Twenty Bakes

This is not a bread-baking course. It is a programme in controlled cereal fermentation, in which baking is the final step and often not the most important one.

Every technique here — souring, scalding, soaking, sprouting, malting, milling, co-fermenting with pulses — existed long before anyone measured what it did. The programme's working assumption is that these techniques persisted because they solved real problems: they made grain digestible, storable, palatable and nutritionally adequate for people who had no alternative. Modern research is used to explain why they worked, to find the dose at which they work, and occasionally to show that a tradition has been misunderstood by its modern admirers. It is not used to grant permission.

The four tiers, and how they are weighted

What this programme does not do

It does not appeal to government dietary guidance as a reason to eat or avoid anything. Where there is a genuine acute hazard — ergot in rye, cyanogenic glycosides in flax, cyclopiazonic acid on badly stored kodo millet, pathogen amplification during sprouting — the hazard itself is named, with its mechanism and, where known, its dose. That is a different kind of claim from a dietary recommendation, and it is treated differently.

It also does not optimise for an open crumb. An open crumb is one property of one family of breads, and pursuing it in rye, millet or heavily seeded doughs actively damages them. A tin is not a compromise; for most of the world's great keeping breads it is the correct technology.

Four things "everyone knows" that this research did not support

1. Scalding does not unlock minerals — heat destroys the enzyme that would have. Cereal phytase works best around 45–55 °C and is progressively inactivated above roughly 60–65 °C, so a scald hot enough to gelatinise starch is also hot enough to start destroying it. The one controlled human comparison is unambiguous at the top end: a rye bread whose flour was scalded at 100 °C contained 38 mg of phytate-phosphorus per 100 g — 31 mg in an 80 g portion, 47 mg in a 120 g portion — while the sourdough-dephytinised comparator measured zero. At the 63–65 °C of a traditional held zavarka, loss is partial and time-dependent rather than total. Either way: scald for texture, flavour and keeping; sour the rest of the grist for the minerals. Established
2. Longer fermentation is not monotonically better. Phytase activity in a mature sourdough is already about half that of the parent flour, and long acidic fermentation depolymerises beta-glucan, and beta-glucan's activity depends steeply on its molecular weight — so souring oat or barley fibre hard can cost you the property that made it useful. Established
3. Psyllium does not absorb 8–15 times its weight in water. The two things that have actually been measured are a pharmacopoeial swelling volume of roughly 65 mL per gram, and a water-holding capacity of 3.42 g per gram. The popular figure matches neither, and appears to be marketing shorthand rather than a measurement. The number a baker needs is the working ratio: 1:8 to 1:10 by weight. Contested
4. Sourdough is not reliably low-FODMAP, and it manufactures a FODMAP of its own. Heterofermentative bacteria reduce fructose to mannitol; measured wholemeal wheat sourdough carried 515 mg/100 g dry matter against 33 mg for the non-sourdough control. All rye breads tested exceeded low-FODMAP thresholds under conventional processing — the authors treat low-FODMAP rye as an unsolved problem, not an impossibility. Established

You already have the four things that matter most: a Dutch oven, 1 lb and 2 lb tins, a digital probe thermometer, and a dehydrator. Everything below is what is missing, ranked by whether it actually changes an outcome.

Buy now — under £60 the lot

Buy later, once the programme has told you whether you need it

• Proofing. Ladder, cheapest first: oven with the light on (often 24–28 °C — measure yours, LED bulbs give almost nothing); a cool box with a hot water bottle or a jar of hot water, refreshed twice a day; a seedling mat with an external thermostat (~£40, and this solves ninety per cent of the problem); a Brod & Taylor folding proofer (~£190). Try the seedling mat first. The proofer is worth it only if you bake weekly year-round and can point to loaves the cold is costing you.
• Pullman tin, 10 × 10 cm, 1 kg capacity (~£27–47). Exactly right for a 500 g-flour rye, and gives the flat-topped brick that slices thinly. Buy once you know you bake rye regularly.
• Banneton. One, not a set — a 500 g smooth round is about £12–15. A bowl lined with a linen tea towel rubbed with 50:50 rice flour and plain flour does the same job for nothing.
• Grain mill. Mockmill 100 around £224, Mockmill 200 around £274–329 depending on retailer — a £55 spread, so shop around. Your existing hand grinder plus blender will make sprouted-grain flour: the dehydrator is the part that matters and you have it. The grinding is the weak link — expect a coarse meal, sift it, and use the tailings as a scald. If you find yourself doing that monthly, that is the argument for the mill.

Do not bother

A pH meter. For rye, pH is scientifically the right variable — you want the dough below about 4.5 to suppress amylase. But cheap meters drift, need calibration and storage solution, are awkward in stiff dough, and are usually abandoned within a month. pH strips are worse than useless against brown dough. Get there instead by building a generous, properly ripe sour and reading its taste and smell. Revisit only if you are baking 100% rye repeatedly and getting gummy crumb you cannot diagnose.

Also skip: banneton sets, "sourdough starter kits" (flour, water, a jar), expensive lame handles, and xanthan gum — psyllium does the same job better in a wholegrain context and without the slick mouthfeel.

Availability and prices checked 30 August 2026. Flour prices move; treat them as indicative.

Shipton Mill — the backbone

No minimum order, £6 delivery under £30 and free above it (up to 100 kg), so build a £30 basket of 1 kg bags. Old /flour-direct/ links now 404 — use the /collections/ pages.

Hodmedod's — pulses, British grains, pulse flours

The standout for anything British-grown and non-wheat, and the only reliable source of British pulse flours.

• Fava bean flour, organic — £2.99/500 g, £5.38/kg. The one that matters for sourdough: a few per cent lifts crumb and colour, and it is the traditional British improver.
• Lentil flour £3.99/500 g; green pea, yellow pea and wrinkled pea flours £2.99/500 g.
• Emmer wholemeal £3.49/kg · rye wholemeal £2.99/kg · spelt wholemeal £2.99/kg · naked barley flour and red millet flour from £4.99 · YQ population wheat from £4.99 · wheat bran £1.69/500 g.
• Grains: naked oats (groats) from £1.99, kibbled naked barley from £1.99, pearled naked barley £2.19, naked barley wholegrain £1.99–2.69.
• Pulses for sprouting: speckled green lentils £4.95/500 g, olive green lentils £4.35/500 g, whole fava beans £2.49/500 g, carlin peas £2.19/500 g.
• British-grown chia from £3.49 — genuinely grown in Essex. Brown linseed £2.24.

Flagged on the day: rye wholegrain and British chickpeas were showing out of stock, and Hodmedod's do not carry mung beans.

Everyone else, and what each is for

A first order

Shipton (~£30 for free carriage): dark rye ×2, emmer, einkorn, wholemeal spelt, extra-coarse wholemeal, chopped rye, semolina. Hodmedod's: fava bean flour 500 g, naked oat groats, kibbled naked barley, whole green lentils, British chia. BakeryBits: Diax diastatic malt, flexible scraper, bench knife. Then: psyllium whole husk 500 g, a 0.1 g scale, a 2 L clear tub. Millets, halim, sabja and gond katira from an Indian grocer in small quantities only.

Flour is the largest uncontrolled variable in home baking, and the commonest reason a recipe that worked last month fails this month.

