Sugar and Alcohol Stabilization of Yeast in Sweet Wine

APPLIED MICROIOLOGY, July. 1968, p 1067-1075 Copyright 1968 American Society for Microbiology Vol. 16, NW. 7 Printed In U.S.A. Sugar and Alcohol Stabilization of Yeast in Sweet Wine RALPH E. KUNKEE AND MAYNARD A. AMERINE Department of Viticulture and Enology, University of California, Davis, California 95616 Received for publication 19 April 1968 Secondary fermentation of sweet wine was prevented by the Delle stabilization procedure. For this procedure, advantage is taken of the inhibitory effects of high concentrations of sugar as well as of alcohol. Thus, relatively small amounts of wine spirits were added to fermenting musts to obtain stability, as compared to the conventional procedure in which larger amounts of alcohol are added and the inhibitory effect of alcohol only is considered. The Delle value is a function of the concentrations in the wine, after spirits addition, of alcohol and sugar. Delle values which gave stable wine were dependent on time of alcohol addition, on strain of wine yeast, and on composition of wine spirits. Fractional addition of spirits, concentration of SO2, and clarity of must had little effect on the Delle value. Sensory comparison of wines especially prepared for tasting by the Delle procedure and by the conventional procedure showed the wines made by the Delle procedure to be superior in quality. Under proper storage conditions, the Delle wines were shown to be microbiologically stable and resistant to wine spoilage organisms. Yeast metabolism is inhibited by high concentrations of sugar and of alcohol. Delle (6) speculated that a combination of alcohol and sugar at levels less than required separately for complete inhibition might be used to inhibit growth and fermentation of yeast in sweet wine. He suggested the following formula: a + 4.5 c = DU, where a is the percentage of sugar (grams of reducing sugar/100 ml), c is the volume percentage of alcohol (milliliters of ethyl alcohol/100 ml), and DU ("Delle units") is a measure of the inhibitory activity of the mixture. Delle had observed that alcohol fermentation in wine musts would not occur at about 18% (v/v) alcohol or at about 80% (w/v) sugar. Thus, the equation is based on the inhibitory activity of alcohol being 4.5 times, in terms of the above units, that of sugar. A combination, then, of sugar and alcohol concentrations to give a DU of 80 should inhibit fermentation. Delle and other Russian workers of that time and since have verified the equation in general and have studied the theoretical and practical aspects of the equation. A summary of this literature has been given by us (2). Nomograms, both linear and circular, have been prepared for easy calculation of Delle units (4, 7, 18). Delle's equation might have special application for stabilization of dessert wines in California. These wines are usually stabilized by addition of wine spirits to give at least 19.5% alcohol (the minimum legal limit in California). Because of the inhibitory effect of the sugar already present in these wines, one should be able to add smaller amounts of spirits to achieve stability. Preliminary studies on the validity of the Delle equation applied to California conditions have been reported by us (2). The equation was found to be useful but somewhat variable. Stability occurred with DU values ranging from 75 to 85, depending on the extent of fermentation when the wine spirits addition was made. In the work given here, we have extended these studies to include several variables, such as yeast strain, concentration of SO2, clarity of 1067 must, fractional addition of spirits, and kind of spirits, and to determine the sensory qualities of wine prepared with the use of the Delle equation. MATERIALS AND METHODS Vinification. Unless noted otherwise, the following wine-making procedures were used: musts were prepared from grapes grown in the University vineyards at Davis and harvested at maturity (Table 1). The grapes were crushed, pressed, and sulfited o Eh same day as harvested. The volume of must used for each experiment was about 500 liters. After being settled overnight at 0 C, the musts were transferred to a fermentation room maintained at 20 to 22 C and inoculated with 1% yeast (Saccharomyces cerevisiae var. ellipsoideus, Montrachet strain, UCD Enology no. 522). Starter cultures were prepared in autoclaved grape juice to which 150 mg of SO2 per liter

