By Dale Ims
Extract is the term used to describe the non-volatile components in a wine—basically, what’s left after the liquids have evaporated. It will include all the non-volatile, non-fermentable components that were present in the juice we started with unless the winemaking process has somehow removed them from solution or suspension. It will also include chemically altered derivatives of some of the components that were present in the juice as well as some intermediate-stage products of the complex, multi-step process by which the enzymes in the yeast we add to do the fermentation convert the sugars to alcohol. It will further include any leftover sugars which the yeast was unable to convert.
Typical values for wine extracts are < 25 gm/L for dry white wines and 25–30 gm/L for dry red wines. (Handbook of Enology, Volume 2: P. Ribereau-Gayon, et al, 2006). Boulton, et al (Principles and Practices of Winemaking) cite a 1982 paper (Winger) which found a mean value of 22 gm/L for the extract of dry white California wines. The composition of the extracts can be expected to vary with type of grape, growing season and fermentation conditions, but we can be sure that there are some pigments in the extracts of red wines, and that at least some of the organic acids and their soluble salts present in the juices will also be present in the wine. Glycerol (a by-product of the fermentation process) is stated to be present at the level of 5–8 gm/L in wines (Boulton, et al), where it contributes to the wines’ perceived sweetness and mouthfeel. Pentose sugars (arabinose and rhamnose, with 5 carbon atoms instead of the 6 in glucose and fructose) pass through the fermentation process unaffected and are likely present at the level of 0.4 to 2 gm/L. Any residual levels of the hexose sugars (glucose and fructose) are likely to be mainly fructose, as it is the more difficult to ferment.
A number of different techniques have been developed for measuring the extract value of a wine, and those values are of more than passing interest in Europe—especially in France where there are strict laws regarding wine production techniques, and a wine’s extract value is used as an indication of use of proper production methods. It would seem that a simple dry-down process would be the easiest and most direct method of determining wine extract values, but that technique is apparently seldom used. Another technique boils a measured volume of wine until the alcohol has all evaporated off, then adds water to restore the original volume and finally uses a hydrometer to measure the weight% solids in the sample. Still other techniques utilize specially prepared nomographs or tables to infer the level of extract from the samples’ alcohol contents and specific gravity readings.
The level of the extract in wines became a subject of some interest to me when I was looking into the origin of the formulas for predicting the potential alcohol of juices. Rational and reasonable expressions for calculating potential alcohol levels based on the sugar content of the juices include a correction of about 3°Brix for non-fermentable components, and I thought that a careful look at the subject of extract might help to resolve any questions relative to the value of that correction term. Having learned a bit more about the composition of the extract, and how it can change during the fermentation, it became clear that the level of extract was going to be of little help in quantifying the level of non-fermentables in the juice.
A literature search revealed that while the non-fermentable components in juices have roughly the same concentrations as the extracts of finished wines, at least some of the components present in the juices are lost and some components of the extract are formed during the fermentation process, so the level of extract is not a reliable proof of the correction term used in the equation used to predict potential alcohol.
Regardless of the utility of the extract data for my intended purpose, I thought that the results of some of my calculations relative to this topic were of some interest, and decided to write them up in the hopes that the other members of RAHW will find them of interest, too. The calculations were done to look at the effects of the alcohol content and the residual sugar level on the specific gravity of a finished wine. For the calculations I assumed the wine to be made up of only water (density = 0.9982 gm/cc), ethanol (density = 0.789 gm/cc), and sucrose (density = 1.62 gm/cc) even though we know that the extract is likely comprised of a number of different components with differing densities. Therefore, what we are calculating here is the residual-sugar-equivalent to the extract and combining it with the actual residual sugar.
For these calculations, it is imperative that we include the excess-volume corrections to the volumes, densities and specific gravities in order to get results at least close to reality. In fact, I found that I needed a more accurate basis for the excess volume calculation for this work, and found it in the form of a table of data in Appendix 6.2 of Wine Science: Principles and Applications by Ron Jackson. Perhaps not surprisingly, my calculations now reproduce exactly the specific gravity values in the above table for water/ethanol mixtures in the concentration range of interest to winemakers.
The calculation process was to assume we had wines with 8, 9, 10, 11 & 12 weight per cent ethanol, and that each of those wines could further have 0, 1, 2, 3, 4, …, 8 weight per cent sugar. For each of these 45 combinations, we calculated the volume % alcohol (ABV), the density of the wine, and finally its specific gravity. The results are shown in the graph below.
Figure 1: Results of calculations
Note that in the figure we have plotted specific gravity against vol% alcohol, and that the data points are organized according to the residual sugar level which—in our calculations—includes the extract. A first result of interest to me is that a “wine” with no residual sugar or extract (labeled as “Dry” in the figure) has a specific gravity of about 0.9840 if the alcohol concentration is 12 vol% and about 0.9820 at 14%! Most of the wines I make—which I assume to be dry, and with alcohol levels in the 12–14% range—end up with a specific gravity of about 0.9950. In order to have an SG that high, we can see from Figure 1 that we would need somewhere around 3% sugar/extract. So, even though we recognize that the non-fermentables for which we correct in the calculation for potential alcohol may not carry over unaffected through the fermentation process, we find that an approximately equal level (3%) of sugar/extract remains in order to get the observed final SG values. An example of a commercial wine which also seems to fit the chart in Figure 1 is Wagner’s Alta B. The Wagner website indicates that the alcohol level of Alta B is 11% and that the residual sugar content is 5%; I measured 1.016 as its SG, which is very close to what we would expect from Figure 1 for a wine with 8% (total) extract/sugar and 11% alcohol.