Why the same formula behaves differently

• Protein. UK strong whites are typically lower in protein than North American ones — the single commonest cause of an American recipe failing in a British kitchen. One experienced baker reports needing to drop hydration by 8 percentage points, from 60% to 52%, when moving to a particular Shipton white. Whether or not that number suits you, an 8-point swing between two white bread flours is the figure to internalise. Practitioner
• Stone-ground versus roller-milled. Stone-ground runs hotter, produces more damaged starch and a wider particle-size spread. It takes up water fast at mixing and keeps drinking. A stone-ground dough that feels right at mix is usually too dry forty minutes later.
• Freshly milled flour absorbs less, not more. Millers report fresh wholemeal wanting about 70% where aged wholemeal takes 80%, and warn that water can leach back out during rests. It also ferments faster, with a weaker gluten network — so the overproofing risk after shaping is raised. Practitioner
• Harvest. A wet British harvest raises sprout damage and enzyme activity, giving stickier crumb and darker crusts from the same recipe. Large millers blend to a spec; small stone mills blend less.

The fresh-flour ageing argument, unresolved

Experienced bakers genuinely disagree. One camp holds that "green" flour needs to oxidise, reporting better behaviour after two to three weeks and sometimes six to eight. Another bakes successfully with same-day flour. The most interesting datum is from a baker who tried the middle: flour aged four to five days gave slack, stretchy, sticky dough with gluten collapse inside an hour. If that replicates, the worst thing you can do is age it a little. Contested

Rancidity constrains the other end: the germ oxidises within roughly two to fourteen days, and King Arthur's own guidance is no more than three days at room temperature. Since you are grinding small batches rather than milling in bulk, the practical answer is simple — mill to order, use within days, freeze anything else.

Storage

Making high-extraction flour at home

Sift wholemeal through a coarse kitchen sieve, weigh what stays behind, and return it as an overnight soaker rather than throwing it away. You get most of wholemeal's flavour and enzyme activity with a larger share of its gluten potential — a sweet spot practitioners consistently report. Note that a fine sieve removes only a few per cent and is barely high-extraction at all; you want a coarse mesh removing 8–15%.

A wholemeal rye mother, built from flour, water and the organisms already on the grain. No kefir, no brine, no yoghurt, no fruit, no commercial yeast — those are later experiments, not shortcuts.

Why rye

Rye carries the highest phytase activity and the richest microbial load of the common cereals, and a stiff rye mother tolerates neglect far better than a liquid white one — which matters in a cool British kitchen where everything runs slowly. Established Practitioner

The build, day by day — and the failure everyone hits

Feed ratios — what each is actually for

There is a real disagreement here: King Arthur states flatly that ratio affects only when a starter peaks, not how it bakes. Debra Wink and Tom Cucuzza both hold that carrying less forward and feeding bigger shifts the yeast-to-bacteria balance. Wink's rule is "feed less, not more" — meaning less frequently, at a larger ratio — when bacteria dominate. Contested

Reading a starter without the float test

King Arthur now calls the float test "simply too unreliable", and Maurizio Leo calls it prone to false positives. Stiff starters sink when perfectly ripe; liquid ones float when already collapsing; stirring before sampling destroys the result. Use these instead:

• Rise and timing. Reliable doubling within eight hours at room temperature is the working definition of mature.
• Peak, liquid mother: domed, then the dome just beginning to flatten; tide marks streaking the jar wall where it topped out; surface bubbles turning from small-and-many to large-and-few.
• Peak, stiff mother: it does not dome or streak the same way. Peak is the surface cracking open in fissures, and a mass that feels airy and shreddy when torn.
• Aroma progression: sweet and wheaty → mildly fruity or yoghurty → vinegary → acetone. Acetone does not only mean hungry — it can equally mean too warm (above 27 °C), too wet, or too acidic. Fix in that order: cool it, go to 1:5:5, reduce hydration, loosen the lid.

Steering the acid profile

This is exactly the logic of the German Detmold three-stage rye sour, which grows yeast in a wet warm first stage, acetic acid in a stiff cool second, and lactic acid in a warm wet third. Note one live disagreement: Peter Reinhart said in 1998 that a firm mother favours lactic, then reversed himself in 2001. The Detmold tradition, Hamelman and the SFBI all say firm-and-cool gives acetic. Take firm = acetic as the working position, and know the history. Contested

Maintenance in a cool kitchen

• Working schedule: keep about 120 g of stiff rye mother (roughly 60–65% hydration, a thick paste) in the fridge. Take it out, refresh, and let it show doubling at room temperature for 12 h before it goes back. Refresh at least every two weeks; weekly is better if you are baking twice a week.
• Warming it: a hot water bottle inside a zipped cool bag with the jar, refreshed twice a day, holds the low-to-mid twenties. This is the cheapest reliable solution and it is what working bakers in cold houses actually do.
• Reviving a neglected mother: pour off hooch, scrape the discoloured top layer away, take 6 g from the pale layer, feed 50 g flour + 50 g water, repeat every 12 h. If sluggish after two to three days, go to three feeds a day discarding two-thirds each time. Use white flour for the rescue, then reintroduce 5% wholegrain. Success is doubling between feeds within two to three days; abandon after two further days of nothing.
• Red flags: black hooch is normal and not death. Pink or orange streaks, or any fuzzy mould, means start again. So does a smell of cabbage or sulphur that does not clear after two feeds.

Dough temperature is the master variable. Almost every "my dough never rises" complaint from a British kitchen is this and nothing else.

At 21 °C, yeast activity is roughly half what it is at 27 °C. At 17–18 °C you are looking at two and a half to three times a 25 °C schedule. A four-hour bulk becomes ten to twelve hours. That is the honest baseline. Practitioner

The water temperature table

The familiar baker's rule — water = (3 × DDT) − flour − room − friction — treats flour and water as equal thermal masses. They are not: water's specific heat is more than twice flour's. In a 17 °C kitchen aiming for 26 °C, that rule calls for 43 °C water, which actually lands the dough at about 29 °C. The table below is computed by thermal mass instead, for the standard formula in this manual (400 g dough flour, 220 g levain, ~260 g dough water, hand-mixed, friction +1 °C).

Read down to your kitchen temperature, across to the dough you are making. Calibrate after three bakes: if your dough consistently finishes 2 °C warm, subtract 2 °C from the water. Warm the bowl and the flour too — a 2 kg bag at 15 °C is a large thermal mass.

Cool-kitchen practice, ranked

1. Oven with the light on. Typically 24–28 °C, but measure yours before trusting it — LED bulbs give almost nothing. A tray of just-boiled water adds humidity and a few degrees.
2. Cool box with a hot water bottle, refreshed twice daily. Holds the low-to-mid twenties. Cheapest reliable option, scales from a starter jar to a whole dough tub.
3. Cool box with a 1 L jar of 50 °C water, swapped every three to four hours. Finer control; lifts a 25 L box roughly 5–7 °C above ambient.
4. Seedling or reptile heat mat with an external thermostat, around £40. The reliable answer. Never run a mat without a thermostat — unregulated mats exceed 40 °C and will kill a levain.
5. Do nothing, and lengthen. Perfectly valid: bulk at 18 °C for 10–14 h to a 75–100% rise, then shape and retard. Flavour is generally better; only the schedule suffers.

The default sequence for Family A. Every wheat recipe card is a variation on this, and the card only states what differs.