1068 KUNKEE AND AMERINE TABLE 1. Must analyses APPL. MICROBIOL. Variety Date of harvest 'Brix Taitdite ph S02 added mg/liter Chardonnay-Pinot blanc (5:2) 27 Sept. 1965 24.9 0.75 3.41 125, 140 Semillon 4 Oct. 1965 25.6 0.69 3.45 125 Sauvignon blanc 11 Oct. 1965 26.5 0.64 3.43 75 Semillon-French Colombard (2:1) 1 Nov. 1965 26.4 0.59 3.57 133 aas grams of tartaric acid/100 ml. had been added after autoclaving. Large samples of the musts were removed during fermentation for spirits addition, and the remaining portions were fermented to dryness. The dry wines were racked and stored in full glass containers at 13 C. Wine spirits addition (fortification). At indicated times during the fermentation, samples of must were removed from the fermentation tanks and divided, by weight, into six, seven, or eight portions of about 6 liters each. Wine spirits were added to each portion to give a calculated final concentration of alcohol of 8 to 20% (v/v). These calculations were made by using a standard equation (13). The percentage (v/v) of alcohol in the must at time of fortification was estimated as 48% of the drop in density, in 'Brix. The fortified samples were analyzed for 'Brix, concentration of reducing sugar, soluble solids (extract plus sugar), and alcohol by procedures given by Amerine (1). After storage for 1 week in full glass containers at 13 C, the wines were racked and analyzed. After at least one additional month of storage, the samples were again analyzed. Delle unit determination. Delle units required for stability were calculated from sugar and alcohol concentrations of wines in which fermentation was completely inhibited after fortification. It was difficult to obtain precise analyses of the samples immediately after fortification because of gassiness resulting from high concentrations of carbon dioxide. Accordingly, it was difficult to determine, by comparison of the initial and final data, which samples had been made stable at the time of fortification. In practice, the samples which had been made stable could be determined from the final analyses alone. For example, Fig. 1 shows typical data obtained with a series of wines which had been fortified to give calculated concentrations of alcohol from 10.5 to 18% (v/v), in 1.5% increments, and then stored. Clearly, in the example, the samples with greater than 15% alcohol at fortification were stable. Taking 15% (v/v) alcohol and 13% (w/v) sugar as minimum concentrations for stability in this example, the minimum Delle units for stability were: DU = a + 4.5 c = 13 + 4.5 (15) = 80.5. Samples with less than 15% alcohol had had some fermentation after alcohol addition; these samples, compared with those that had 15% alcohol or greater immediately after fortification, had decreases in 'Brix and in concentrations of sugar and soluble solids after storage. Also, there was an increase in alcohol content resulting from the continued fermentation. vs In 15 0 z0 20 - z u 10 z 0 z U *u 5 I. IOY2 12 13Y 15 16/2 18 PERCENTAGE OF ALCOHOL DESIRED AFTER FORTIFICATION FIG. 1. Effect of alcohol concentration obtained at fortification on final concentrations of alcohol (%, v/v, A), soluble solids (%, wlw, sugar plus extract, A), reducing sugar (%, w/v, 0) and density ('Brix, 0). Arrow indicates level of alcohol, at fortification, required to obtain stability. At this concentration of sugar and alcohol, DU 80.5. = Samples with 15% alcohol, or greater, after fortification showed alcohol concentrations corresponding to the calculated amount of alcohol. In the samples with greater than 15% alcohol, even though they were stable, there was a slight decrease in soluble solids and reducing sugar from dilution by the larger amount of alcohol added. This also caused a large drop in 'Brix in these samples because of the effect of alcohol on the density. Often the shape of the curves indicated that a sample with an alcohol concentration midway between those actually obtained would have been stable. In these cases, the amounts of alcohol and sugar were estimated from this midpoint. Thus, the precision of selection of sample could be as erroneous as onehalf of the alcohol addition increment, and the error in DU could be as much as Y2 X 1.5 X 4.5 = 3.4.

