"a professional winemaker here from BC mentioned that they were getting higher alcohol levels for the starting Brix than they used to, often by 1% or even more. She said this was confirmed by other winemakers from the area. Her hypothesis was that the yeast were getting more efficient in alcohol production. This could have large repercussions if that were indeed the case. I'm wondering if this is happening in other, hotter areas, as well?"
In addition to wine I make a lot of beer. I use Wyeast liquid yeast and prepare a starter for my 6 gallon batches. Once fermentation is complete I reuse the yeast for another batch. I do this again for a third batch. To reuse the yeast I pour the new batch of wort onto the yeast cake from the previouse batch. There is a lot more yeast present for the second and third batches. These batches always ferment down lower than the first batch, thus creating more alcohol. Makes me wonder if the winemaker mentioned above is using variable amounts of yeast and perhaps that is the reason for the increased alcohol. Beer never ferments to zero like wine. Seems like the only way you can end up with more alcohol in wine is to start with more fermentables.
Frederick, It is easy to find references to respected authors that use the equations. Can you provide references to respected authors that publish information indicating that these equations are wrong?
Only the winemaker could answer this for sure, but I doubt she's making simple errors like this. She's one of the best commercial winemakers in BC, with a degree from UC Davis. As for lab results, I would think she's doing that already but that's just a guess.
As for the PA definition, it sounds reasonable, but I'm wondering what formula is being used to equate 22Brix with 12PA? Unless we know that we can't be sure the hydrometer PA scale is correct. Going back to Lum's post about alcohol yield from sugar, we should be able to calculate theoretical maximum PA for a given Brix - I think I saw the result in Margalit's Concepts in Wine Chemistry, but I'm waiting to get it from library, so can't answer that right now. Anw, I'm willing to bet that value is higher than 12PA.
I have some info sheets at home from mostly Napa wineries with Brix, TA and final alcohol numbers, I'll try to look those up to bring some more data into this discussion.
We have _lots_ of experts in this group, and these experts have access to a wide variety of references. So why don't we just ask the folks here _point blank_, and see what they can come up with.
Earlier, Andy wrote: "...My literature from UC Davis says (paraphrasing):
The theoretic maximum yield of ethanol is around .6 times the initial Brix. This would give a maximum yield of 13.2 ABV for a 22 Brix must. My reference goes on to say that .55 times the initial brix is really all we can get in practice, which would yield around 12.1 ABV. This is because a varying percentage of the sugar is used for other things and even if the fermentation does not stick, there is a percentage of sugar that ends up as other end products (like glycerol, pyruvate, acetate, acetaldehyde)...."
**Question to the group: Does _anyone_ out there have any modern reference that_contradicts_what UC Davis has to say on this subject in any substantive way ????? TIA
Alright, here is how the hydrometer PA scale does against some real data. The numbers are from info sheets provided by Napa, Sonoma, and 1 Washington state winery. The wines are marked R (red) and W (white) as this is significant.
The columns are: Brix, final alcohol, PA per hydrometer, delta (hydrometer - actual).
The PA per hydrometer was calculated by B * (12/22) = B * 0.545 (as
22B = 12PA).
The table is ordered by the delta.
Where B is given as a range, avg value was taken for simplicity.
Type B Final PA Delta R 25.5 13.5 13.9 +0.4 R 26-27 14.5 14.5 0 W 25.6 14.2 14 -0.2 W 25.2 14 13.8 -0.2 R 24.8 13.9 13.6 -0.3 R 24-25 13.7 13.4 -0.3 R 26.8 15 14.6 -0.4 R 24.2 13.6 13.2 -0.4 R 26.4 15 14.4 -0.6 W 25 14.5 13.7 -0.8 W 23.8 13.9 13 -0.9 W 23.5 13.9 12.8 -1.1 W 23.5-25 14.5 13.3 -1.2 W 23.5 14.2 12.8 -1.4
Average delta all: -0.53; reds: -0.23; whites: -0.83
I'd say not too good overall, particularly for whites. The higher alcohol in whites vs. reds makes sense and is well supported in literature. But the formula systematically underestimates even reds in this sample. Out of 14 wines, 6 are outside of the reasonable +-0.5 delta, with the worst case being -1.4 (that's Napa, Matanzas Creek Sauv B 2001).
Based on this, I think I could argue that the PA scale on all of the world's hydrometers leaves something to be desired.
Would be interesting to see some data from Brix values closer to 22 Brix. Do you have or could you get data in this area. 25+ Brix values are not representative of grapes grown in the East. I am wondering if the error might not be non linear and may in fact be opposite for lower Brix juice.
I would be careful about using data sheets like this too. Even wine labels can have an inacurate ABV listed on them. For all the crap the US government puts wine makers through concerning labels, they allow a pretty big margin of error when it comes to alcohol content.
