pressure gradient former and Kratos 470 autosampler with 20µl loop (ABI Analytical, Warrington WA3 7PD, UK). Compounds were separated on a Spherisorb S5 ODS2 column (Phase Separations Ltd, Clwyd CH5 2NU, UK) using a 0.1 M orthophosphoric acid: methanol gradient (Casteele et al., 1983) with detection by LC-UV detector (Pye Unicam Ltd, Cambridge CBI 2PX, UK) at 300 nm and integration. Samples were diluted to 30% v/v ethanol and analysed in triplicate with peak areas standardized on 2-naphthol (400 ng internal standard on column). Available standard substances were used for calibration and gave mess: area correlation co-efficients of >0.997. Variation between replicates was ±5% standardized peak area, or better.
|1||(a)iso-amyl alcohol||25||'oak' lactone (cis) (T)|
|3||1-amyl alcohol||27||'oak' lactone (trans) (T)|
|5||1,1,-diethoxy-2 propanone (T)||29||phenol|
|6||ethyl lactate (T)||30||(b)unknown phenol|
|8||ethyl octanoate||32||octanoic acid|
|9||furfural||33||tetradecyl acetate (T)|
|12||ethyl nonanoate||36||ethyl hexadecanoate (T)|
|14||5-methyl furfural||38||2-phenylethyl octanoate (T)|
|16||ethyl decanoatc||40||dodecanoic acid|
|17||(a)iso-amyl octanoate (T)||41||(b)unsat. C-18 ethyl ester|
|18||(c)diethyl succinate/isovaleric acid||42||vanillin|
|19||ethyl decenoate (T)||43||2-phenylethyl decanoate (T)|
|20||1-decanol||44||ethyl vanillate (T)|
|21||2-phenylethyl acetate (T)||45||tetradecanoic acid|
|22||hexanoic acid||46||hexadecanoic acid|
|23||ethyl dodecanoate||47||hexadecenoic acid (T)|
|24||(a)iso-amyl decaneate (T)||48||syringaldehyde|
|a||mixture of 2-methyl-butyl and 3-methyl-butyl.|
|b||incomplete identification from mass spectra.|
A panel of 20 experienced assessors described the aroma of the samples, using a vocabulary of 24 terms (Piggott & Canaway, 1981; Piggott, 1991), listed in Fig. 4, on an intensity scale from 0 to 5. The samples were presented to the panel at 23% v/v alcohol concentration in tulip shaped nosing glasses similar to standard wine tasting glasses (BS5586: 1978) but of approximately 150 ml capacity, covered with watchglasses and assessed in individual sensory booths under red lighting to minimize colour differences. Each sample was assessed independently in duplicate together with similar samples from another maturation trial, eight samples per session, over three sessions on consecutive days. Data were collected either on machine-readable cards (Piggott, 1983) or using the PSA-System (Oliemans, Punter & Partners, PO Box 14167, 3508 SG Utrecht, The Netherlands).
Chemical analytical data and mean panel aroma description intensities were separately analysed by principal components analysis (PCA; Piggott & Sharman, 1986) with rotation according to the Varimax criterion (Kaiser, 1959). The principal components are linear combinations of variables, calculated so as to describe as much of the variance of the original data as possible. This allows the original multidimensional matrix of samples on chemical measures or sensory descriptors to be simplified without substantial loss of information. The results of PCA can be graphically displayed as two sets of plots. In the first, the loadings of the individual chemical or sensory variables on successive principal components (i.e. correlations of the variables with the components) can be plotted to aid interpretation of the components in terms of the original variables; in the second the individual sample scores on each principal component can be plotted to show relationships between samples and changes over time. Thus a loadings plot shows which of the original variables are important in a particular principal component, and a component scores plot (not to be confused with the raw sensory scores) shows the calculated values of the samples for the particular combination of variables. A principal component consists largely of a group of correlated variables; since the variables are correlated they must, in general, change in the same way. Thus samples having different cal- culated component scores must, in general, differ in the same way on all the variables contributing strongly to the component. This multivariate data analysis approach has two major advantages compared with a conventional univariate approach. Firstly, it eases interpretation by simplifying complex data matrices; secondly, it detects trends in data sets which might not be apparent in a univariate analysis by displaying the systematic variation in a number of variables, and ignoring the unsystematic variation.