VOL. 16, 1968 SUGAR AND ALCOHOL STABILIZATION OF YEAST 1069 Determination of volume of spirits addition for a given Delle value. In the Delle equation, a, the percentage (w/v) of sugar after spirits addition, can be substituted by the percentage of sugar before fortification to give: a'[v/(v + X)] + 4.5 c = DU, where a' is the percentage (w/v) of sugar before spirits addition, V is the volume of fermenting must, and X is the volume of spirits to be added. (As is usual, the volume contraction resulting from the mixing of alcohol and water is considered negligible.) When a' and V are known, X can be determined in terms of c, the percentage (v/v) of alcohol after fortification, for any given DU value. Now c is identical to C in the fortification equation (13): X = [V(C - A)]/ (B - C), where the percentage (v/v) of alcohol in the fermenting must is A and in the high-proof spirits is B. Using these two equations with two unknowns (X and C c), one can determine the volume of spirits to add to give a desired Delle value; however, these calculations involve solving a quadratic equation, and they are unnecessarily complicated and awkward. In practice, the volume of spirits to be added, X, can first be estimated from previous fortifications to give an approximate value for c in the modified Delle equation (above). This value can then be substituted as C in the fortification equation for a precise evaluation of X. An example of the calculations is given in Appendix I. Alcohol and sugar tolerance tests. Alcohol and glucose were added to grape juice (Semillon, 1964; 23.0 'Brix; ph 3.46; total acid, 0.49%, as tartaric acid; 150 mg of SO2 per liter; stored at 0 C) as follows: (i) 15 ml of 95% (v/v) ethyl alcohol was added aseptically to 200-ml volumes of autoclaved grape juice to give a final concentration of about 7% ethyl alcohol; or (ii) glucose was added to grape juice to raise the 'Brix from 23 to 32, and this mixture was divided into 200-ml lots. Yeast and bacteria were inactivated in the latter samples by addition of 1 g of diethyl pyrocarbonate (Chas. Pfizer and Co.) per liter and by pasteurization in flowing steam for 20 min. Controls were 200-ml samples of autoclaved grape juice. The solutions were inoculated with 1 ml of starter cultures of various yeasts and incubated at 25 C. (The starter cultures were prepared by inoculation with a loop from a slant into 30 ml of autoclaved grape juice and incubation at room temperature for several days.) The 'Brix was determined at the cessation of gas production and again after an additional 2 weeks. Other analyses. Nitrogen and aldehyde concentrations were determined by methods given by Amerine (1). Fusel oil (higher alcohols) was determined colorimetrically (11). Yeast cell concentration was determined microscopically with the aid of a Levy- Hausser counting chamber. Turbidity was determined as absorbancy at 650 m,u in a Spectronic-20 colorimeter; the high wavelength was used to minimize variations in absorbancy resulting from differences in wine color. RESULTS Effect of yeast strain. To discover which wine yeasts might be especially sensitive to alcohol and sugar, several strains were tested for their sensitivity to alcohol and to sugar. [We have already shown (2) that the Montrachet and Champagne strains of the UCD Enology yeast collection have the same DU values for inhibition, but that the Burgundy strain requires a higher value under the same set of conditions.] The yeasts which were most sensitive to initially high concentrations of alcohol were Distillers' and Tokay strains of S. cerevisiae var. ellipsoideus (Table 2). Those most sensitive to initially high concentrations of sugar were the Tokay strain and the TPF strain of Schizosaccharomyces acidodevoratus. Accordingly, these three yeasts, and the Montrachet strain as control, were tested for their DU value for stability in sweet wine production. Starter cultures were added to about 70 liters of a mixture of Chardonnay and Pinot blanc juices (see Table 1); 10% starter was used in all cases except with Montrachet (for which 1% starter was used). Figure 2 shows the change of 'Brix of the four fermentations, with time. The DU values required for stability of the yeasts are given in Table 3. The Tokay strain required the lowest DU value for stability. Effect of SO2. Sulfur dioxide inhibits yeast fermentation and might have an effect on the Delle value needed for stability. In the above experiments with Chardonnay and Pinot blanc grapes, one sample of must was fermented with the addition of 40 mg of S02 rather than 125 mg per liter. The Delle units required for stability, with the Montrachet strain of yeast, are given in Table 4. For the wine with low SO2, a somewhat higher Delle value was required for stability, but TABLE 2. Sensitivity of various yeasts to alcohol and sugar Yeast Final 'Brix c High High Species Strain Con sugarb Saccharomyces Burgundy -2.0 3.0 14.0 cerevisiae var. Bordeaux -1.5 6.0 16.4 ellipsoideus Distillers' -1.4 10.3 16.8 Montrachet-2.0 3.2 12.3 Tokay 5.4 16.4 21.4 Schizosaccharo- -2.0 4.4 6.0 myces pombe Schizosaccharo- TPFC -2.0 4.5 17.5 myces acidodevoratus Initially about 7% (v/v) ethyl alcohol. b Initially 32 Brix. c From Long Ashton Agricultural and Horticultural Research Station, Univ. of Bristol (England).