The data sheets on many web sites are only general information on the parameters. I'm sure detailed data sheets are kept on each fermentation, but the stuff published is generalized for public consumption. Some times the brix values are nothing more than the last field tests or sampling tests conducted during the crush.
Ray,
You may have posted it before and I missed it, but care to share your calculation for (estimated) alcohol content.
I can vouch for that. Last year I worked part time at a small commercial winery in Northern Virginia. The winemaker would "SWAG" (Scientific Wild Ass Guess" the alcohol content based on the initial brix at crush for each variety.
I doubt seriously if many or any of the wineries in that area (Northern Virginia) actually do a laboratory test for the finished alcohol content but that does not stop them from putting a number on the label.
They are also not checked for a log of sprayings and a log of the fermentations and additions for each wine. The winery I worked at was owned by a Pulmonary Doctor so one would think that detail and record keeping would be something with which he had familiarity.
I have heard that winery and vineyard records ARE checked in wineries out west. Anyone have any comments or experience in the record keeping and required testing for commercial wineries?
It definitely would be informative, but can't help you there, these were all the data I could find among my notes. However, the theory is that in colder climates the alcohol yield is better than in hotter climates, so if anything, the error should be even more pronounced for lower B values, more common in colder climates. But that's just a guess, no data to support that.
On your comments concerning the accuracy of the PA scale on the hydrometer. I think you will find that this is the same scale that is found in the common SG/Brix/Sugar/PA table that is published all over the place. Fred and I have been having an extended discussion of this off line. He has one view of it's meaning and I have another. Neither of us have found diffinative data showing which one is right. We are both looking for such now but it may take time. The data you include does bear on the discussion.
Fred maintains that the PA values in the table represent the maximum possible alcohol yield and anyone who uses or supports a calculation that gives values higher than these is wrong by deffinition. He has not produced any data to directly support this. He has arguements that do but they have not convinced me. I will leave it to Fred to argue his point.
If you look at many published accounts of the table the instructions suggest that if your SG drops below 1.000 you have to calculate the extra alcohol that is being generated. This suggests that the table is based on frementing, not to dryness, but to an ending SG of 1.000. This is the interpretation that I use. It is supported by other authors that look at the SG/PA relationship in different ways. But, in all honesty, I cannot say that I can prove my view any better than Fred can. All I have to go on is the comments of well respected authors who Fred discounts.
The calculation I use is that published by Duncan and Acton in Progressive Winemaking. PA = (G begining - G ending) / F Where G = 1000 * (SG - 1) = gravity and F = 7.75 - 3*(G begining - 7) / 800
The F term corrects for non-sugar solutes in the wine. Depending on the media you may need to adjust the value 7 up or down. It is interesting to note that Duncan and Action did not mension the common PA table at all but if you assume and ending SG of 1.000 you will get the values in the common table using thier equation. (See "The Unified Theory of Gravity" in April-May WineMaker Mag.)
I have really not found any data that proves that this is true but you will typically get 1% higher calculated alcohol useing this meathod over just looking up PA in the table. The numbers you posted do seem to indicate that you can get numbers higher than found in the common table. That is interesting. But I am not going to say that they prove my side of the arguement. I want a bit more before I claim that.
So, by Freds interpretation, he would say that the table is correct and the measurements you quote must be wrong. By my interpretation, I would expect you to get a higher alcohol level than given in the table if you ferement to dryness.
----------------------------- Paul wrote: "I am wondering if the error might not be non linear and may in fact be opposite for lower Brix juice."
IMHO: I suspect that the difference is linear as the other relations in the table are linear. What is probably not linear is error caused by using different yeast strains and differen media. Different yeast may be more or less efficient as converting sugar to alcohol and different media may have a different correction in the F term.
--------------- JEP wrote "I would be careful about using data sheets like this too. Even wine labels can have an inacurate ABV listed on them. For all the crap the US government puts wine makers through concerning labels, they allow a pretty big margin of error when it comes to alcohol content."
IMHO: Regardless of what is put on the labels, they do perform laboratory analysis on wines out west to determine the true alcohol level. This is for tax purposes. I have been in communication with a number of agencies, labs, and some wineries out there recently. The government is strict on them. Back east they seem to ignore the goverment requirements. One of these days the Feds are going to come down on them. I think the main reason for the difference in enforcment is that Cal. grapes are much higher sugar content and are at risk of making higher than 14% wines. The goverment taxes these at a higer level. East coast grapes are rarely in danger of having this problem.
I was thinking about the problem yesterday and reread your Winemaker article and it seems to me both of you might be right, just the basic assumption is wrong.