1. Levain, the night before. 20 g ripe mother + 50 g wholemeal rye + 50 g wholemeal wheat + 100 g water at 30 °C. 10–14 h at 20–22 °C. Ripe when domed and just flattening, streaked, smelling of yoghurt and green apple.
2. Fermentolyse. Flour + water + levain, mixed to no dry flour. Rest 30–45 min. (Salt held back; this is a fermentolyse, not a true autolyse — see §22 for why the distinction matters and when to test it.)
3. Salt. Add with a splash of reserved water and squeeze through until no grit remains. Take dough temperature now and write it down.
4. Bulk with folds. Into a straight-sided tub, level marked, 35 g aliquot pulled into a narrow jar and pressed free of air. Three or four sets of coil folds at 30-minute intervals, then leave it alone.
5. Bulk endpoint. Read §19. Percentage rise on the aliquot, plus a domed edge, a whole-mass jiggle and bubbles through the side wall.
6. Pre-shape, bench rest 20–25 min uncovered, final shape.
7. Proof: 1–1½ h at room temperature, or 8–14 h at 4 °C. Cold retard is the default for boules in this programme — it is more forgiving and it moves the bake to the morning.
8. Bake per the card. Probe. Cool on a rack.

How much to handle it. This is grain-dependent and the ranking inverts. High-bran wheat rewards mechanical development, because there is a gluten network to build. Einkorn, emmer and high-rye do not, because there is not — working them only smears. Wholemeal wheat: three to four fold sets. Spelt: two, gentle. Einkorn and emmer: one, or none. Rye: mix only.

A tin is not a lesser vessel. It carries the structural load, which means it forgives a dough whose own gluten cannot — and that describes rye, high-millet, heavily seeded and ancient-wheat doughs, which is to say most of this programme.

Sizing

The fill rule: dough should occupy no more than two-thirds of the tin after shaping. If a formula gives you more than about 1,050 g for a 2 lb tin, take 80–100 g out and bake it in a greased ramekin as a canary roll — cut that after two hours to preview the crumb without violating the 24-hour rest on the loaf. This is the single most useful habit for rye.

Method

1. Grease well; line the base for anything above 50% rye or with a scald.
2. Flatten to a rectangle the width of the tin, roll up tightly along the short edge, seal the seam, place seam-down, and press into the corners. For batter-doughs, scrape in and level with a wet spatula.
3. Proof endpoint: the dough crowns 1.5–2 cm above the rim and the surface loses its shine. Tins tolerate substantially more proof than a free-standing loaf, so err later here than you would with a boule.
4. Bake on the middle shelf with a tray of boiling water on the oven floor for the first 15 minutes, then remove it and drop the temperature. Tent with foil if the top darkens before the centre is done.
5. Turn out immediately onto a rack. Leaving a loaf in its tin steams the crust soft.

Rye is not weak wheat. Treating it as such produces the gummy, sticky, wet-streaked crumb that convinces people they cannot bake rye.

The mechanism you must understand

Rye has no functional gluten. Structure comes from pentosans — arabinoxylans — which form a viscous water-binding gel. Acidity is not a flavour decision; it is enzyme control. Rye starch gelatinises at a lower temperature than wheat's, and that range overlaps alpha-amylase's activity window: the enzyme is most active roughly between 60 and 77 °C and is not fully inactivated until about 82 °C — exactly the range the crumb passes through during baking. Unchecked, it dismantles the starch framework faster than the starch can set, and the framework cannot then bind the dough's water. Alpha-amylase's pH optimum is around 5.0–5.5, and its activity falls away steeply below that, becoming negligible by about pH 3.5. That is why the target is around pH 4.3–4.5, and why a young or under-acidified sour cannot make good rye however well it has risen.

Two honest qualifications. Those temperature and pH figures come from the rye-baking technical literature rather than from a primary enzymology paper, so treat them as a working model rather than as measured constants — the direction is not in doubt, the exact numbers are. And acidity is not doing only this: lower pH also increases pentosan solubility and water binding, which is the other half of why rye needs a sour at all. Practitioner Traditional

The rules

• Preferment a lot. 25–35% of the rye for a light rye bread, 40% for a rye-dominant mixed bread, 50–60% for a wheat–rye. This manual uses 50% at 70% rye and 40% at 100% rye with a scald.
• Mix, do not knead. No folds. Two to three minutes with a dough whisk. There is nothing to develop and working it smears the gel.
• Wet everything — hands, spatula, the top of the loaf. Constantly.
• The proof endpoint is visual and unmistakable: the floured surface develops cracks at least 1 cm wide across the whole top. When they are open, bake immediately. Rye gives no second chance — the cracks close again as it over-proofs.
• Falling bake profile, not optional. 250 °C to set the crust, falling to 200 °C and then 180 °C, for a 60–90 minute total. This is the tradition across Latvia, Russia, Denmark and Germany, and it dries the crumb through without burning the outside.
• Internal 96–98 °C — lower than wheat.
• Rest 24 h at 60% rye, 48–72 h at 100%. Three converging reasons: amylase remains active until late in the bake; the arabinoxylan gel and retrograded starch must set; and moisture must equilibrate from crust to crumb. Cut it warm and you get a smeared face and a wasted loaf.

The one-stage sour (Detmold), and where it comes from

Total dough yield 180–200 — that is 80–100% hydration — with 10% mother on the sour's flour, held 15 h at 26 °C or 20 h at 18–22 °C for a more acidic result. Its virtue is high processing tolerance and a fairly constant acid level, which is what a home baker wants.

The three-stage Detmold system is more aromatic and much less forgiving. Its stages have distinct jobs: a wet, warm Anfrischsauer at 100–110% hydration and 25–26 °C for 5–8 h grows yeast; a stiff, cooler Grundsauer at 50–70% and 23–28 °C for 6–10 h grows acid; a wet, warm Vollsauer at 80–100% and 25–32 °C for 3–10 h restores leavening power. German sources describe its weakness bluntly: geringe Verarbeitungstoleranz — very low processing tolerance. Try it once you have the one-stage working.

Note two things for honesty. First, the stage names conflict between sources — the IREKS compendium calls them Grundsauer / Mittelsauer / Vollsauer. Second, and more important: Detmold, the Berliner Kurzsauer (1942) and the Monheimer Salzsauer (1958) are twentieth-century cereal-institute standardisations, not folk practice. They are authoritative technical sources. They are not ancient. Traditional Contested

A scald pre-gelatinises starch outside the dough. The gelatinised starch binds far more water, feeds the Maillard reaction with liberated sugars, and slows staling — which is why the great keeping breads of northern and eastern Europe almost all contain one.

The German family, with the ratios that matter

The dosage ceilings are not arbitrary. Gelatinised starch is enzymatically defenceless, so over-dosing a scald into a rye dough hands alpha-amylase a pre-digested substrate and produces exactly the dense, damp crumb you were trying to avoid. Total dough yield including scald water should not exceed about 185–190 — roughly 85–90% overall hydration. Traditional

The Russian and Baltic zavarka, and its one critical temperature

Rye flour, malt and spices are scalded with near-boiling water so that the mixture settles at 63–65 °C and no higher, then held there for two hours. That number is not a preference. Rye starch is gelatinised at that temperature so the amylases can reach it, while beta-amylase — the maltose-producing one — denatures at 65 °C. Above the line you kill the enzyme; below it the starch is not accessible.

Two hours is the whole benefit. Measured sugar content rises from about 10–14% at fifteen minutes to 25–26% at two hours, and no further. Holding a scald longer does nothing, which contradicts the folklore directly. Traditional Established

The fermented scald

Cool the finished scald to 30 °C, stir in a spoonful of ripe sour, and hold at 28–30 °C for 10–12 hours. It becomes fruity and apple-like, and the residual malt sugars buffer the perception of acid so the finished loaf tastes rounder rather than sharper. The acid also accelerates saccharification.