|Maturation time||12 months||24 months||36 months|
|Total phenols (mg garlic acid l-1)||0.24||0.26||0.17||0.18||0.29||0.31||0.23||0.22||0.30||0.31||0.24||0.23|
|Galloyl esters (mg garlic acid l-1)||0.017||0.019||0.024||0.021||0.029||0.031||0.023||0.022||0.030||0.031||0.024||0.023|
|Gallic acid (mg l-1)||3.4||4.5||3.7||5.7||4.0||6.0||4.9||6.9||4.7||6.0||5.2||9.3|
|Vanillin (mg l-1)||1.7||2.2||1.7||1.7||3.2||4.2||3.1||3.3||3.4||3.2||2.9||3.5|
|Syringaldehyde (mg l-1)||6.2||8.8||4.4||4.5||13||16||9||10||14||18||10||11|
|Vanillic acid (mg l-1)||3.2||3.8||8.1||8.1||2.7||3.7||4.2||4.9||2.6||2.7||4.4||4.3|
|Syringic acid (mg l-1)||1.7||2.1||1.7||1.6||2.6||3.2||2.4||2.6||2.8||2.3||2.4||2.5|
|Coumaric acid (mg l-1)||0.29||0.41||0.17||0.36||0.24||0.42||0.32||0.56||0.37||0.51||0.13||0.32|
|Ferulic acid (mg l-1)||0.18||0.15||0.22||0.24||0.73||0.47||0.70||0.52||0.90||0.51||0.70||0.51|
|(a)Ellagitannins (mg l-1)||5.8||8.1||11.0||9||10||14||13||10||11||16||19|
|(b)Coniferaldebyde (mg l-1)||7.5||6.9||10.3||10.0||6||7||10||10||11||10||16||17|
|(b)Sinapaldebyde (mg l-1)||9.5||9.4||14.6||13.0||4.6||5.1||6.7||7.0||13||13||20||21|
|(a)Ellagic acid (mg l-1)||0.57||1.1||0.13||0.44||0.20||0.18||0.15||0.25||0.26||0.18||0.44||0.25|
(a) Estimated from response factor for garlic acid.
(b) Estimated from response factor for ferulic acid.
Results of the non-volatile and HPLC analyses, for the 1-, 2- and 3-year old samples, are shown in Table 2. Analysis of variance for the cask effect over all time periods showed that the charred cask samples had significantly higher levels of total phenols, absorbance at 520 nm, and syringaldehyde (P < 0.001), and significantly lower levels of vanillic acid (P<0.025), coniferaldehyde (P<0.025) and sinapaldehyde (P < 0.05). All of the variables in this set increased with maturation time (P < 0.001) with the exception of coumaric and ellagic acids, which did not change significantly with time for either charred or uncharred cask samples, and absorbance which remained constant for the charred cask samples.
Trends in the data with time were examined more closely by PCA. Variable loadings on the the first two rotated components (which accounted for 70% of total variance) are shown in Fig. 1, and the sample scores in Figs 2 and 3. The principal component scores for the charred cask samples were consistently lower than the uncharred cask samples (Fig. 2), corresponding to generally higher values for the variables which were negatively correlated with component 1 (Fig. 1), particularly total phenols, galloyl esters, vanillin, syringaldehyde and syringic acid. Samples from both types of cask followed a similar trend with maturation time. The principal component 1 scores for the uncharred cask samples were consistently lower than the charred cask samples (Fig. 3), corresponding to generally higher values for garlic and vanillic acids, ellagitannins, coniferaldehyde and sinapaldehyde, which had negative loadings on component 2 (Fig. 1). Again samples from both cask types followed similar trends.
The effect of ethanol concentration was less obvious, but Fig. 3 shows that in both cask types the 63.4% distillate had consistently larger principal component scores from 12 months maturation onwards, corresponding to generally lower concentrations of the compounds negatively correlated with component 2. Inspection of Table 2 shows that this effect was primarily due to garlic acid, suggesting that the lower ethanol concentration was less effective at extracting these compounds, garlic acid, vanillic acid, ellagitannins, coniferaldehyde and sinapaldehyde.
Analysis of variance showed that the charred cask samples were described as more smooth, vanilla, sweet, malty (P < 0.001), spicy, fruity (eatery) and floral (P<0.01), and in many respects the longer-matured samples were reminiscent of Bourbon whiskeys. In contrast the uncharred cask samples were described as more pungent, grainy, sour, oily, sulphury (P<0.001), catty (P<0.01), meaty and fishy (P<0.05). Similar changes occurred in both charred and uncharred cask samples during maturation. Ratings of spicy, fruity (other), smooth, sweet and vanilla increased (P < 0.001) as maturation progressed. Ratings of pungent (P < 0.001) and sour (P < 0.05) also increased significantly during maturation for the uncharred cask samples. Summaries of the sensory data at 12, 24, and 36 months maturation are shown in Table 3.
