Fermenting · build log
Salt is two to two and a half percent of the weight of the shredded vegetables. That single number is the whole sauerkraut method. Everything after it is deciding when to stop. This page writes down the method, the yogurt protocol we have not run yet, and what the human trials on fermented food actually found, including the one that came back close to empty.
One batch of sauerkraut exists. It was 54 hours old when this was written on 2 August 2026 and nobody has looked in the jars since, so that hour count is the last observation and not a live one. Four 2L jars, cabbage and carrot, bubbling, tasted good on day 2. That is the entire sensory dataset: one person, one taste, one day. Zero batches of yogurt have been made. The yogurt section below is a protocol we have read, not a thing we have done.
No pH has been measured. No weights were recorded for this batch. Nothing here is for sale.
Fact-checked 3 August 2026, and it changed the page. Every citation was pulled and read. The trial numbers held up. Seven other things did not, including the mould rule, the brine top-up arithmetic and how much of the yogurt method actually came from the extension guide it was credited to. The corrections are marked where they happened rather than quietly swapped in.
This is not medical advice and nothing on this page treats or prevents anything. The evidence section reports what specific trials measured. It does not turn that into a recommendation, because the trials do not support one yet.
Fermented food gets sold as a settled thing. It is not settled. Two good human trials point in different directions, and the honest version of this page is to put both of them up rather than quote the flattering one.
Wastyk and colleagues, Cell, 2021, run at Stanford. A 17 week randomised trial, 18 people per arm, with a 10 week diet inside it.
The fermented food arm went from 0.4 to 6.3 servings a day across yogurt, kefir, fermented cottage cheese, kimchi, other fermented vegetables, brine drinks and kombucha. Gut microbiota diversity rose steadily. Of 93 inflammatory serum proteins measured, 19 fell, the paper naming IL-6, IL-10 and IL-12b. Signalling activation dropped in 14 of the 60 cell-type-specific responses tested, spread across all four cell types the study looked at, which were CD4+ T cells, CD8+ T cells, B cells and classical monocytes.
The comparison arm doubled fibre, from 21.5 to 45.1 grams a day. Those consumers split into three different immune trajectories that tracked their starting microbiota diversity.
The headline result is a secondary outcome. The trial's stated primary outcome was the cytokine response score, and the full text says the change from baseline was not significant for either arm. The abstract mentions that only against the fibre arm, which is how the fermented arm ends up looking cleaner in most retellings than the paper reads.
Schropp and colleagues, Microbiome, 2025, and the companion inflammation paper, 2026. Same trial. 87 people randomised and 84 completing everything, in a randomised crossover: 100 g of sauerkraut a day for four weeks, fresh in one phase and pasteurized in the other, with a four week washout before each phase.
Their conclusion, verbatim: the gut microbiome of healthy individuals is "rather resilient to short-term dietary interventions even though single species might be affected by sauerkraut consumption." Pasteurized sauerkraut moved more markers than fresh, and only the pasteurized arm raised serum short chain fatty acids.
On health outcomes: a small drop in systolic blood pressure from both forms, no change in gut barrier integrity, and changes in inflammation and glucose markers only inside subgroups. The authors concluded that live bacteria are not essential for the blood pressure effect, and that four weeks of daily sauerkraut "was not linked to an appreciable systemic health benefit in healthy individuals."
They are not the same experiment. The Stanford arm reached six servings a day of many different fermented foods for ten weeks. The sauerkraut trial gave one food, 100 grams a day, for four weeks. If dose, variety and duration matter, and there is no reason to assume they do not, both results stand. Nobody has run the trial that separates them. That is the actual state of the field.
The review that maps the mechanisms says the same thing more bluntly. Caffrey, Sonnenburg and Devkota, Cell Metabolism, 2024 is the paper behind the interesting molecular story. D-phenyllactic acid, made by lactic acid bacteria, is the only known ligand for the HCA3 receptor, an active version of which is carried only by the great apes, humans among them, and the siamang. It is elevated in plasma after sauerkraut is eaten. Their own summary of the field: "The rise in health claims about fermented foods is not matched by an increase in clinical evidence." They also name the lack of an adequate animal model and the absence of any known HCA3 inhibitor as the reason the biological relevance is still open, so the mechanism is suggestive rather than demonstrated.
So the fair reading is narrow. In one well-run trial, a high and varied fermented food intake for ten weeks raised microbiome diversity and lowered a set of inflammatory serum proteins in healthy adults, while the outcome that trial had named in advance did not move in either arm. That is a real finding about secondary markers, in healthy people, at a dose most people do not eat. It is not a treatment, it is not a cure, and one food alone at a normal portion for a month did not reproduce it. Cabbage is cheap and it tastes good, which is reason enough to make it.