The Ethiopian parallel is worth knowing. In injera, a portion of the already-fermented batter is cooked into a gel and stirred back — the absit step. It is functionally a scald that also re-inoculates and does not dilute the acidity. Two traditions, half a continent apart, arrived at the same solution; one of them arguably arrived at the better version. Traditional

Scalding oats and barley: the coarse-particle rule

Use coarse jumbo oats or kibbled barley, not fine oatmeal, and add the scald to the final dough rather than the levain. The reason is a steep and under-appreciated dose relationship: the modelled minimum dose of oat beta-glucan needed to blunt a glucose response is about 0.2 g for high-molecular-weight, 2.2 g for medium and 3.2 g for low, per 30 g of available carbohydrate — a tenfold to fifteenfold penalty for depolymerisation. Beta-glucanases are active at sourdough pH and temperature, and larger bran particle size measurably inhibits that degradation. Coarse particles, heat to denature, add late, keep the ferment moderate. Established

Read those three numbers as a ratio, not as doses. They come from a meta-regression and are extrapolations — the point at which the upper confidence bound crosses no-effect — not thresholds anyone observed. 0.2 g is well below any dose actually tested, and the authorised European health claim for oat beta-glucan uses 4 g per 30 g of available carbohydrate. What is solid, and what this manual acts on, is the relative penalty: degrade the molecular weight and you need many times more of it. Contested

Treat these as functional ingredients with measurable jobs, and account for their water separately from the dough's — which is why every recipe card in this manual splits the water four ways.

Reference table — dosing, ratio, and what each one actually does

Working rules

• Pre-gel psyllium; never add dry husk to mixed dough. It hydrates in 15–45 minutes in dough, but a pre-made gel disperses evenly and — importantly in sourdough — stops it stealing water from the levain during bulk.
• Whisk psyllium into cold water hard and fast. It lumps within seconds. Whole husk gives an open, bread-like crumb; fine powder gives a tighter, more gelatinous one and is roughly 15% stronger by weight.
• Grind flax coarsely, the same day. Whole seed passes through undigested; flour-fine oxidises and turns bitter. A few seconds in the blender.
• Adding a second mucilage: cut cereal water by 10–15 g and reassess at the first fold.
• Gond katira is for gluten-poor doughs only. Wasting it in strong wheat bread tells you nothing. Reserve it as the psyllium comparison in a high-millet loaf — and note the swelling fraction (bassorin, 60–70% of the gum) is slow, so hydrate it for hours, not minutes. Dry crystals swallowed unhydrated are a genuine choking and obstruction hazard.
• Isolated inulin stays out. Fibre diversity comes first from wholegrains, rye, oats, barley, pulses, seeds and psyllium. Concentrated inulin is a later test with a specific purpose, not a default addition.

When mucilage fails

Gummy, damp, slightly rubbery crumb — dose too high for the water available; cut the mucilage before adding water. Wrinkled crust and a "juicy" crumb — over-dosed; in wholemeal wheat 5% psyllium on flour is optimal and 10% is too much. Killed oven spring — the gel is competing with starch and protein for water, and a too-stiff gel raises extensional viscosity beyond what the cells can expand against. Purple or grey crumb — anthocyanins in the husk responding to pH; use blond husk and avoid aluminium-based raising agents. Sourdough acidity normally suppresses it entirely. Structural collapse above ~90% hydration in starch-based systems — the gel goes from elastic film to slurry.

Two different targets, and confusing them is the commonest error. Flour for bread wants a root barely emerged. Malt for enzymes wants a shoot grown to nearly full kernel length. They are not the same protocol.

Protocols

Dehydration — the temperature that decides everything

This is where most home protocols go wrong, and the error is understandable: amylase denaturation temperatures quoted as roughly 65 °C for beta-amylase and 70–75 °C for alpha are hydrated-system values. In a dry matrix enzymes are far more thermostable. That is why the maltster's rule is dry cool first, heat only once dry — and why the dangerous window in a dehydrator is the first few hours, when the grain is still 40%+ moisture.

Flagged: a widely shared traditional-foods protocol specifies drying sprouted grain at 63 °C for 12–24 h "to preserve enzymes". That temperature, applied to still-wet grain, is above the moist-heat inactivation point for beta-amylase. It makes perfectly good sprouted flour; it is not a diastatic protocol, and calling it enzyme-preserving is wrong. Contested

Endpoint, grinding, storage

• Dry enough? Bite one. It must snap, exactly like unsprouted grain. If it dents, it is not dry. Roughly 10–12% moisture. Run a fan-only cycle for a further 12 h after the heated stage.
• Grinding with what you have. Hand grinder: two passes, coarse then fine, and expect it to be hard work for 500 g. Blender: 60–80 g batches, pulsed to avoid heating, then sieved at about 500 µm with the tailings re-milled or used as a scald. Never mill damp grain — it clogs burrs and can burn out a motor.
• Storage. Freeze, and use within six months. Note an honest correction to received wisdom: a 2023 review found free fatty acids formed more slowly and oxidation products were lower in sprouted grains than raw over twelve months' storage. Milling is what starts the clock, not sprouting. Sprouted whole grain keeps at least as well as raw; sprouted flour still deserves the freezer. Contested

Inclusion rates

Keep this separate from the twenty-bake programme. Attempt it after Bake 20, when your endpoints are reliable and you can afford a failed batch.

Diastatic malt

1. Steep at 10–15 °C in alternating rests: wet 8 h (16 max) → air rest 8–10 h → wet 8 h. Total 48–72 h. Target 42–46% moisture at steep-out.
2. Germinate at 13–18 °C for 3–5 days, turning daily. Endpoint is the acrospire — the shoot under the husk — at three-quarters to full kernel length. This is the brewing standard and it does apply here, because you want maximum enzyme. It is the opposite of the sprouted-flour target.
3. Kiln. Free-dry below 52 °C until moisture reaches 10–12%; only then raise to 60–71 °C to finish below 6%. For maximum diastatic power, the home simplification is simply ≤40 °C for 3–4 h in the dehydrator, then fan-only.
4. Remove the rootlets before milling. They are bitter. Rest the grain unmilled for a few weeks, then grind.

Dosing, and why it goes wrong

Diastatic malt: 0.1–1% of flour, with 2% the outer limit. Dose must scale inversely with the malt's Lintner value: pure undiluted malt runs 140–200 °L, while commercial "diastatic malt powder" is often heavily diluted to 20–40 °L. That is why one brand is safe at a teaspoon per three cups and another is disastrous. Above the threshold: sticky slack dough, damp greyish gummy crumb, rubbery texture, and a reddish rather than golden crust from excess reducing sugars. Wholegrain and home-milled flours already carry substantial native amylase and usually need none at all. Non-diastatic malt is a different thing entirely and is used like sugar, at 1–5%.

Testing whether your malt is actually diastatic

An iodine micro-mash, borrowed from brewing. Make a thin slurry of 10 g plain flour in 100 mL water, heat to gelatinise, cool to about 65 °C and hold. Split into two cups; add 1 g of ground malt to one. Spot a drop of each on a white saucer at 0, 5, 10, 20 and 30 minutes and add a drop of tincture of iodine. Blue-black means starch intact; amber or no colour change means converted. A genuinely diastatic malt clears within 15–20 minutes; the control stays blue-black indefinitely. The malted slurry should also visibly thin and taste sweet.

