A bentonite dose copied from another wine wastes the trial: the clay can miss the protein haze in this one, or take aroma compounds out with it. Match the addition to the wine in the vessel, then record a bench trial instead of a single spoonful.
Bentonite timing follows the wine in the vessel. Keep a kit on the day and the amount on its own sheet. No kit dose is copied here. Juice before fermentation is a separate job: the Australian Wine Research Institute says a juice addition can work, and that much higher rates are often required because many juice proteins drop out during fermentation without clay. Purdue Extension FS-53-W (New 03/10) writes that this early fining may lead to a sluggish fermentation. For a finished wine, AWRI places protein fining on the final blend, after fermentation and after other protein fining. Purdue’s figures for that protein job run from about 10 to 300 mg/L of protein and about 60 to 1,800 mg/L of bentonite across wines. One spoonful does not cover that span. Leave every result cell blank until the sheet or the trial fills it.
This page compiles the AWRI fining-agents page and Purdue Extension FS-53-W (New 03/10), read on 14 October 2026 from copies saved on 11 October 2026. The AWRI copy prints no revision date. It is a way to choose the timing and file the trial. It is not a dose for every wine, and it does not rank clay brands.
Key points:
- Three jobs, not one day. A kit sheet, juice before fermentation, and a finished blend are different additions.
- The rate is a test result. AWRI uses a heat test at 80°C for six hours. Purdue describes other heat examples and calls a single routine addition a path to over- or under-dosing.
- Name the pass line. AWRI reports a 0.5 NTU criterion in some labs and 2.0 NTU as what other practitioners call more reasonable. They are not the same line.
- Retest after the wine changes. A blend, a sweetening, a later protein fining, or a pH change can undo a result you already accepted.
- Blank stays blank. If the bag, the sheet, or the test does not state a cell, do not fill it from the other column.
| Wine in the vessel | Where the addition sits | What to copy into the record | What stays blank |
|---|---|---|---|
| A kit with its own instruction sheet | The day and the amount printed on that sheet | Product name, the sheet’s day, the sheet’s amount | Any figure from the rows below, including Purdue’s 60 to 1,800 mg/L span |
| Juice or must, before fermentation | AWRI: the addition can work, and much higher rates are often required because many juice proteins precipitate during fermentation anyway. Purdue: fining juice first may lead to a sluggish fermentation | That the treatment was on juice, and that the finished wine still needs its own test | How much higher. Neither source prints a multiplier |
| Final blend, after fermentation | AWRI: best after fermentation and after protein fining, on the final blend. A later blend, sweetening, or protein fining can call for another light addition | Blend identity, the named test, and the result | A dose taken from a different wine |
A bowl of fine beige powder, a wooden spoon, and a spill on the bench. The bowl has no label. The picture does not show a weight or a wine.
Which wine are you treating?
Start from the vessel, not from a habit. The three rows above are the split. A kit sheet is an instruction for that product. AWRI’s final-blend timing is an instruction for protein stability of a finished wine. Purdue’s juice warning is an instruction about fermentation speed and leftover fructose. Using one sentence for all three is how a copied spoonful misses.
The supplier pages saved with this research, from Midwest Supplies and a homebrewing clarifier page, stop at the site menus. They do not include the instruction paragraphs, so this page does not state a kit day or a kit amount. If the sheet is not in your hand, leave the kit cells blank and do not borrow the cellar row to fill them.
Juice is the early case both institutes discuss, and they discuss different consequences. AWRI says bentonite on juice can succeed, then notes that juice holds many proteins that drop out during fermentation with no clay at all, so much higher rates are often required. No number is attached to “much higher” on that page. Purdue’s consequence is the ferment itself: fining juice before fermentation may lead to a sluggish fermentation, because the clay clarifies the juice and can strip growth factors such as fatty acids, phospholipids, and sterols. Purdue adds that a longer fermentation can leave more residual fructose, and that fructose is twice as sweet as glucose, which changes how dry the wine tastes. Neither sentence is a reason to skip a later test on the finished wine.
The finished wine is where AWRI puts the protein job. Bentonite fining is best performed on the final blend, after fermentation and after all protein fining. The same page says some producers trial an addition toward the end of fermentation to help yeast settle and to cut the number of later racking steps. That trial is a settling aid. It is not a substitute for the final-blend test. If the wine is later blended or sweetened, or if a protein fining agent goes in afterward, AWRI says an additional light fining may be required. “Light” is not given a milligram figure on that page. The figure comes from the new trial.
Takeaway: Write “kit,” “juice,” or “final blend” on the record before you write an amount. If you cannot name which of the three you are holding, the amount cell stays blank.
What do you record before any addition?