17KUNKEE AND AMERINE 1070 APPL. MICROBIOL. 20 1 0 2 3 4 Day 5 6 7 8 FIG. 2. Fermentation rates with several yeast strains. Density of solution, in 'Brix, is plotted against days. Montrachet time, in (0), Distillers' (A), and Tokay (0) are strains ofsaccharomyces cerevisiae var. Schiz. ellipsoideus. is Schizosaccharomyces acidodevoratus TPF (A). Arrows indicate times ofsample removal Numbers in parentheses forfortification. refer to yeast cell concentration after inoculation. TABLE 3. Effect of yeast strain on Delle value TABLE 4. Effect of S02 on Delle value required required for stability for stability Yeast Montrachet Distillers' Tokay Schizosaccharomyces acidode'oratus 'Brix ati fortifi- Sugar, AlcoholD cationa % Del_ (w/v) (v'/v) units 20.1 15.1 20.0 15.3 20.0 15.0 19.9 15.0 16.1 12.9 15.5 13.1 16.2 11.8 12.2 15.2 12.2 14.2 12.2 12.9 15.5 13.9 12.1 '17.3 71.0 81.4 70.5 77.4 71.3 70.0 78.0 90.0 a Wine spirits of 1870 proof were used for fortification. the differences were not much greater than the precision of the determination. Most of the SO2 in both samples was probably in the bound form at the time of the fortification (16), making the inhibiting activity at those times very slight, regardless of the amount of SO2 added. Effect of fractional fortification. Fractional fortification is sometimes used in production of sweet wines (17); the spirits addition is made at several times during the fermentation, with each addition bringing about a greater and greater inhibition of fermentation until it is S0 Brix ata SOs fortifications Sugar, Alcohol, Delle % (w/v) % (v/v) units mg/liter 40 19.5 15.1 15.4 13.0 13.1 15.7 74.6 83.9 125 20.1 16.1 12.2 71.0 15.1 12.9 15.2 81.4 a Wine spirits of 1870 proof were used for fortification. finally stopped. The effect of fractional fortification on sugar-alcohol stabilization was determined by fortification of Semillon must (Table 1). Samples were removed for Delle unit determination at time of initial addition of spirits. The remainder of the fermenting must was divided into four samples, of about 60 liters each, which were fortified to give an added amount of alcohol of 0, 1, 2, and 3% (v/v). After the four samples had showed a loss in 4.50 Brix, each was again tested for Delle unit value required for stability. The Delle value for stability in these wines was approximately the same, ranging from 80.1 to 80.5 (Table 5). [The desired Brix at time of second fortification was not precisely 4.50 lower in each case than that at time of first fortification. The differences (Table 5) reflect

VOL. 16, 1968 SUGAR AND ALCOHOL STABILIZATION OF YEAST 1071 (i) an overall drop of about 1.10 Brix occurring during the time of the first fortification and division of samples (approximately 1 hr), and (ii) a drop of 0, 0.4, 0.8, and 1.20 Brix, in the respective samples, resulting both from dilution by the addition of spirits and from the effect of the alcohol on the density of the solution. The time required for these reductions in 'Brix were 12.0, 14.5, 17.0, and 18.0 hr, respectively.] Kinds of spirits. Materials in high-proof wine spirits other than ethyl alcohol sometimes are inhibitory to fermentation rate (12) and may have an important effect on amount of alcohol needed for stability in the presence of sugar. Accordingly, comparisons were made between wine spirits (1880 proof) and laboratory ethyl alcohol (1900 proof) addition. Also, we