Red (fermented) rye malt

The full traditional process, which is worth attempting precisely because the ingredient is so hard to buy in Britain:

1. Steep rye 6–8 h at ~18 °C.
2. Germinate 2–4 days at 15–16 °C, rinsing twice daily. Sprouts must not exceed the grain length.
3. Stew — the defining stage. Hold the wet green malt in a closed vessel at 45–65 °C for 2–3 days. This is not mainly microbial fermentation; it is enzymatic release of sugars and amino acids at a temperature where they immediately brown — Maillard and melanoidin chemistry. Native lactic bacteria contribute acidity. A practitioner variant that is easier at home: 40 °C for 24–48 h (longer is more sour), then 50 °C for 12 h.
4. Dry in three stages: 45 °C down to about 30% moisture, then 70 °C down to about 10%, then 80 °C to develop aroma. Roughly two days.
5. Condition unmilled for 3–4 weeks. Remove rootlets before milling.

Result: deep red-brown, non-diastatic, with a chocolate-and-sour aroma. Dose at 1–2% of total flour generally, or 3–5% of the rye when used inside a zavarka as in Bake 19.

Acrylamide caveat. Acrylamide forms from free asparagine and reducing sugars above about 120 °C at low moisture — exactly the conditions of a dark malt roast and a dark crust, and rye is high in free asparagine. Fermentation genuinely helps, because yeast and bacteria consume free asparagine during proofing. Keep home roasting to the lowest temperature and shortest time that achieves the colour you want, prefer the low-temperature stewing route over a hard roast, and bake to colour rather than beyond it. Established

Four non-oven sessions that run alongside the programme. These are not loaves and they should not be judged as loaves — they are the grains' own technologies, and in several cases they are the better use of the grain.

B1 · Kisra — Sudanese fermented sorghum flatbread

Run in Week 6 (starter) and Week 7 (bake). The one genuinely maintainable non-wheat culture in this manual.

• Batter: sorghum flour to water at 1:2 by weight — pourable, thinner than crêpe batter. Start with 200 g flour and 400 g water.
• Starter (ajin): traditionally a retained piece of the previous batch, at about 5% of the flour weight. For your first batch use 20 g of your rye mother — it will be displaced by the sorghum's own organisms within two or three back-slops.
• Ferment: traditionally 12–16 h at 32–37 °C, reaching pH 3.4–4.0. In a 19 °C British kitchen expect 30–40 h, or 16–20 h in a warm spot. Raising the inoculum to 10–20% is a better fix than simply waiting longer.
• Bake: spread very thin on a carbon-steel crêpe pan or large flat non-stick at 160–180 °C — a drop of water should skitter, not explode. One side only, 60–90 seconds. The sheet lifts at the edges and goes matt. Use a plastic dough scraper or the back of a metal ruler as a gargariba spreader, or pour and tilt like a crêpe. Stack the sheets.
• Back-slop: retain 30 g of fermented batter for the next batch. Lactobacillus ratios have been shown to hold constant over four consecutive fermentations, so this is a genuinely stable culture, not a one-off.
• Keep a dried backup. Smear a thin film of active batter and dry it. Sudanese practice sun-dries the ajin, and here is the interesting part: measured side by side, wet household starters carried lactobacilli but no detectable yeast, while sun-dried starters carried both. Drying is what preserves the yeast fraction.

Why it matters nutritionally: fermentation raises sorghum's in-vitro protein digestibility from about 44% to 65% — the kafirin cross-linking that makes sorghum protein hard to digest is partly undone by acidification. That is the strongest nutritional argument for the whole technology, and it is not available from putting 20% sorghum in a wheat loaf. Established Traditional

B2 · Bhakri and rotla — the hot-water dough

Run in Week 8. Jowar, bajra and ragi bhakri/rotla are unleavened, but the dough is made with boiling water — roughly 1 cup flour to ¾ cup hot water, kneaded while still hot, rested to room temperature, then patted out by hand, water smeared on the surface, and finished over a direct flame to puff.

This is a scald in disguise, and it is the point of the session. Gluten-free flours have no protein network; hot water gelatinises a fraction of the starch, and that gelatinised starch becomes the binder. Cold-water jowar dough cracks; hot-water dough does not. It is the same physics as a rye Brühstück, a Japanese tangzhong and an Ethiopian absit — four traditions, one mechanism.

Run the rancidity test at the same time. Buy 500 g of bajra flour, make bhakri on day 1 and again on day 10, and taste them against each other. If the second is bitter, you have learned the most useful practical fact about pearl millet in Britain, and you will never buy a large bag again.

B3 · Ragi ambali, koozh and ragi malt

Run in Week 8–9. The traditional Indian technologies for finger millet are fermented porridges and malted weaning flours, not loaves.

• Ambali: ragi flour slurried, part-cooked, left overnight in a covered pot to sour, thinned next morning, salted, often finished with buttermilk, onion and green chilli. The practice is genuinely traditional; the hour-counts and probiotic claims circulating online are not sourced. Treat it as a practice to learn, not a protocol to cite.
• Koozh is better documented: flour cooked to a gruel, cooled, fermented 24 h; pH falls from 6.5 to 3.5–3.7, with lactic bacteria peaking at 15 h. In a cool kitchen allow 36–48 h.
• Ragi malt, and the mechanism worth stealing. Sprout ragi 48 h at ~30 °C, dry, roast briefly, grind. Then note what a small amount of it does to a thick porridge: in the amylase-rich-food work, adding 0.4–0.8 g per 100 mL collapsed a 10% rice gruel's viscosity from 2,780 cP to 632 and then 312 cP — an 89% reduction — and roughly doubled how much an infant could actually drink. Established

For a baker, that is directly transferable: a pinch of malted ragi or bajra flour will thin an over-stiff porridge or slacken a gluten-free batter without adding water. It is the cheapest viscosity control you have.

B4 · Injera-style with absit — optional, Week 10

Teff (or a teff–sorghum blend, which is itself traditional at lower altitudes) at 1:2 flour to water, with 10–16% ersho — the yellowish liquid retained from a previous batch — fermented 48–72 h. Optimum in controlled work is 25 °C for 64 h; in a British kitchen unassisted, allow four to five days.

Then the absit step, which is the reason to run this session at all: pour off the supernatant, take about 8% of the batter, cook it into a gel for ten minutes, stir it back, and rest for a further 4 h. Bake on a very hot flat pan — 235 °C for 3 minutes or 255 °C for 2 — pouring in a spiral from the outside in, lid on for the final minute.

Domesticated version: whisk 130 g of fermented batter into 250 mL simmering water, cook 30 seconds, cool 25 minutes, slacken with 300 mL cold water, return to the main batter, rest 1–2 h.

Why this is the most transferable idea in the manual. European bakers scald with water. Ethiopian bakers scald with their own already-fermented, acidified batter — so the step gelatinises starch and re-inoculates and does not dilute the acidity. Arguably a better design than a plain water scald, and worth testing in a rye. Traditional

Prep sessions

P1 (Week 6): sprout and dehydrate green lentils for Bakes 14, 15 and 20. Make 150 g of flour — more than you need, because you will want to repeat. P2 (Week 7): mung and chickpea, for comparison. P3 (post-programme): diastatic barley malt. P4 (post-programme): red rye malt, if you could not buy any.

Twenty oven bakes over eleven weeks at roughly two a week, plus four bench sessions and two prep sessions that run alongside. One major variable at a time, on a 500 g flour basis throughout, with explicit controls.

