Much more information is available on wine yeast than ale yeasts. While a mead maker can consult the Winemaking Handbook from Scott Labs to look up the nutrient requirement of most used wine yeasts, no similar bible exists for ale yeasts. I have personally contacted three ale yeast manufacturers, and they said that nutrient requirement information was not available for their ale yeasts. One example of the lack of data is the little-noticed but big change between TOSNA 2.0 and TOSNA 3.0, that reduced the recommended level of nutrients for ale yeast from high to low. Moreover, with the range of levels for wine yeasts its unfortunate that all ale yeasts get clumped together, an English ale yeast may have different nutrient requirements than a Kviek yeast in nutrient calculators. As nutrients are a major determinate of proper mead fermentation, mead makers really need more data for ale yeasts.
In this experiment, a high versus low nutrient rate is tested for its flavor and aroma contributions in a 4.5% ABV, carbonated, dry traditional mead fermented using an English ale yeast, Safale S-04. The ale yeast strain, S-04 is noted to have a clean flavor, contributing apple esters, and has a fast fermentation with good flocculation. Other than the nutrient level, all other variables were identical, and the batches were fermented side by side. Triangle tests are conducted to see if participants can correctly identify the difference between the two meads. Respondents also provided feedback on the differences perceived in the two meads.
My hypothesis was that it may affect the ester levels, but given my prior success using medium and high levels of nutrients, I thought that a low level of nutrients could cause a sluggish fermentation and could also have potential off flavors.
Recipe: 4.5%, Dry Traditional Short Mead, January 2021, 1 gallon
- OG = 1.034
- 0.33 liters of 2020 golden wildflower Honey from Nith Valley apiaries
- 4 liter water spring water in 8 liter jug
- 0.35 grams of calcium chloride
- 0.2 grams of potassium bicarbonate
- 0.12 grams of Himalayan sea salt
- 1.5 grams of Safale S-04
Honey tasting notes:
- Moderate floral
- High perceived sweetness,
- honey comb, honey like, bees wax
Nutrients (calculated using The MeadMakr BatchBuildr):
- Low nutrient requirement: 65.2 ppm YAN
- Fermaid-K: 0.5 grams
- DAP: 1 grams
- High nutrient requirement: 108.7 ppm YAN
- Fermaid-K: 0.9 grams
- DAP: 1.7 grams
- Mixed honey, water, salts.
- Shook jug for two minutes to aerate
- Sprinkled yeast on top of must
- Degassed by swirling the jug every 12 hours for the first two weeks.
- +2 hours, first 1/3 nutrient addition. 64.3°F
- +24 hours, second 1/3 nutrient addition. 65°F
- +48 hours, third 1/3 nutrient addition. 65°F
Regular temperature, pH, gravity and aroma tests were taken. The table below summarizes the observations.
|+0 hours 63°F||pH 7, |
|pH 7 |
|+20 hours 67.6°F||Honey/ fermentaion/ yeast||Raw honey / light fermentation|
|+36 hours 66.9°F||Some ferm stress/ light rubber, 1.026||Raw honey / clean, 1.028|
|+3 days 66.6°F||Clean/ light esters, 1.012||Clean, honey, 1.014|
|+4 days 64°F||fruity esters, 1.006||Raw honey, 1.014|
|+5 days 66.6°F||fruity esters/ clean, 1.004||Clean/ honey, 1.012|
|+6 days 66.6°F||Clean/ floral/ light pomme esters. No heat/ smooth, 1.000|
Transferred to secondary
|Light esters/ honey, 1.008|
|+7 days 66.6°F||Great/ honey/ some esters, 1.006|
|+9 days 66.6°F||Hight raw honey clean, low pomme esters, 1.004|
|+13 days 66.6°F||clean, light honey, low/no esters, 1.000 |
Transferred to secondary
Basically, the fermentation with the low nutrients took twice as long to ferment. Throughout the fermentation, the low nutrient regime smelled more like honey, whereas the high levels of nutrients has more pear and apple esters.