|Maturation time||12 months||24 months||36 months|
|Fruity (estery)||1.0||0.9||0.9||0.7||1.0||1.0||0.8||0 7||0.9||1.1||0.8||0.8|
The first two components from PCA of the sensory data explained 56% of thetal variance; loadings of the descriptors on the first and second rotatedmponents are shown in Fig. 4, and sample scores in Figs 5 and 6. Componenthad high loadings for spicy, smoot h, vanilla, woody and sweet. Guy et al. (1989) found thate first principal component, from PCA of trained panel descriptive sensory data a set of whiskies, matched the second axis from analysis of consumer free- choice profile data on the same whiskies, which was correlated with the descriptors 'maturity' and 'smoothness'. Given the nature of the highly loaded descriptors on the first principal component of the present analysis, and the steady increase in component scores with time, this component can be regarded as representing 'maturity' in the positive direction. Component 2, with high loadings for pungent, soapy, sour, grassy and oily, can be regarded as representing 'immaturity' in the negative direction (Fig. 4)
The charred cask samples had higher principal component scores than the uncharred cask samples on component 1 (except for the 12-month samples). This difference corresponded to generally higher scores for the correlated descriptors, associated with 'maturity' (Fig. 5). The uncharred cask samples had consistently lower principal component scores than the charred cask samples on component 2, similarly corresponding to generally higher scores for the descriptors associated with 'immaturity' (Fig. 6). Similar trends with maturation time were evident for both charred and uncharred cask samples. For the charred cask samples there was a consistent increase in 'maturity' coinciding with a decrease in 'immaturity' up to 18 months. The principal component scores, and therefore the mean panel ratings, for mature characteristics generally levelled off at 30 months for both charred and uncharred cask samples (Fig. 5). For the uncharred cask samples substantial development of mature characteristics was not evident until 18 months of maturation. The ethanol concentration had little effect on the charred cask samples, but the high strength samples from the uncharred casks had consistently lower scores on both components 1 and 2.
It is clear that differences between samples from charred and uncharred casks were present after 3 months maturation, and remained for the 36 months of this study (Figs. 2, 3, 5, and 6). The differences in the non-volatile compounds fell into two groups; those measures increased by charring and associated with principal component 1, and those decreased by charring and associated with principal component 2 (Table 2 and Fig. 1). However these two groups were not simply opposed, or they would have been described by only one principal component. The two principal components required are by definition independent, so two separate effects must have been present in the data. The 'char effect' (Fig. 2) was present after 3 months, and there was continuing change up to about 24 months. This would correspond to charring having caused a rapid release of observed coloured material (presumably polymeric) and other compounds (particularly phenols, galloyl esters, vanillin, syringaldehyde and syringic acid) into the maturing spirit. The differences between the cask types then persisted for the three years of this study, but the process of active extraction seems to have ceased after about 24 months.
On the contrary, the 'plain wood effect' was small and there was very little change in the levels of materials extracted up to 18 months, and thereafter a more rapid change and greater difference developed between cask types (Fig. 3). This could correspond either to material requiring longer to diffuse out from the wood (not damaged by charring), or to time being required for polymeric wood components to break down.
The sensory data similarly showed two independent sets of changes. The 'maturity' character shown in Fig. 5 can be seen to roughly parallel the 'char effect' (Fig. 2). The sensory descriptors associated with principal component 1 (Fig. 5) could well have been related to the aromatic aldehydes (vanillin and syringaldehyde) found at higher levels in whisky from charred casks (Table 2). However, the aromatic aldehydes (coniferaldehyde and sinapaldehyde) found at higher levels in uncharred cask samples, and associated with principal component 2, may contribute to the developing 'maturity' in the uncharred cask samples after 18 months (Fig. 5). The origin and meaning of the second sensory principal component (Fig. 6) is less clear.
On the basis of the associated descriptors it was dubbed 'immaturity', but it is obviously not a simple case of immature character reducing during maturation. This component seemed to be associated with an initial reduction for about 18 months in notes characteristic of immature distillates, followed by an increase. This could be explained as an initial decrease in pungency and sourness in the charred cask samples, as concentrations rise of sweet-smelling compounds such as vanillin and syringaldehyde. However, once the char layer is exhausted of readily extractable materials, the predominant sensory impression is of increasing pungency and sourness as the second wave of compounds is extracted into the spirit (Fig. 3).