Before Bake 1 — a fortnight of starter

The programme does not begin with a loaf. Build the rye mother first (§5) and bake nothing until it doubles reliably within eight hours at room temperature. That takes seven to fourteen days. Starting the bakes on a starter that is not yet ready makes the first three loaves uninterpretable and wastes the calibration phase.

Four deliberate departures from a purely sequential plan

• Bakes 3, 7, 12, 13, 14, 16 and 17 are twin bakes — one session, two 250 g-flour loaves in 1 lb tins, baked together. Seven sessions therefore deliver fourteen conditions, which is how twenty bakes covers this much ground. A side-by-side comparison in the same oven on the same day is also a far better experiment than two loaves a week apart.
• Emmer and einkorn get their own session (Bake 7) rather than being folded into a single "ancient grains" bake. They behave differently enough that lumping them together would teach you nothing.
• Bajra and kodo millet are handled in the bench sessions, not as loaf additives — which follows the programme's own principle that a grain should be treated according to its own traditional technology. Both belong in hot-water flatbreads and fermented porridges, and bajra's two-week rancidity clock makes it a poor candidate for a slow experimental loaf anyway.
• Ragi and sorghum each get a two-way bake (16 and 17): once as a 20% guest in a wheat dough, once as the substance of the bread with psyllium doing the structural work. This is the programme's central thesis put to the test twice.

The calendar

The four families

Every formula below was computed and validated rather than typed — flour percentages sum, preferment flour is drawn from the stated basis, and water is accounted in four separate streams. The dough weights are what you will actually get.

Everything below is done the evening before the bake. If you can only remember one rule: the levain goes on last, so that its ripeness lands where you want it in the morning.

Not the day before: psyllium gel (20 minutes ahead on the day) and halim soaker (2–4 hours ahead). Both are made fresh.

Three templates. Individual cards state only what differs from these.

A · Wheat loaf, cool kitchen, same-day bake

In a warm kitchen (24–26 °C) compress the bulk to 4½–6 h and start correspondingly later. Do not compress the bulk and keep the cool-kitchen rise target — read the endpoint, not the clock.

B · Wheat loaf with an overnight retard — the default for boules

Identical to A up to shaping, but bring the shaping forward to late afternoon: mix 07:45, bulk to target by about 16:00, shape 16:30, then straight into the fridge at 4 °C for 12–16 h. Bake from cold the next morning — no need to warm it up. This is more forgiving, gives a better crust, and moves the bake to a civilised hour. Note that in a cool kitchen the fridge is a weaker brake than in a warm one, so shape at a slightly later point in bulk than you would in summer.

C · Rye tin loaf

Never the clock alone. Every endpoint in this manual is an approximate time and an observable state, and when they disagree the observable state wins.

Percentage rise is not a property of the dough

It is a property of dough × temperature × what happens next, which is why a single number cannot be borrowed between kitchens. Two anchors:

Self-correction rule: if a loaf overproofs, drop the target by 10 points next bake. If it underproofs, add 10. Three bakes calibrates you.

The aliquot jar

Take 30–40 g of dough after the first fold set. Press it firmly into a narrow, straight-sided jar, eliminating air pockets, and mark the level. Refinement: fill a second identical jar with water to the same level and mark it, so you are measuring volume rather than estimating it. Keep the jar in the same place as the dough — the small sample is more temperature-sensitive and cools faster. Under a 2 °C retard the aliquot will actually fall below its start line, which is expected.

What to look at besides the jar

1. Edges. Domed and rounded where the dough meets the container = ready. Flat or pulling away = under. Sunken or soupy at the edge = over.
2. Jiggle. Shake the tub: a ready dough moves as one puffy mass.
3. Bubbles. Visible at the surface and through the side wall of a clear tub.
4. Surface. Smooth, elastic, with defined edges that resist stretching.

The poke test is for final proof, not bulk. Springs back immediately = under; springs back about halfway, slowly = ready; the indentation stays = over. It does not work on cold dough — straight from a 4 °C retard the reading is meaningless, which is exactly why the endpoint decision for a retarded loaf has to be made before the fridge.

Final proof by vessel

Levain and sour endpoints

Wheat levain, liquid: domed and just beginning to flatten; tide marks streaking the jar; bubbles turning from small-and-many to large-and-few; aroma of yoghurt and green apple. Rye sour: risen, then cracked across the surface; sharply sour and fruity; properly acidic on the tongue, not merely tangy. Stiff mother: surface fissured, mass airy and shreddy when torn — it will not dome or streak the way a liquid one does.

The house baselines

Five rules that cover almost every case

1. Changing flour brand or bag: hold back 5% of the water. Mix, rest 20–30 min, then decide, adding back in 1–2% increments during the first fold. The failure is asymmetric — too dry is recoverable, too wet is not.
2. Adjust the dough water, never the levain, scald or gel water. Those are fixed ratios doing specific jobs. Changing them changes the experiment.
3. Adding a mucilage: leave the cereal hydration alone and account the gel water separately. Adding a second mucilage: cut cereal water by 10–15 g.
4. Fresh-milled or fresh-ground flour: start 8–10 points lower than the aged equivalent, and expect water to leach back out during rests rather than being absorbed.
5. Read the dough at the second fold, not at mixing. Bran and stone-ground flour keep drinking for forty minutes. If it tears rather than stretches at the second fold, add 10–15 g there and record it. If it will not hold a fold at all, carry on, note it, and take 20 g out next time.

Work from the symptom to the question. The key move is separating fermentation faults (wrong amount of gas or acid) from structure faults (gas in the wrong place) from bake faults (heat and steam) — their crumb signatures overlap and they have completely different fixes.

Swapping flours

Swapping functional ingredients

• Psyllium powder for whole husk: use 85% of the weight, and expect a tighter, more gelatinous crumb.
• Chia for flax: works nutritionally at up to 5% of flour, but chia is not an anti-staling agent — measured as having no effect on staling over 15 days. Do not expect the keeping benefit.
• Basil seed gum or cress seed gum for psyllium: only at low dose (0.3–1.5% of flour) and only for dough stability and anti-firming. Neither builds the structural film psyllium does.
• Gond katira for psyllium: reserve for gluten-poor doughs, hydrate for hours not minutes, and expect anti-staling rather than volume. Honestly: it is understudied in bread and you will be generating your own evidence.
• Red rye malt: if unobtainable, 20 g of dark malt extract plus 5 g of cocoa gives approximate colour and some depth — and the loaf is then a scalded rye, not a Borodinsky. Say so.
• Diastatic malt: almost never needed. Wholegrain and home-milled flours already carry substantial native amylase.

Open disagreements, and the experiment that settles each

Where experienced practitioners disagree, the difference is usually explained by a variable neither of them controlled. These are the ones worth resolving in your own kitchen, and several are already built into the programme.

Within coherent culinary traditions, at doses that taste of something. Ten spices in one loaf for an antioxidant claim is not a technique.

Traditional combinations, with the tradition named

Working notes

• Toast and crush. Caraway, fennel, coriander and cumin all gain substantially from thirty seconds in a dry pan and a light crush. Grind coriander fresh — pre-ground coriander is nearly inert.
• Rye takes more spice than wheat. A dose that reads as assertive in a wheat loaf disappears into a dark rye.
• Ajwain and methi are potent and bitter in excess. Start at 2 g per 500 g and go up only if you want more.
• Seeds are ingredients, not decoration. Toasted pumpkin and sunflower at 20–30 g per 500 g add magnesium, zinc and texture; sesame at 15–20 g adds calcium and a flavour that suits both Mediterranean and Indian breads. Black sesame is stronger and worth using where you want it noticed.
• Do not claim a tradition you have invented. Rosemary-and-fennel in a durum loaf is historically grounded. Rosemary-and-ajwain-and-caraway in the same loaf is your own idea, and there is nothing wrong with that as long as it is labelled as such.