- Cold crashed at 58°F for 2 weeks.
- Bottled after one month to 2.5 volumes.
The mineral profile of the spring water, contribution of the salt additions and the final water profile was as follows.
Initial Tasting Notes
At bottling, I tasted both meads and they both seemed different but were both excellent. The mead with the low nutrient levels had mild diacytal/butterscotch and some sulfur. However, it was sweeter, and had more raw honey, and minerality. In contrast, the high nutrient level was all apple and pear ester character. The high level nutrient mead also had more noticeable alcohol, was dryer, and tasted like fermented honey rather than raw honey.
Due to Covid-19, and in consultation with statisticians, every participant was sent two bottles and completed up to five triangle tests. Participants were sent four or five experiments and knew that I was testing something around nutrient regimes, clarifying agents, and acidity levels. The bottles were labeled experiment A, B, C etc as well as being labeled as treatment or baseline. Every participant was also sent enough identical red solo cups.
One scoresheet was filled out by each participant for each experiment. Participants were asked their experience level with meads, how blown their palate was, and their status as judges and home/professional brewers. Experience was given a value from one to five where one is first time having a mead to five being well-experienced. Palate was given a value from one to five where one is having had nothing to drink yet, and five was that they’ve already had too much (like just drank an IPA and sitting in a brewery). Participants were asked to say which mead they preferred and just select one if they couldn’t tell the difference.
There were 6 participants and 26 triangle tests were completed overall. We had a good selection of mead experts and mead enthusiasts. There were two BJCP certified beer judges and two BJCP mead judges.
On average, people were experienced with meads and their palate was not tired. All had some experience with meads and off flavors. Basic summary statistics on the self reporting of participants experience and palate when taking the triangle tests:
All responses were collected within the same month. Tests were evaluated when the meads were 4-5 months.
There was a significant difference between the two meads. Out of the 26 triangle tests 24 could identify the odd mead out! A test that the results were from random guessing is rejected with near 100 percent confidence. Moreover, all participants preferred the low nutrient rates. The null hypothesis of equal preference between the meads is rejected with 96.8 percent confidence. Here is a summary of the results.
I did three triangle tests and found them effortless. Participants commented that it could be done by aroma alone.
What people described as the difference between the two meads is summarized below. The high nutrient level mead was described less pleasantly, with alcohol and off flavors described. In contrast, the low nutrient as being more honey like and was smoother. I was surprised at the results, as my tasting at bottling was quite different.
I was really surprised by how different the two meads were. When I began making mead, I often used high level of nutrients with ale yeasts because that was what was recommended. Now, there is clear evidence that for me that when staggering with inorganic nutrients it’s better to go low. It’s also interesting that despite all the focus on quickly turning around meads, it is clear that you can lose out from too aggressive of a fermentation. Is it worth taking half the time in fermentation if the product is half as good?
I focused this test on inorganic nutrient because I want people who do not have access to Go-ferm and Fermaid-O to still be able to make great mead. That said, the use of Go-ferm and some organic nutrient may have further benefited the meads, especially for the first nutrient addition. More testing is still needed for this.
Moreover, I have done bench trials on the timing of the nutrient additions for short meads since this experiment. The timing of the nutrient additions was quick and could have benefited from staggered out more. As summarized in the metadata of Short Mead Recipes, S-04 is particularly susceptible to nutrient burn from staggering with inorganic nutrients when done within the first 24 hours. The yeast is one of the few I know that can handle all the nutrients up front. I plan to replicate this experiment but with a better staggering regime and compare a low and medium level of nutrients. It would also be interesting to compare all nutrients upfront versus stagger when using S-04.
I still use a medium to medium-low nutrient level when staggering inorganic nutrients in short meads with S-04 as I do like the extra apple esters that results. Of course, it is also needed to test nutrient levels for different ale yeast, including Kveik and US-05. What is clear is that when it comes to ale yeasts, there is still much to learn.