Method one
It is a weight method, not a recipe, which is why it survives any change of scale or vegetable. NC State Extension gives the salt for sauerkraut as a final concentration of 2.0 to 2.5 percent. Two percent is the bottom of that range and the crunch-friendly end of it.
Read that as a floor, not a target. The NCHFP says the salt in fermented kraut and brined pickles "not only provides characteristic flavor but also is vital to safety and texture", and carries the caution "Do not attempt to make sauerkraut or fermented pickles by cutting back on the salt required." Its own recipe is salted by volume rather than by percent, so the two numbers are not directly comparable, and we are not going to publish a conversion we would have to invent. If you are choosing, 2.5 is the conservative end.
When you move it to the fridge is not a matter of taste preference. It decides whether you made a preserved food or a live salad.
| Fridge at | Acidity | What you have |
|---|---|---|
| Day 2 to 4 | around 0.3 percent, pH still above 4.6 | Mild, fizzy, very crunchy, and not preserved. Above pH 4.6 the acid is not doing the preserving, so the fridge is. Keep it cold and eat it within a couple of weeks. |
| Week 3 to 4 | 2 to 2.5 percent, pH near 3.5 | Properly sour, softer, keeps for months cold. This is sauerkraut in the sense the pH table means it. |
These four jars are at day 2 and taste good, which is what the fizzy end is supposed to taste like. The plan is to split them: two jars to the fridge now for crunch, two left to run all the way to sour. Same cabbage, same day, two products, and it is the cheapest way to learn what the other three weeks actually do.
Cold slows a ferment rather than stopping it, so a jar in the fridge should keep getting sourer over months. We have not held a jar long enough to watch that happen and we have no extension figure for the rate, so treat it as the expectation we are working to, not a measurement.
Do not pour it away. It works anywhere you would use vinegar and it is a drink. If you eat a jar down, push the rest back under the liquid, because UC Davis documents white mould growing on sauerkraut that was stored out of its brine. The other thing people do with old brine is splash it into a new batch as a starter, on the theory that it shortens the slow first day. We have not tested that and have found no extension source for it.
This is the part of the page worth getting right, so it follows one source rather than the fermenting internet. UC Davis Food Science and UC Agriculture and Natural Resources publish Common issues with fermented fruits and vegetables, and it is unambiguous.
On mould, verbatim: "Mold growth can occur anytime during the fermentation process and is a sign of a failed fermentation. If you confirm mold growth on any part of a ferment, it should be immediately discarded." The whole jar, not the top of it. That is an extension rule, and an earlier version of this page wrongly said no extension published one and that discarding was a house preference. It is not a preference.
On the flat surface film, the same document says white, grey or pink films are typically yeast, the thing home fermenters call kahm, and that for sauerkraut the yeast should be periodically removed because heavy growth causes off flavours and textures and reduces acidity. It also says to check that the pH does not rise above 4.6.
Colour will not tell you which one you have, and this page previously implied it would. UC Davis lists pink and grey as yeast colours, and separately photographs white mould on sauerkraut. So the honest heuristic is texture and nothing else: a flat film lying on the surface is the yeast case, anything fuzzy or raised is the mould case, and if you cannot confidently call it flat, the jar goes out. Ohio State Extension explains why the surface is not the whole story, saying a mould filament "can penetrate through the surface of softer products" and that with soft products the toxin "cannot be eliminated by simply removing the mold and surrounding product". A jar of kraut is brine, so it is the soft case rather than the firm-vegetable case. A putrid smell, per UC Davis, is also a failed ferment and also goes out. A cabbage costs almost nothing. A wrong call is the expensive one.
Both problems are oxygen problems, and both stop happening when the solids stay under the brine.
Four levers, strongest first. Weigh the salt rather than eyeballing it, because the NCHFP calls salt "vital to safety and texture" and warns against cutting it back. Stay under 75 F, above which the NCHFP says kraut may become soft. Then two we have no source for and are carrying as working theory: shred thicker, on the reasoning that thin shreds collapse sooner, and fridge earlier, on the reasoning that crunch declines the whole way through. Neither of those last two is sourced or tested here. Tannin sources like bay, oak or grape leaf are the traditional firmness fix, we have not tested it, and we have found no extension figure for it.