Two things worth knowing

Never refrigerate bread. Starch retrogradation is fastest just above 0 °C, so the fridge stales bread faster than the counter. Freeze instead — freezing arrests retrogradation by immobilising water.

And a popular claim that is half right, which is the more interesting case. The advice to freeze bread and toast it "to double the resistant starch" gets the mechanism backwards: RS3 formation needs molecular mobility, which is maximal at refrigeration temperatures, and controlled comparisons of 4 °C, 25 °C and −18 °C find the slowest starch digestibility at 4 °C. Freezing arrests retrogradation rather than driving it.

But the effect itself is real. A human trial found freezing and defrosting lowered the glycaemic response to homemade white bread substantially — incremental area under the curve falling from 259 to 179, and to 157 when toasted after freezing — and concluded that all three procedures favourably altered the glucose response. So: the observation stands, the resistant-starch explanation for it does not, and something else (probably changes in starch accessibility and moisture) is doing the work. Freeze bread because it is much the best way to keep it, and take the glycaemic effect as a bonus with an unsettled mechanism. Established Contested

Reheating

• Whole loaf: 180 °C for 8–12 minutes, ideally with a splash of water on the crust first. Restores a day-three loaf almost completely.
• Frozen slices: straight into the toaster from the freezer. No thawing.
• Rye: do not reheat. It is at its best on days two to five and gains nothing from an oven.
• Freezing well: cool completely, slice, interleave nothing, bag, and press out the air. Whole loaves freeze well too but you commit to thawing the lot.

A bake number completed is not a stage passed. Every bake ends in one of four verdicts, and the default when a result is ambiguous is to repeat, not to advance.

The rules that make the verdicts mean something

1. A combination must beat its best single component on a measured criterion, not merely equal it. If it only equals it, take the simpler bread. This is the rule that stops the House loaf becoming a list of good ingredients.
2. Repeat any bake where you changed more than one thing, including accidentally — a different flour bag counts.
3. Repeat any bake where the aliquot and the dough disagreed and you followed the wrong one.
4. Two consecutive REPEATs on the same variable means DROP the variable, not the programme. Move on and come back.
5. Judge keeping quality at 72 h, not 24 h. Scalds and mucilages are anti-staling technologies; testing them at 24 h tests the wrong thing.
6. Blind-test flavour where you can. Get someone else to plate two slices unlabelled. You will be surprised how often the loaf you expected to prefer loses.
7. Do not advance a phase with an unresolved failure behind you. Bakes 8–11 are meaningless if Bake 4 never worked.

Eight criteria, scored 1–5, entered with each bake in the batch log below. Score before you know what you want the answer to be — ideally on the day after the bake, from the loaf rather than from memory.

Also record, in the notes rather than as a score: digestibility or tolerance if you observe anything, the nutritional rationale you are actually testing, and — importantly — whether you would choose to eat this loaf again. That last question overrides the arithmetic more often than you would expect.

Log every feed during the first fortnight, then whenever behaviour changes. The columns that matter most are room temperature and hours to peak — together they are your kitchen's fermentation calibration curve, and they are what let you predict a bake rather than watch one.

The two fields people fail to record and later wish they had: sprout length at the moment you stopped, and drying temperature. Those two decide whether you made bread flour or malt, and whether the resulting loaf is gummy.

One row per loaf — for twin bakes, one row per half, labelled 07 A and 07 B. This is the record the whole programme depends on: without it, Bake 20 is guesswork.

The formula is Bake 20, in full, with gram weights, water allocation and timings. What follows is the reasoning — why each component is present, and what would have to be true for it to be removed.

Nothing on that list is present because it is a superfood. Every entry names a job. If a component's job was not demonstrated in the bake that tested it, it should not be in V1 — and the fact that this default formula lists all of them reflects what the evidence predicts, not what your results will necessarily show.

Subtract before you add. A V2 that removes a component and loses nothing is a better result than a V2 that adds one and gains a little.

V2 — three candidates, in order of expected value

1. Remove the psyllium entirely and see whether you can tell. It is the component with the strongest theoretical case and the weakest sensory footprint at 1.2%. If the loaf is unchanged at 72 h, you have simplified the formula permanently. If it is measurably worse, you have proved the ingredient rather than assumed it.
2. Rye to 30%, millet out. This tests the programme's deepest question against itself: does flavour and mineral diversity from a third grain matter more than more acid-driven cereal chemistry from rye? Both are defensible. Only your loaf can answer it.
3. Move the spelt entirely into the levain across two feeds, testing the single-practitioner claim from Bake 6 that this eliminates slack spelt doughs. Mechanistically plausible — acidification protects fragile gluten from protease before it has to do structural work — but unreplicated. A clean, cheap experiment.

V3 — the longer branches, once V2 is settled

• Home malt. Both kinds. A little diastatic barley malt at 0.3% for crust and crumb, tested against none; and red rye malt made by the stewing route, which would let you bake a genuine Borodinsky from ingredients you cannot currently buy in Britain.
• The fermented scald, transplanted. Take the Ethiopian absit idea into the House loaf: cook a portion of the already-fermented levain into a gel and return it, instead of scalding plain flour with water. It gelatinises starch, re-inoculates, and does not dilute the acid. As far as this research found, nobody has tested it in a Northern European loaf.
• Bran fermentation as a separate stage. The single highest-yield phytate technique found anywhere in the literature: LAB-fermenting the bran fraction on its own has achieved up to 90% phytate breakdown, against roughly 70% for ordinary sourdough acidification. Sift the bran, ferment it warm at 40–50 °C with a spoonful of levain, and return it. This is the most promising untested idea in the manual.
• High-extraction milling. Sift your own wholemeal to 85% extraction, use the removed bran in the fermented-bran stage above, and see whether you get wholemeal's flavour with more of its gluten.
• The Family C branch on its own terms. A maintained kisra culture, a teff-and-sorghum injera with absit, and a ragi malt you made yourself — judged as their own cuisine rather than as gluten-free bread. On the evidence, this is where the largest untapped nutritional and culinary return in the whole programme lies.
• Vitamin C at the meal, for the high-millet and high-sorghum breads specifically. It is not a baking technique, but the human data are striking: with a high-polyphenol grain, ascorbic acid at the meal restored iron absorption from 2.7% to 13.6%, where iron fortification and enzyme treatment did essentially nothing. A tomato salad alongside a jowar loaf outperforms any processing trick you can do to the flour.

Grouped by what kind of claim each source can support. A traditional source is authoritative about practice and not about mechanism; a mechanistic study is authoritative about mechanism and often silent about whether it matters at the dose you are using.