I would agree with Fred's definition of PA being the maximum potential alcohol once can get by fermenting to dryness (I hope I'm paraphrasing this ok?). But what's shown on the hydrometer is NOT the *theoretical* maximum PA bur rather an estimate of *practical* PA that one can normally achieve. Taking this number as maximum PA that one can achieve is incorrect, as demonstrated by the data that I sent in.
Going back to Lum's post - the maximum theoretical yield of alcohol form sugar is 51% by weight, which translates to 65% by volume. For
22B, this gives 14.3-14.4, which agrees dead on with Duncan and Acton. But this value can never be achieved, as some sugar (8-10%) is used to produce other things than alcohol, so a *practical* maximum is more around 60-59% by volume. We are of course more interested in the practical value, so this suggests the D&A numbers have to be adjusted downward, unless this is somehow built into their formula in that F coefficient?
Going back to the hydrometer PA, the 22B = 12% gives B * 0.545 = % alcohol as the underlying formula, so this is a significant down adjustment from 0.65. My beef was this was that we didn't know where this number came from. If this estimate is too low, then the scale systemically undervalues the practical PA, and then it's possible that one can get a better estimate by *including the final gravity* using this scale than just basing it on the initial gravity.
If, on the other hand, a different formula is used, the starting gravity might be good enough. Somebody was asking for modern references - Margalit in Concepts of Wine Technology (2nd edition) has
2 formulas: B * 0.57 and a variation of this - (B-2.1) * sg * 0.57. These incorporate the sugar loss for byproducts and other solids in must. He claims one of these - I think the second one - is +-0.2% on average from the finished measured value for B range 18-25. Both formulas should give a close result in this range. The B*0.57 formula would add about 0.5% to the hydrometer estimate - incidentally, if this was used against the data that I provided (deltas -0.83, -0.53,
-0.23), the estimate would have been much better (-0.3, 0, +0.2).
So, the point, which I still have some trouble articulating, is that depending on the formula one uses, a good estimate can be achieved either just from the starting gravity or one also has to include the finished gravity. With the hydrometer scale, it seems both of those options are about the same, with the first one underestimating and the other likely overestimating by about the same amount for a dry wine.
I think I'm done with this one, back to making wine!
I ran my numbers for a Chardonnay in long-term aging to see what PA the formula above reports. I must be doing something wrong because the answer I'm getting is an impossible PA. Can someone run-through how they're determining the answer based on my SG? Either the formula is wrong or I'm wrong. Considering what I do for a living, I lean on the latter; I'm NOT a mathematician! Rather, I was a Liberal Arts major. :-)
Beginning SG: 1.098 Ending SG: .998
The final alcohol rating I calculated with a hydrometer after fermentation completed was 13.75%.
I think we now have the answer to this question. It's what they used to call "the sound of silence". No one who actually understands this stuff is going to contradict what the research scientists at UC Davis have to say on this subject (except yourself).
We covered this material in our private discussion and our discussion ended several weeks ago when you refused to accept this work as a modern, authoritative reference. This same material has now been covered in both this thread and the thread titled "Planning a ferment", and you have once again ignored or rejected all arguments.
There is nothing more I can do for you, Ray. You will have to somehow work this out on your own.
No serious student of winemaking would consider this to be "real data". But I know that you are serious about this and I will try to give you a serious answer here. To this end, let me give you an easy way to evaluate this data for yourself.
Keep in mind that the maximum *theoretical* conversion rate is about 0.60, and the maximum *realistic* conversion rate is about 0.55. It is this realistic rate that we find on our hydrometers.
All you have to do is divide the end alcohol by the original BRIX to see how this data compares to reality.
In your fist example we would divide 13.5(ABV) by 25.5(BRIX) and get 0.529. This is slightly less than the expected 0.55 rate and would lead us to expect that a small amount of sugar was unconsumed and was left in the wine as "residual" sugar.
In your last example we would divide 14.2(ABV) by 23.5(BRIX) and get 0.6043. This exceeds even the theoretical maximum (even without _any_ losses) !! Obviously there is something seriously wrong with these numbers.
I might also point out that in at least 3 of these examples they are unsure of their original BRIX numbers, which automatically indicates that this "real data" isn't credible.
There is of course a much easier way to do this. Since the "realistic" calculations have already been done for us and appear in the PA scale on our hydrometers, simply compare the end alcohol to the original PA for that wine. If the end alcohol exceeds the predicted (potential) alcohol, you automatically know that something has gone wrong with your end alcohol calculation. HTMS, HTH
Ok, I _think_ I found what you want. Principles and Practices of Winemaking, by Boulton et al, Chapter 5, section 2 page 195. It's way too deep to go into detail here, but it's exactly what you and Fred are looking for. I have a copy and will sum up:
The French have done a ton of work on this and so have the Californians since the 50's. The bottom line seems to be that PA is nothing more than a ballpark value. The variety of wine must, area grown in and seasonal weather variations all impact alcohol conversion, not to mention fermentation practices.