Six fields. The row is a form. It is not an example result, and no cell below is a dose from this site.
| Field | What to write | What to leave blank |
|---|---|---|
| Product and batch | The name on the bag. Sodium or calcium if the bag says so. AWRI says the lab trial and the cellar addition have to be the same batch | The form, if the bag does not state it |
| Rehydration | Water or wine, water temperature, how long it stood, and the slurry strength you mixed | Any of those you did not write down at the time |
| Treated sample and control | The addition you pipetted into the sample, plus a sample with no addition. AWRI says a no-addition control should always be set up so the sensory effect can be judged | A control you did not pour |
| Stability test | The named test, copied from the next section, including temperature and time | A nickname such as “heat test” with no source |
| Result | Clear or hazy, or the NTU increase, and the pass line you are using | A result you have not read. Do not enter 0.5 or 2.0 unless that line is the one you chose from the AWRI sentence |
| Retest trigger | Blend, sweetening, a later protein fining, or a pH change, once it happens | The word “done” written in advance |
Rehydration is part of the dose, because the clay has to swell. AWRI’s laboratory stock is 5 percent weight/volume: 5 g sprinkled into about 85 mL of water at about 60°C, cooled, stood overnight, then made to 100 mL. One millilitre of that stock in 100 mL of wine equals 500 mg/L. That sentence defines the pipette. It is not a rate AWRI tells you to add. The rate, on that page, is whatever the heat test says is enough.
Purdue’s bench slurry is a different stock. A water-to-bentonite ratio of 16 to 1, stated as 60 g per 1 L, is a 6 percent weight/volume slurry. FS-53-W points to a Table 1 of additions for that slurry. The text copy used here does not include that table’s cells, so they are not reprinted. Purdue’s cellar practice is different again: about 8 to 1, stated as 1 kg of bentonite per 8 L of water, or 1 lb per gallon, which Purdue says allows additions up to 120 g/hL (10 lb per 1,000 gallons) while keeping added water under 1 percent of the wine. Above 60 g/hL of the 6 percent slurry, Purdue says the water addition exceeds 1 percent. Those are limits on water, not a target dose.
Two preparation rules sit beside the numbers. Purdue says rehydrate in clean, chlorine-free water at 140°F (60°C), add the clay to the water rather than the water to the clay, let it swell at least four hours, and do not leave the slurry past overnight because of microbial growth. Rehydrating in wine, Purdue says, does not let the clay swell fully. The maximum water Purdue allows for dissolving the clay is 16.7 L per kilogram (2 gallons per pound), and water from all processing sources should stay at or under 1 percent of the wine. AWRI asks for the same water, the same temperature, the same makeup, and the same rehydration time in the lab trial and in the cellar, plus the same batch of clay. This site’s fermentation guide already records a Connecticut station line of at least three hours of hydration. Purdue’s minimum here is four hours. Write the time you waited. Do not average three and four.
Purdue also says the bench mixing has to match the cellar. The reaction is quick and not instant. At least 10 to 15 minutes of vigorous mixing, with the wine above 50°F (10°C). If the trial is mixed harder or warmer than the tank, Purdue says the trial will underestimate how much clay the tank needs. After the addition, Purdue allows about one week, depending on tank height, for gravity settling, and says limiting contact time limits lead residues extracted into the wine. No lead concentration is printed in the sheet read here.
Takeaway: The record is ready when the six fields are filled from the bag, the slurry you mixed, and a test you named. An empty cell stays empty.
Which heat test are you actually running?
Name one test. AWRI and Purdue do not describe the same heat exposure, and Purdue does not treat a heat result as the same thing as a wine’s life on the shelf.
| Test, as stated | Conditions in that source | What counts as a result there | Do not copy across |
|---|---|---|---|
| AWRI heat stability test | 80°C for six hours | A haze in the fined, heated wine means it is not stable and more bentonite is required. With a turbidimeter, some labs use an increase greater than 0.5 NTU against the unheated control. Other practitioners call 2.0 NTU a more reasonable criterion | Purdue’s 49°C or 90°C examples, and Purdue’s ethanol assay |
| Purdue heat examples | 49°C (120°F) for 48 hours, or 90°C (194°F) for one hour, then cooling | Purdue calls these the quickest tests and says it is an uninformed assumption that they match protein precipitation over the life of a properly stored wine | AWRI’s six-hour test, and either NTU figure |
| Boulton ethanol assay, as Purdue describes it | Titration with successive quantities of ethanol, read on a nephelometric turbidity meter, for all colloidal materials | No volume series and no pass number in the sheet read here | A heat-test temperature, or 0.5 NTU, or 2.0 NTU |
Three glasses of pale wine. The left glass looks cloudier than the right. Nothing in the picture is labeled, and the cloudiness is not a heat-test reading.
The pass line is part of the result. An increase of 1.0 NTU fails the 0.5 NTU line and passes the 2.0 NTU line. That comparison only places the two AWRI criteria against each other. It is not a reading from a wine. If the meter printout does not state which line you used, write “NTU line not stated” and do not insert 0.5 or 2.0 to finish the cell. If you have no meter, AWRI’s non-instrument reading is the haze itself after 80°C for six hours. That visual call is not a Purdue 49°C result with the clock changed.
AWRI also says to adjust pH and titratable acidity before the fining, because bentonite is more effective at lower pH and stability might differ after the adjustment. Purdue makes the retest explicit: even a small pH change can alter protein solubility, high pH is likely to raise bentonite demand, and protein stability has to be checked again after any treatment that changes acidity. Cold stabilization is one of those treatments when it moves the pH. The heat test you run before that step does not answer the wine you have after it.