It has been shown (Piggott et al., 1992) that one of the effects of maturation of spirits in wood is to reduce the headspace concentration of ethyl esters, and probably of other compounds of similar polarity. No allowance was made for this effect in the analyses used here, the extraction method for &C analysis providing a near-complete extraction of the volatile compounds present. Therefore, it is likely that, if a different extraction method or direct headspace sampling had been used, a different effect of maturation or charring would have been found in the GC data, and might have been related to the aroma descriptive data. As it was, the GC data showed little change which could have been related to the changes in the aroma descriptive data. It is tempting to suggest that better relationships should be found between flavour-bymouth and total (volatile plus non-volatile) chemical data, and between aroma and volatile data. However, Piggott & Jardine (1979) showed that little extra information was provided by tasting whisky, compared with nosing, and the limited relationships found here suggested that changes in the non-volatile and less-volatile compounds (Table 2) were paralleled by changes in aroma.
This work was based on the 3 year maturation of new distillate in industrial size casks and so is directly comparable to commercial maturations of Scotch whisky. At the end of the maturation two entirely different whiskies were produced. The sensory profile of the charred product was closer to that of a Bourbon than a Scotch whisky. The sensory profile of the whisky matured in new uncharred casks resembled that of an immature whisky. Therefore there are benefits from maturing whisky in new charred wood but the resulting product differs from a standard Scotch whisky.
Conner, J.M., Paterson, A. & Piggott, J.R. (1990).
A comparison of charred and uncharred American white oak (Quercus alba). In: Proceedings of the Third Aviemore Conference on Malting, Brewing and Distilling (edited by I. Campbell). Pp. 457 - 459. London: Institute of Brewing.
Guy, C., Piggott, J.R. & Marie, S. (1989).
Consumer profiling of Scotch whisky. Food Quality and Preference, 1, 69 - 73.
Kaiser, H.F. (1959).
Computer program for Varimax rotation in factor analysis. Educational and Psychological Measurement, 19, 413-420.
MacDougall, D.B (1989).
Measurement of food and beverage colour. In: Distilled Beverage Flavour (edited by J.R. Piggott & A. Paterson). Pp. 85 - 96. Chichester: Ellis Horwood.
Moutounet, M., Rabier, P., Puech, J.-L., Verette, E. & Barillere, J.M. (1989).
Analysis by HPLC of extractable substances in oak wood-application to a Chardonnay wine. Science des Aliments, 9, 35-51.
Nishimura, K. & Matsuyama, R. (1989).
Maturation and maturation chemistry. In: The Science and Technology of Whiskies (edited by J.R. Piggott, R. Sharp & R.E.B. Duncan). Pp. 235-263. Harlow: Longman.
Nishimura, K., Ohnishi, M., Matsuda, M., Koya, K. & Matsuyama, R. (1983).
Reactions of wood components during maturation. In: Flavour of Distilled Beverages: Origin and Development (edited by J.R. Piggott). Pp. 241-255. Chichester: Ellis Horwood.
Perry, D.R. (1986).
Whisky maturation mcchanisms. In: Proceedings of the Second Aviemore Conference on Malting, Brewing & Distilling (edited by I. Campbell & F.G. Priest). Pp. 409 - 412. London: Institute of Brewing.
Philp, J. (1989).
Cask quality and warehouse conditions. In: The Science and Technology of Whiskies (edited by J.R. Piggott, R. Sharp & R.E.B. Duncan). Pp. 264- 294. Harlow: Longman.
Piggott, J.R. (1983).
Automated data collection in sensory analysis. In: Flavour of Distilled Beverages: Origin and Development (edited by J.R. Piggott). Pp 215-218. Chichester: Ellis Horwood.
Piggott, J.R. (1991).
Selection of terms for descriptive analysis. In: Sensory Science Theory and Applications in Foods (edited by H.T. Lawless & B.P. Klein). Pp. 339-351. New York: Marcel Dekker.
Piggott, J.R. & Jardine, S.P. (1979).
Descriptive sensory analysis of whisky flavour. Journal of the Institute of Brewing, 85, 82 - 85.
Piggott, J.R. & Canaway, P.R. (1981).
Finding the word for it: methods and uses of descriptive sensory analysis. In: Flavour '81 (edited by P. Schreier). Pp. 33-46. Berlin: Walter de Gruyter.
Piggott, J.R. & Sharman, K. (1986).
Methods for multivariatc dimensionality reduction. In: Statistical Procedures in Food Research (cditcd by J.R. Piggott). Pp. 181-232. London: Elsevier Applied Science.
Piggott, J.R., Conner, J.M., Clyne, J. & Paterson, A. (1992).
The influence of non-volatile constituents on the extraction of ethyl esters from brandies. Journal of the Science of Food and Agriculture, 59, 477 - 482.
Reazin, G.H. (1983).
Chemical analysis of whisky maturation. In: Flavour of Distilled Beverages: Origin and Development (edited by J.R. Piggott). Pp. 225-240. Chichester: Ellis Horwood.