determined the effect of wine spirits to which had been added acetaldehyde (100 mg/liter) and fusel oil (200 mg/liter). The fusel oil addition was a 1:7 mixture of commercial isobutyl alcohol and commercial isoamyl alcohol; the latter contained about 20% active amyl alcohol. The acetaldehyde and fusel oil content, determined analytically, of the high-proof alcohol before and after additions are given in Table 6. The results with Sauvignon blanc must (Table 1) showed an effect when wine spirits were used (Table 6); a lower Delle value was needed for stability as compared with ethyl alcohol as the fortification material. The amounts of acetaldehyde and fusel oils added had little effect on stabilization. Effect of clarification. Settled, nonsettled and filtered juices were compared for Delle value requirement. Settled juice was stored for 1 week in the cold. It was then filtered with use of filter aid and asbestos pads. No pectinase was added and several changes of filter pads were required to obtain 45 liters of juice. The Delle values for settled, unsettled, and filtered Sauvignon blanc juices (Table 1) were identical within the range of the precision of the determinations (Table 7). Sensory evaluation of stabilization methods. Sweet wines were prepared by the conventional California method to have at least 19.5% (v/v) alcohol and by the Delle method (whereby the lowest concentration of alcohol required for stability is used) in order to compare their sensory qualities. For these tests, two 100-liter samples of fermenting must were removed at three different times from a tank containing a fermenting mixture of Semillon and French Colombard musts. At each removal, one sample was fortified to give a calculated final concentration of 20% (v/v) alcohol, and the other sample was fortified to give a final concentration of alcohol calculated for a certain Delle value. For the latter, the Delle value required for stability was estimated from the previous experiment with Semillon. When fortified at about 200 Brix, the Semillon must required 72.6 DU for stability (Table 5). To be assured of stability in the present experiment, we increased the Delle value to 76. In previous work (2), we found the Delle value required for stability was about 4 units greater when fortified at 160 Brix and about 7 units greater at 120 Brix than that required at 200 Brix. Accordingly, we attempted to fortify the must at 200, 160, and 120 Brix to give Delle values of 76, 80, and 83, respectively. In another series, the musts were fortified to give a final alcohol concentration of 20% (v/v). The procedure for determination of amount of alcohol to add and TABLE 5. Effect offractional fortification on Delle value required for stability r- Alcohol _ added at 1st Fortifi- 'Brix at fortification, cation fortification' Sugar, Alcohol, % (v/v) Delle, % units (W/V) (v/v) 1st 19.4 15.1 12.8 72.6 0 2nd 14.3 11.6 15.4 81.1 1 13.9 12.0 15.4 81.5 2 13.5 11.6 15.2 80.1 3 13.2 11.2 15.5 81.2 Wine spirits of 1860 proof were used for fortification. TABLE 6. Effect of kinds of spirits on Delle value required for stability Fortification alcohol 'Brix at fortification Sugar, Alcohol, Delle Type Proof Aldehyde Fusel oil % (W/V) % (v/v) units mg/liter mg/liter Ethyl alcohol 1900 NDa ND 19.6 17.4 13.7 79.2 High-proof 1880 250 215 19.5 16.8 13.0 75.4 High-proof 188 295 365 19.3 16.0 12.9 74.3 Not determined.