Hera are a few formulas: Dujardin: (They make a wonderful ebulliometer too by the way) PA= 0.0595* [2.66*Oe- 31.8]
Ok, I _think_ I found what you want. Principles and Practices of Winemaking, by Boulton et al, Chapter 5, section 2 page 195. It's way too deep to go into detail here, but it's exactly what you and Fred are looking for. I have a copy and will sum up:
The French have done a ton of work on this and so have the Californians since the 50's. The bottom line seems to be that PA is nothing more than a ballpark value. The variety of wine must, area grown in and seasonal weather variations all impact alcohol conversion, not to mention fermentation practices.
Hera are a few formulas: Dujardin: (They make a wonderful ebulliometer too by the way) PA= 0.0595* [2.66*Oe- 31.8]
Here is what I was saying before about potential alcohol;
Dujardin (1913):
PA (V/V)= 0.0594*[2.66*Oe-31.8]
Where Oe = Oechsle [(s.g.(20/20C-1.000]*1000
Ribereau-Gayon (1975): PA (V/V)= 0.0595*[2560*((S.G 20/20 C)- 1)-22.2] These two come out pretty close to one another.
Californian Data (Bioletti (Marsh 1958):
PA (V/V) = 0.592*[(sugar w/w%)-3.0] (Huge difference)
From Ough, Amerine and Jones (1963-1985)" PA (V/V) a+b* brix
Where 'b' varies by year, cultivar and growing region and 'a' varies from -4.92 to + 4.37. I can't give you what a and b are, they are measured values that are not stated clearly (at least to me).
The bottom line is they saw mean ethanol yields vary from 0.665 to
0.588 from region I to region IV over 9 years. That's a variation on the order of 1.5% varied around 12.5% V/V.
The modulus 145 values of Baume roughly correlate to PA if you exclude nonsugar extract, a bad idea. The good news is its easy to measure that one in finished wine, but by then its a little too late. You could guesstimate at 2 or 3%, but that is what the other calculatins are compensating for and they got those values from somewhere.
I guess what I'm trying to say is as far as i am concerned Potential Alcohol is just a best guess. Too many factors can influence the final yield of a wine. It's hard to estimate the unfermentable dissolved solids and that will impact a hydrometer, period. That number seems to bounce around at least 1 % that's a big deal to me.
As to the prefered methods of analysis, ebulliometry and distillation/ hydrometry are really the only approved methods of measuring alcohol by the BATF. Ebulliometry is affected by sugar at 2% or greater, distillation/ hydrometry are impacted by excess SO2 (>200 mg/L) and excess acetic acid (>0.1%). As long as those issues are deal with, you can get accurate data to at least 0.3% which is good enough for BATF. (I can't speak to other countries regulations, I just don't know for sure.)
I have a chart of ebulliometric values once barometric pressure is compensated for from Wine Analysis and Production by Zoecklein et al that I am going to transfer to Excel as an FYI.
Our discussion ended several weeks ago when you refused to accept the UC Davis material as a modern, "respected" source. I covered this material with you, and that material has again been covered in both this thread and the thread titled "Planning a ferment". You continue to refuse to accept this. I can no longer help you,Ray. I said this then and I will repeat it here, your argument is no longer just with me but is now with the research scientists at UC Davis !! Please take your arguments up with *them*.
I am no longer looking for additional data. I am completely satisfied that _all_ of my previous answers to you were correct. I suggest you drop that D&A reference as the sole support of your theory, and proceed by using any and all _other_ references you can find.
Please stop misrepresenting what I have said. The PA values in the table represent the maximum _realistic_ amount of alcohol that can be expected if the wine goes to dryness. There is of course the _theoretical_ maximum, but *all* sources will tell you that this theoretical maximum is not achievable in any real world situation.
What do you call the UC Davis material ??
This is the false assumption that all of Ray's work is based on. The _only_ reason SG drops below 1.000 is because of the effects of the alcohol !! Alcohol is a NON-fermentable solute and contributes _nothing_ to the resulting end alcohol level.
Oh yes they did !! All of the information on the Duncan and Acton chart conforms to the "standard" tables with the exception of a grotesquely distorted PA column. Here are some examples taken from the D&A table: SG1.080...19.8Brix...12.8PA SG1.090...22.0Brix...14.4PA SG1.100...24.2Brix...16.0PA
I'm told that is document was originally published in the 1960s. If, after all this time, it has had no effect on modern thinking, I feel it is safe to assume that it has been rejected by modern science. I mean, does any one out there have a hydrometer that reflects the above PA information ?? BTW - I finally located where I can get a copy of this reference. I found it on AMAZON.COM. It is long out of print and is on sale there for $1.75(US). ;o)
DrinksForum website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.