Takeaway: Write the test’s temperature, time, and pass line in the same cell. A haze from a different time and temperature is a different test.
What forces a second test?
Four changes, each from one of the two sources. If none of them has happened, do not invent a second dose.
| Change | What the source says | What you do |
|---|---|---|
| The wine is blended, including two wines that each passed | Purdue: a shift in alcohol, pH, or protective colloids can leave the blend unstable. The mixture needs its own test, and more fining might be needed | New row. The old pass does not transfer |
| The wine is sweetened, or a protein fining agent is added later | AWRI: an additional light fining may be required. No milligram figure for “light” | New trial. Leave the amount blank until that trial |
| pH or titratable acidity changes | AWRI: do the adjustment before fining. Purdue: reassess after the change. Calcium bentonite is not recommended above pH 3.4 | If you already fined, the old result is for the old pH |
| The clay batch or the rehydration changes | AWRI: lab and cellar must use the same batch, water, temperature, makeup, and soak time | The old trial does not license the new bag |
A rack by itself is not on that list. The question “before or after racking” is a search phrase, and neither page answers it with a dose. Purdue’s week is a settling time after the clay is in, and it depends on tank height. AWRI’s extra fining is tied to blending, sweetening, or a later protein fining, which a rack might accompany and might not. If all you did was move the wine off lees, the record does not gain a new amount.
The bottling guide is where protein stability sits next to the other checks that decide a bottling date. A bentonite row does not clear residual sugar, malic acid, cold stability, or sulfur dioxide. Those stay on that page.
Takeaway: Open a new row when the blend, the sugar, the pH, or the clay changes. A quiet rack does not fill the amount cell.
What changes if the wine is red, or the clay is calcium?
Red wine is not a white-wine dose with the color ignored. AWRI says use in reds should be limited because bentonite adsorbs anthocyanins and reduces color. Purdue says tannins in Vitis vinifera reds usually denature grape proteins during aging, so those wines often need less of this fining, and then names the wines that still need a protein test: anthocyanin-rich, tannin-poor red European-American varieties, blush and rosé, and red vinifera from very cool climates. For blush and rosé, Purdue says the bench samples have to be read for color change as well as haze. A trial that only records “clear” has skipped the color check Purdue asks for on those wines.
Sodium and calcium are not a winner and a loser in the sheets read here. Purdue says the choice does not matter as long as you find the smallest dose that passes the test you are using. Purdue also reports the supplier arguments without adopting them as measurements: suppliers of sodium bentonite argue that it has about twice the protein-fining capacity of calcium bentonite, and suppliers of calcium bentonite argue that it swells less, makes fewer lees, and loses less wine at racking. AWRI’s own contrast is practical: calcium bentonites often need higher addition rates, hydrate easily, and give compact lees with less wine loss, and calcium can be extracted into the wine, so calcium instabilities are the risk to watch.
Purdue’s calcium illustration is not a recipe. An addition of 1,920 mg/L (16 lb per 1,000 gallons) of calcium bentonite, which Purdue calls equivalent to 960 mg/L (8 lb per 1,000 gallons) of sodium bentonite, on a protein-rich Gewürztraminer, would add the potential for 114 mg/L of calcium tartrate. Calcium tartrate, Purdue notes, does not drop out in a cold stabilization the way potassium bitartrate does. That paragraph shows why a heavy calcium addition is a tartrate question. It is one stated example. It is not the dose for the next carboy, and the 114 mg/L figure belongs to that example only.
A jar of thick beige slurry next to a corked bottle of pale wine. The glass has no graduations. The picture does not show a slurry strength or a wine treatment.
Takeaway: On a red, blush, or rosé, write down the color of the treated sample next to the haze result. On calcium clay, the pass line still comes from the test, and Purdue’s Gewürztraminer figures stay in that example.
How this page was put together
The claims above come from two public documents read on 14 October 2026 from copies saved on 11 October 2026. AWRI states the final-blend timing, the juice caveat with no multiplier, the 5 percent stock and the 500 mg/L pipette meaning, the 80°C six-hour test, the 0.5 and 2.0 NTU criteria, the same-batch rule, and the limit on red-wine color. Purdue FS-53-W (New 03/10) states the 10 to 300 mg/L protein span, the 60 to 1,800 mg/L bentonite span and its pounds-per-thousand-gallons pair, the juice-fermentation warning, the heat examples and the criticism of them, the ethanol assay without volumes, the rehydration times and water limits, the pH 3.4 calcium line, the blend retest, and the Gewürztraminer calcium-tartrate example. The Connecticut three-hour hydration line is the one already cited on this site’s fermentation guide; the Connecticut document was not re-read for this page.
Midwest Supplies and the homebrewing clarifier page were in the research set. The saved copies contain the site menus and not the instruction paragraphs, so no kit day and no kit spoonful appear here. A mead-forum thread was in the same set and is not used. No figure on this page was measured here. FS-53-W’s Table 1 cells were not in the text copy, so they are absent on purpose. The photographs show a settled jar and a carboy, a bowl of powder, three glasses, and a jar of slurry beside a corked bottle. They are not a cellar log and not a test result.