1072 KUNKEE AND AMERINE APPL. MICROBIOL. an example of this procedure are given in Materials and Methods and Appendix I. (The data used for these fortifications and the values obtained are given in Appendix II.) After storage in glass for 2 months, these six wines were filtered and transferred to new, alkali-treated, oak barrels (capacity, 56.8 liters). The barrels were maintained completely filled. After 3 months in wood, the wines were subjected to sensory analysis. At this time, the free SO2 was found to be about 10 mg/liter in each wine, and an addition of 75 mg (per liter) of S02 was made for each. After a total of 9 months in wood, the wines were bottled in glass and corked. Two months later the wines were tasted again. For the first tasting, 5 months after fortification, an experienced taste panel of five judges was used (3). The sensory evaluations (scores) are given in Table 8. The differences in sensory scores for each pair of wines fortified at the same time by two different fortification procedures were analyzed for statistical significance (3) The Student's t values are also given in Table 8- TABLE 7. Effect of clarification of must on Delle value for stability Treatment N' fortificationb Sugar, Alcohol, Delle % % unt (W/V) (V/V) nt Not settled 75 20.1 15.8 13.7 77.4 Settled 76 19.4 16.8 13.0 75.4 Filtered 56 16.6 13.8 14.5 76.7 a As milligrams of (NH4)2SO4 per liter. b Wine spirits of 188' proof were used for fortification. In all instances, the judges placed a greater value on the wines fortified by the Delle procedure, and, in the case of fortification very early during fermentation, the greater value was statistically significant. In the second tasting, 13 months after preparation, the tasting was held on an informal basis with 30 members of the California wine and allied industries as judges. The samples were coded, but the conditions of judging were not as rigorous as those for the first tasting. Again, in all cases, the average scores of wines prepared by the Delle method were greater than those by the conventional procedure. All of these wines prepared for tasting were microbiologically stable. After 2 years of storage, 250 ml of each wine was filtered through a sterile membrane filter (Millipore; 0.45-, pore size), and the membranes were placed on nutrient medium. No yeast or bacterial colony formation occurred. Furthermore, there was very little increase in the volatile acidity of the wines during this storage period (Table 8). Turbidity measurements of the 2-year-old wines were practically identical for each of the wines (Table 8). Reinoculation of the wines with Montrachet yeast brought about no yeast growth in the 2-year-old wines after several weeks of storage at room temperature. Casual tasting of the wines showed them to have remained fresh and sound. DIscUSSION It has been established that tolerance to sugar and to alcohol is variable from yeast strain to yeast strain (8, 9). Furthermore, the degree of tolerance to each of these materials in a single strain can change depending on the age of the cultures and the previous conditions of growth TABLE 8. Sensory evaluation, volatile acidity, and turbidity of Delle and conventional wines Brix at fortification Determination 200 160 12 Delle Conven- Delle Conven- Delle Conventional tional tional Taste panel score (mean). 15.0 13.8 14.0 12.6 13.8 13.6 Informal tasting score (mean). 12.0 9.8 11.6 8.8 10.4 8.0 Volatile acidityb After 1 year.0.042 0.033 0.036 0.034 0.036 0.039 After 2 years.0.062 0.045 0.047 0.043 0.048 0.049 Turbidity' After 2 years... 0.022 0.020 0.025 0.020 0.022 0.023 Five judges. Scale: 1 (low) to 20 (high). The Student's t value between pairs at 20' was 3.21 (significant at 95% level of probability); at 16', 2.34; at 12', 0.53. b As grams of acetic acid per 100 ml. Absorbancy at 650 mg.

VOL. 16, 1968 SUGAR AND ALCOHOL STABILIZATION OF YEAST 1073 (9, 10, 14). Considering differences of DU > 3.4 as significant, we have confirmed previous results (2) showing that the minimum Delle value required for stability was dependent on time of fortification and yeast strain. Lowest Delle values were required when fortification took place near the beginning of fermentation and when a Tokay strain of wine yeast was used for fermentation. The change in Delle value required at different times during the fermentation is probably even greater than it appears to be. The temperature of the fermenting must, which was not controlled, was less (22 to 24 C) at the time of early fortification than those made later (24 to 26 C). Since alcohol has a greater inhibitory