Method two, never run
Zero batches made. Steps 1 to 5 and step 8 are transcribed from the University of Alaska Fairbanks Cooperative Extension guide. Steps 6 and 7 are not, and that matters: the UAF guide covers yogurt and does not cover straining, Greek style yogurt or cheesecloth at all. The straining steps are ordinary kitchen practice plus one inference from a dairy paper, and they carry no extension backing. An earlier version of this page said the whole section came from UAF. None of it is our own measurement.
The only measured straining numbers we have come from Karastamatis and colleagues, 2022. Strained straight after incubation, 100 g of set yogurt gives 30.5 g of strained yogurt and 69.5 g of acid whey. Strained after 24 hours chilled, the same 100 g gives 33 g of yogurt and 67 g of whey, which the authors describe as about 8 percent more yogurt and about 11 percent less acid whey.
Read the method before you read the numbers. That study used a commercial freeze dried starter and bovine milk standardised to 3.2 percent fat, and it separated the whey by centrifuging at 5500 rpm for 15 minutes. It is not a bowl and a cloth in a fridge. Whether the ratio or the chilling advantage survives the move to gravity straining at home is an assumption on our part, not a finding, and the first real batch is what will settle it.
| Milk in | Set yogurt | Light strain, about 2 to 1 | Hard strain, about 3 to 1 |
|---|---|---|---|
| 1 L | about 1 L | about 500 g | about 330 g |
| 2 L | about 2 L | about 1.0 kg | about 660 g |
The 3 to 1 column is a centrifuge result from somebody else's lab, carried across to a kitchen on the assumption that the ratio holds. The 2 to 1 column is an expectation for a short home strain and nothing more. Neither column has been weighed here. The first real batch gets weighed at every stage and this table gets replaced.
The whey is not waste. It goes into bread, marinades and smoothies, and it can top up a kraut jar. Several litres a month down the drain would be a silly habit to acquire.
A quarter cup of your own batch will start the next one, and UAF does say to save that much from a quart batch. But UAF's actual instruction is stronger than the reculturing habit, and we had softened it: it says "Use a fresh, recently purchased culture from the grocery store each time you make yogurt." Reculturing indefinitely is our intention, not extension guidance, and it runs against what UAF prints.
The number you see repeated online is five generations before the strain balance drifts. We have no source for that figure and we have not tested it. The plan is to number each generation and log set time and thickness, which is the only way to find out what happens in this kitchen, with the UAF instruction on the record as the cautious alternative.
Long life UHT milk is already heat treated, so step 1 looks redundant with it. The reason to heat anyway is that UAF gives the purpose of the heat as protein behaviour rather than sanitation, saying the milk must be heated so the proteins bind together instead of forming curds and whey. UAF does not discuss UHT milk, so the read across is ours. We have not tested skipping it, so the first batch runs the full heat and gives us a baseline.
"How To Heal 20 Years of Gut Damage in 30 Days" by Nick Norwitz MD PhD, published 27 July 2026, running 18 minutes 19 seconds. Oleg watched it and the jars followed. That is the honest extent of the causal chain.
We are not going to quote the video, because we could not verify a transcript. An earlier version of this page put words close to quotation marks around advice about starting with the blandest fermented food and building up, and around a remark about eating it in Greek yogurt. We could not retrieve the captions to check either one, so both are gone. Everything below is from the video's own public description, which we did read.
The description sets the video up as what the author would do if he could go back to the day he was diagnosed with inflammatory bowel disease and start over. It lists the ground covered as the gut brain axis, vagus nerve signalling, breathwork, stress reduction, time restricted feeding, fermented foods, probiotics and microplastics. Its four chapters are the recovery story at 0:00, free foundational tools at 1:55, dietary principles at 7:32, and a peptide stack at 12:57.
Two things worth stating plainly rather than laundering.
The title is a healing timeline. We have found no published trial that supports healing twenty years of gut damage in thirty days, and none of the four papers on this page comes close to testing such a thing. We are not characterising what the video argues, only saying that the claim in the title is not one the literature we read supports. Treat it as a title.
Second, the last chapter is a supplement stack, described in the author's own words as BPC-157, Larazotide and KPV. The description offers a discount code for those three peptides from a named vendor at 10 percent off, and a separate code for a probiotic from another vendor at 15 percent off. He discloses both in the description himself, which is where we read them. It is a commercial interest sitting next to health content, and it is worth knowing about when you weigh the rest. None of it is on this page and none of it is in this kitchen. The fermented food part stands on its own published evidence, which is the part we followed and the part we checked.
Method
Evidence