Peer Review 1: Justin Angevaare, PhD, Statistician, author of p-value calculator and award winning homebrewer, https://onbrewing.com
Is each triangle test considered an independent observation here, or is some kind of trial-dependence built in to your analysis?
- Response: Yes, every observation was considered independent.
Would it be helpful for consistency if the mead was decanted into a serving vessel first, and exact measured quantities added to the cups (cups on a scale while being filled perhaps?) I’m wondering if a bias in fill level could be picked up by participant due to having 7 of one to fill and 8 of the other (FWIW seems unlikely to me). Maybe they should fill 8 from both, and randomly select one to discard after having done so?
- Response: Each participant was given a 12 oz bottle and told to pour 1 to 1 1/2 of an oz in each cup, with the expectation that they would pour the same amount in each cup. I doubt the participants poured less for the sample with 7 vs 8 cups, but it’s an excellent idea to just say 1 oz for each cup from now on and do 8 to avoid the potential for this.
Was the identification rate equal on the early trials vs. later trials? I can imagine some learning rate by the taster based on them learning results after each trial. I can also imagine certain volatile compounds that may help in identification may have flashed off by trial #5.
- Response: The incorrect responses were given on trial 2 and 5. Yes, there could be a time fixed effect from learning or volatilization. I also suspect there could be palate fatigue. I suspect there are forces in each direction but given the low failure rate, I suspect controlling for it would not matter for this experiment. It would be interesting to test for this in an experiment that exhibited more trial time variation.
Are the temperatures listed fermentation chamber temps or mead temperatures? Can it be assumed the faster fermentation resulted in a higher temperature for that mead?
- Response: Mead temperatures. I did not detect temperature variations across meads possibly due to the small volume.
Is phenol production characteristics of yeast generally a matter of gene promotion rather than gene presence/absence? Asking as I can’t remember having had a non-phenolic yeast produce perceptible phenol.
- Response: No idea. S04 is phenolic prone and in experience more sensitive to phenolic production relative to say US-05.
Are any components of your nutrient regime phenol precursors/directly used by yeast in phenol production? (ferulic acid on my mind)
- Response: The exact formula of Fermaid-k is not published, so I do not know. It would be interesting to replicate the experiment, but with different nutrient sources. I suspect the use of DAP, particularly when combined with staggering in the first 48 hours, was a major reason for the phenolics. It’s an interesting question though, maybe S04 does have higher nutrient requirements, but it’s just that S04 is sensitive to DAP levels or a certain range. This could explain the sluggish fermentation and off-flavors noted at bottling. Replication with other yeast and nutrient regimes could give us a better idea of this.
Side comment: Reading this reminds me of something I’ve run into repeatedly with sourdough cultures – feeding mine with high nutrient flours like rye or fresh whole wheat seems to result in increased vigour, but also more sulfur, solvent aroma, and less depth.
- Response: Interesting. When people advocate for certain nutrients by showing better fer mention speeds, like in the case for Go-ferm in the Scott labs Handbook, I always wonder to myself: “yea but what does the difference taste like?”
Peer Review 2 Chris Kwietniowski, award winning home brewer, triangle test participant.
Being one of the tasters having done the 7 of this and 8 of that pouring myself, I think for certain people there may be an element of “counting cards” when you reach the last trio of samples. I certainly didn’t, but maybe someone reaches the end knowing which sample must be the one with 2? But in red solo cups with 1/8th of a bottle, the filling level is like 1/2″ to 1″ with sort of random inconsistencies.
- Response: It was good to hear your experience and that this did not affect your triangle tests. It will be a good idea from now on to just ensure the instructions insist the pour is identical across cups and not give the 1 to 1 1/2 of an oz range. I will also ask for the order of the tests to be randomized or make sure that 8th cup gets added randomly.
And about the side comment on speed… I think the assumption is right that “faster is better” when comparing to a miserable, sluggish, or incomplete ferment. But at some point you can probably make gains in speed without improving the product.
- Response: Agree, and the focus on fermention optimization needs to continue to find that sweet spot.