effect at higher temperatures (15), a lower Delle value is also probably required at higher temperature. Thus, from considerations of temperature, a lower Delle value rather than a higher value should be required later in the fermentation. (The rise in temperature during fortification resulting from mixing of alcohol and water may also have an effect on stability; these temperatures were not recorded.) Ough (15) has shown that, at 21 C and ph 3.5, the inhibitory effect on fermentation rate is linearly dependent on concentration of alcohol present initially. One might expect that fractional fortification might require less total wine spirits, the initial fortification rendering the yeasts more susceptible to a second addition, in the presence of sugar, as compared to the addition of the same amount of spirits at one time. Adaptation of yeast to the inhibitory effect of alcohol might be thought to be responsible for the change in minimum Delle unit value required for stability during fermentation. However, this ought to be reflected, but was not, in a change in Delle value with fractional fortification (Table 5). More likely, the inhibitory effect of the sugar in the presence of alcohol is not linear over a wide range of sugar concentrations, just as it is not linear over a wide range in the absence of alcohol. The loss of suspended solids, which influence aeration (5), or of nitrogenous material by settling or filtration had surprisingly little effect on stability. The empirical procedure for predetermination of desired Delle value, as given in Materials and Methods, seems to be valid. Appendix II shows that the Delle values obtained in the first two fortifications of wines for tasting were nearly precisely that desired (75.0 cf. 76.0; 80.5 cf. 80.0). The last fortification showed a discrepancy (DU 89.2 cf. 83). This reflects the error in time of fortification. The calculations in this case were made for a fortification at 120 Brix; in fact, the fortification was made at 11.50. Although the Delle wines were shown to resist a second fermentation by yeast, they still must be considered potentially bacteriologically unstable. One of these wines had less alcohol (12.8%) than many table wines. Not unexpectedly, this wine showed a slightly greater increase in volatile acidity on storage than did the wines with higher alcohol concentrations (Table 8), although the increase was not large enough for concern. Under proper conditions of storage, there was no problem with either lactic acid or acetic acid bacteria spoilage. Preparation of sweet wines by the Delle method, then, is quite practical wherever legal. The amount of spirits to be added is about half that used for conventional fortification in California. Not only is the Delle method more economic, but the resulting wines are most pleasant-tasting. Furthermore, sweet wines which can be legally classified as table wines (<14% alcohol) can also be prepared and made stable by this procedure. ACKNOwLEDGMENTS We thank R. W. Goswell of John Harvey and Sons, Ltd. (Bristol, England), for supplying the Schizosaccharomyces acidodevoratus yeast. We also thank H. C. Brenner and W. A. Winton for assistance in the winery operations, M. M. Morris for many of the analyses, and E. A. Crowell for aldehyde and fusel oil determinations. APPENDix I Calculations for determination of volume of spirits addition for a given Delle value. As an example of the calculations used for determination of volume of spirits addition for a desired Delle value, the calculations used for the last experiment reported in this paper were as follows. It was desired to determine the volume of wine [188.40 proof = 94.2% (v/v)] needed to add spirits to 100 liters of fermenting must at 20 Brix to get a Delle value of 76 in the fortified wine. The must was initially 26.10 Brix and had 25.7% (w/v) reducing sugar. The concentration of alcohol in the must at time of fortification, A, was estimated as 26.1-20.00 Brix X 48% = 2.9% (v/v). For evaluation of X, the percentage (w/v) of sugar in the fermenting must, a', can be estimated from the percentage (w/v) of sugar in the original must (determined analytically) and from the projected drop in Brix-from the initial 0Brix to that at time of fortification-in the following manner. The drop in Brix is first corrected for change in density resulting from the presence of alcohol by subtraction of 0.20 Brix for each 1% (v/v) alcohol present (as estimated above). The 0Brix is then converted from g/100 g to g/100 ml by multiplication by the specific gravity at that Brix [see Table 3 of Amerine (1)]. This final value is then subtracted from the initial percentage (w/v) of reducing sugar in the must to give the percentage (w/v) of sugar at time of fortification. Thus, the corrected reduction in Brix needed for estimation