It should be noted that while extract levels may be in the range of an equivalent of 2.2% to 3% for grape wines, those made from herbs or flowers may have lower levels. One of Jill Misterka’s lilac wines finished with an SG of 0.986 and—based on the recipe—about 15.5% alcohol; from the chart in Figure 1, we estimate the extract to be about 1.5%.
Since we had some degree of success above in fitting a few wines to the chart in Figure 1, I was emboldened to consider the possibility that it might be possible to get reasonably-accurate estimates of alcohol content and residual sugar levels of our wines from their specific gravity measurements.
There are a number of options open to home winemakers who are interested in quantitative measurements of their wines’ properties. Accuvin makes a residual sugar tester similar in operation to those used to measure SO2 levels, but at about $3 per test, they are relatively expensive—especially for small batch sizes. A similar test can be done using Clinitest tablets at much more reasonable costs (about $0.50 per test), but red wines complicate the color-matching required to obtain the result. The only home-level tool for measuring the alcohol content of a wine is the Vinometer and similar devices which use the fact that the surface tension of a water/ethanol mixture varies with the composition of the mixture. However, Vinometer readings are unreliable when residual sugar is present in the wine.
It is clear from Figure 1 above that if all the information we have about a wine is its SG, we have a problem since different combinations of residual sugar/extract and alcohol content can give the same SG. We need more information in order to make this approach work, but as home winemakers, we have the ability to get that information without buying a new instrument, tool or test kit—assuming we have a hydrometer.
When I was researching the origin of the equation for potential alcohol, I consulted a number of winemaking books, and in at least one of those I remembered finding a simple formula for calculating the alcohol content of a wine from the change in SG of the juice as it was converted to wine. An approach like that made sense to me, and some time ago I built a simple spreadsheet calculator which utilized the SG of the wine in question, the SG of a dry wine, and the total change in SG expected (juice to finish) if the fermentation had gone to completion to estimate the residual sugar of my wines. At the time I reasoned out the structure of the calculator, I was unsure of the form of the relationship between the sugar content of the must and its SG, but I assumed it to be linear. While doing the calculations shown in Figure 1, it occurred to me that I could also do the calculations required to test the linearity of the sugar content vs SG relationship as a fermentation proceeds, and also determine how the SG change during fermentation is related to the alcohol content.
To do this new round of calculations, we again used the spreadsheet approach with incremental quantities of sugar converted to alcohol in each step, with the quantities and concentrations of each component—as well as the specific gravity—calculated at each step. We started with solutions of sugar in water at 17, 18, 19,…, 25 Brix and followed the calculations until all the sugar was gone (no extract here). The results are shown in the chart in Figure 2, which is included here largely to illustrate that the linearity is pretty good (i.e. that the lines are pretty straight).
Figure 2: Calculated Specific Gravity vs Volume % Alcohol
Actually, a careful examination of the individual curves on the Figure 2 plot show that there is a little curvature in each (very little!), and that there is also a small but systematic variation in slope: from -0.00745 for the 17 Brix juice to -0.00759 for the 25 Brix. A good mean for the slope is -0.00752 (units)/Vol%, and the reciprocal of that is 133 Vol%/(unit). This suggests that if we multiply 133 times the change in SG from juice to finished wine, we will get the Volume % alcohol in the wine. This would imply that a juice which started out with an SG of 1.100 (approximately 24° Brix) and which produced a wine with a final SG of 0.995 would have an alcohol content of:
Volume% = 133 * (1.100 – 0.995) = 14.0%
By chance, after completing the above calculations, I found the abbreviated notes I had taken from the reference mentioned above relative to the calculation of a wine’s alcohol content from the change in SG through fermentation. Apparently that reference was in a book by CJJ Berry, and his equation was:
ABV = 131 * (Starting SG – Ending SG)
That is pretty close to my result: 131 for his multiplier vs 133 for mine. I would probably be willing to agree to a multiplier value of 132.
Assembling all the pieces above, we arrive at the following process for estimating the residual sugar and alcohol contents of a wine.
We found that a set of SG calculations based on model wines produced results in general agreement with wines of known compositions, and that with additional information from the literature relative to extract levels as well as estimates of ABV from changes in SG during fermentation, we could develop a process to allow meaningful estimates of ABV and residual sugar of wines using only hydrometer measurements. This process may prove useful to those of us who would like to know something more about our wines with no additional investment in tools or technology.
Note: If you would like to test the above process but have only Brix data for the starting condition, the following conversion is accurate to 3 places over the range of 17–26 °Brix:
S.G. = 0.00448 * °Brix + 0.9935
See a Quick Reference Guide for Estimating ABV and Residual Sugar that is based on the research from this article.