1074 KUNKEE AND AMERINE APPL. MICROBIOL. of sugar concentration at time of fortification was (26.1-20.0) - 0.2 (A) = 6.1-0.2 (2.9) = 5.5' Brix. This is equivalent to 5.5' X 1.082 g/ml (density at 200 Brix) = 6.0% (w/v). From this, the sugar concentration at time of fortification was estimated as 25.7-6.0 = 19.7%7 (w/v) = a'. From previous information, the volume of alcohol, V, should be approximately 12 liters. Substitution in the modified Delle equation gave: 19.7 [100/(100 + 12) ] + 4.5c = 76. Solving for c gave 13.0. Substitution in the fortification equation gave: X = 100 (13.0-2.9)/(94.2-13.0) = 12.5 liters. Thus 12.5 liters of spirits was to be added to each 100 liters of fermenting must to give a final DU of 76. The results of the calculations for each fortification for the last experiment are given in Appendix II. 6. Delle, P. N. 1911. The influence of must concentration on the fermentation and composition of wine and its stability. Odessa, Otchet "vinodiel'cheskoi stantsii russkikh" vinogradarei i vinodielov" za 1908 i 1909 g., p. 118-160. 7. Ghimicescu, G., and S. Freund. 1963. Nomogramme pour calculer l'indice de stabilite des vins. Bull. Office Intern. Vin 36:611-613. 8. Gray, W. D. 1941. Studies on the alcohol tolerance of yeasts. J. Bacteriol. 42:561-574. 9. Gray, W. D. 1945. The sugar tolerance of four strains of distillers' yeast. J. Bacteriol. 49:445-452. 10. Gray, W. D. 1946. The acclimatization of yeast to high concentrations of glucose: the subse- APPENDIX II Estimation of spirits volume required for Delle and conventional fortifications Brix desired at fortification Determination 200 160 12 Delle Conven- Delle Conven- Delle Conventional tional tional Delle units desired... 76 80 83 Estimateda Before fortification Alcohol (%, v/v)... 2.9 4.8 6.8 Sugar (%, w/v)... 19.7 16.0 12.4 After fortification Sugar (%, w/v)... 17.6 14.3 11.1 Alcohol (%, v/v)... 13.0 20.0 14.6 20.0 16.2 20.0 Spirits to addb... 12.5 23.1 12.3 20.5 12.0 17.8 Actual Brix at fortification... 20.1 20.1 16.3 16.3 11.5 11.5 After fortification Sugar (%, w/v)... 17.4 16.0 14.0 12.5 10.0 9.6 Alcohol (%, v/v)...... 12.8 19.5 14.8 19.7 17.6 21.3 Delle units... 75.0 80.5 89.2 a Estimated for must initially 26.1 Brix with 25.7% (w/v) reducing sugar. I Volume of 188.40 proof spirits to add to 100 volumes of fermenting must. LITERATURE CITED 1. Amerine, M. A. 1965. Laboratory procedures for enologists. Univ. California, Davis. 2. Amerine, M. A., and R. E. Kunkee. 1965. Yeast stability tests on dessert wines. Vitis 5:187-194. 3. Amerine, M. A., E. B. Roessler, and F. Filipello. 1959. Modern sensory methods of evaluating wine. Hilgardia 28:477-567. 4. Cotea, V., and G. Rican. 1964. 0 noua nomograma de stabilire a unitatilor conservante Delle. Gradina, Via Livada 13(7):37-38. 5. Crowell, E. A., and J. F. Guymon. 1963. Influence of aeration and suspended material on higher alcohols, acetoin, and diacetyl during fermentation. Am. J. Enol. Viticult. 14:214-222. quent effect upon alcohol tolerance. J. Bacteriol. 52:703-709. 11. Guymon, J. F., J. L. Ingraham, and E. A. Crowell. 1961. Influence of aeration upon the formation of higher alcohols by yeasts. Am. J. Enol. Viticult. 12:60-66. 12. Guymon, J. F., and J. A. Nakagiri. 1957. Effect of acetaldehyde, acetal, and ethyl acetate upon alcoholic fermentation. Am. J. Enol. 8(1) :1-10. 13. Joslyn, M. A., and M. A. Amerine. 1964. Dessert, appetizer and related flavored wines. The technology of their production. Univ. Calif., Div. Agr. Sci., Berkeley, p. 207. 14. Nosiro, K., and K. Ouchi. 1962. Fermentation

VOL. 16, 1968 SUGAR AND ALCOHOL STABILIZATION OF YEAST 1075 activity of yeasts and its alcohol tolerance. I. Alcohol tolerance of the fermentation activity of yeasts. J. Soc. Brewing, Japan 57:824-830. 15. Ough, C. S. 1966. Fermentation rate of grape juice. III. Effects of initial ethyl alcohol, ph, and fermentation temperature. Am. J. Enol. Viticult. 17:74-81. 16. Paul, F. 1958. Ober den Acetaldehyd im Wein. Seine Entstehung wahrend der alkoholischen Garung. Mitt. Klosterneuberg Ser. A Rebe u. Wein 8:123-134. 17. Singleton, V. L., and J. F. Guymon. 1963. A test of fractional addition of wine spirits to red and white port wines. Am. J. Enol. Viticult. 14:129-136. 18. Stanescu, C. 1965. Disque pour le calcul de lindice de stabilisation biologique des vins. Bull. Office Intern. Vin 38:1428-1429.