Traditional knowledge — documented practice

• German rye technique. The IREKS Kompendium on why rye must be acidified (alpha-amylase pH optimum 5.0–5.5, pentosan solubilisation) and on Quellstück, Brühstück and Kochstück ratios, temperatures and dosage ceilings. Plötzblog on the Detmold three-stage and one-stage. Sauerteigführung for the Berliner Kurzsauer (Pelshenke, 1942) and Monheimer Salzsauer (1958).
  Caveat carried into the manual: these are twentieth-century cereal-institute standardisations, not folk practice. Also, the stage names conflict between sources — IREKS uses Grundsauer/Mittelsauer/Vollsauer where Plötzblog uses Anfrischsauer/Grundsauer/Vollsauer.
• Borodinsky and the zavarka. Bench & Bowl's documentary history — no support for the 1812 legend; the name appears around 1922, standardised 1934, first published recipe Auerman 1935. Ginsberg's rendering of Auerman 1935, which contains no coriander and no wheat flour. Plotnikov & Kolesnikov (1940) for the 63–65 °C / 5–6 h production scald.
• Scalding and saccharification kinetics. Ginsberg on scalds — amylase peaking 60–77 °C and ceasing above 80 °C, and a 70 °C hold for 3–18 h. The two-hour saccharification ceiling and the white-versus-red malt enzyme distinction come from Russian technical sources.
• Baltic and Nordic. The Latvian Culture Canon on rupjmaize and saldskābmaize (caraway mandatory in the latter). Danish rugbrød soakers and the wrap-and-wait-a-day rule. Finnish taikinajuuri — the unwashed dough trough as a continuous back-slop, the functional ancestor of a maintained starter.
  Two flags: the widely circulated rugbrød recipe uses instant yeast, which is a modern shortcut; and rieska is an unleavened flatbread, not sourdough rye — a frequent English-language error.
• Mediterranean. The Pane di Altamura DOP specification — lievito madre at 20% with three renewals, 2% salt, water at 18 °C, and a bake that vents for 15 minutes before closing, which is the inverse of artisan steam practice and is what builds the keeping crust.
• Sudan and Ethiopia. Kisra microbiology across three sorghum varieties (1:2 flour:water, 5% starter, pH break at 16 h); Hamad et al. 1992 on wet versus sun-dried ajin and four stable back-slop cycles. Standardisation of injera making — 25 °C for 64 h, 8% absit cooked 10 min, baking at 235–255 °C.
• India. FAO on Asian fermented cereals — the organisms arising during the soaking of the black gram, not the rice. FAO on sorghum and millets — decortication methods, and malted finger millet being considered superior to malted sorghum or maize in India.

Practitioner experience — repeated, convergent, and labelled where single-source

• Starter microbiology as bakers meet it. Debra Wink's Pineapple Juice Solution — the Leuconostoc false rise, the pH 4.5 / 3.5 dead zone, and the acids she tested and rejected. The most cited amateur microbiology in the hobby, and the reason day 3–4 defeats beginners.
• Endpoints. The Sourdough Journey on percentage rise — 30% at 27 °C against 75–100% at 21 °C, and the 20% post-shaping cooldown rise that explains the gap. The aliquot-jar thread, including the live methodological dispute about marking with water. Maurizio Leo on why percentage rise misleads across flours.
• Rye. A 90% rye at 90% hydration — mix only, dough whisk, half-inch surface cracks as the proof endpoint, 24 h rest. Hamelman's rest rule — 24 h at 60% rye, 48–72 h at 100%.
• Ancient wheats. Jovial's einkorn guidance — reduce liquid 15–20%, no kneading, proof to 40–50% only. The einkorn hydration argument (60–85%, unresolved) and the emmer one (60% for less stickiness versus above 68% for elasticity).
• Bran and wholegrain. The four mechanisms by which bran costs volume, including the counter-intuitive point that finer grinding may worsen the mechanical damage. The bran-scald disagreement in full.
• UK-specific. A British baker on Shipton flours — lower protein than expected, hydration cut by about 8 points. BakeryBits on dough temperature. Weekend Bakery's rye starter and the hot-water-bottle-in-a-cool-bag method.
• Single-source Three high-value claims that rest on one named practitioner and have not been independently replicated: the "70% spelt behaves like 80% wholewheat" rule; the all-spelt-in-the-levain fix for slack spelt dough; and the report that flour aged four to five days behaves worse than either fresh or three-week-old flour.

Established research — well replicated, mechanistically coherent

• Phytase and phytate. Leenhardt et al. 2005 — a moderate drop to pH 5.5 already removes about 70% of whole-wheat phytate. Layer-resolved phytase — rye 12,000, wheat 9,300, barley 8,400, oats 2,400 U/kg at peak. Rye bread making by grist coarseness — 99% / 94% / 82% degradation as grain gets coarser, and mature sourdough carrying about half the phytase of the parent flour.
• Absorption versus status — the central distinction. A 2025 double-isotope crossover in rye bread — fractional iron absorption 7.0% → 19.1%, and the finding that scalded bread retained its phytate. Against that: a 2026 systematic review of eight human studies finding no improvement in long-term iron status, with one arm recording ferritin falling from 32 to 27 µg/L; and Brnić et al. 2018, a twelve-week RCT with no status improvement.
• Glycaemia. Östman et al. — vinegar at 18, 23 and 28 mmol acetic acid, with a significant dose-response for both glucose and satiety. The cleanest dose-response in the area. Sandberg et al. 2016 on the rye second-meal effect — glucose iAUC about 20% lower 10–13 h later, with raised plasma butyrate, PYY and GLP-1.
• Beta-glucan molecular weight. A meta-regression over 59 trial comparisons — minimum effective dose 0.2 g (high MW), 2.2 g (medium), 3.2 g (low). And coarse oat bran particle size inhibiting beta-glucan degradation in sourdough bread. Together these are why this manual specifies coarse oats and a short, late scald.
• Resistant starch. Structural basis of RS in bread — ordinary breads under 2.5% RS, and RS3 greater with refrigeration than at ambient or frozen. Liljeberg et al. — 3.0% → 5.4% (long low bake) → 6.6% (plus lactic acid) → 7.7% (plus high-amylose barley). Real, directionally correct, and small.
• Psyllium and mucilage. 5% optimal and 10% too much in wholegrain bread, with absorption rising from 75% to 125%; water-holding capacity 3.42 g/g and structural collapse above 90% hydration; a psyllium gel scald cutting acrylamide 1.53–3.80×.
• Sprouting and malt. Falling number 400 → 112 → 62 over two days of sprouting. Germinated lentil flour at 2.5–10%, optimum 7.5%. Malting — steep to 42–46% moisture, acrospire three-quarters to full length, and the rule that moist heat destroys enzymes so you dry cool first.
• Hazards. Ergot alkaloid exposure. Flax cyanogenic glycosides — risk from up to 30 g of ground flaxseed per meal "can be largely ruled out" for adults and adolescents from 13; whole unground seed and flax oil are excluded from concern; an adverse effect in younger children cannot be ruled out on current knowledge. Cyclopiazonic acid in kodo millet — a storage-mould contamination problem, not a property of the grain. Sprout outbreak data and 3–5 log pathogen amplification.
• Tannins and sorghum. A human dose-response — iron absorption falling 8.5% → 3.2% → 2.7% with rising polyphenol load, restored to 13.6% by ascorbic acid but not by fortification or enzyme treatment. The genetic basis of the pigmented testa, which is why pericarp colour is a poor guide to tannin content.

Emerging and contested

• FODMAPs. Yeast, not lactic bacteria, does the fructan degradation — 83% for a sourdough S. cerevisiae isolate over about eight hours, but only 10–30% for some wild yeasts. And the complication: sourdough manufactures mannitol — 515 mg/100 g dry matter in wholemeal wheat sourdough against 33 mg for the control — with all rye breads exceeding low-FODMAP thresholds regardless of process. "Sourdough is low-FODMAP" is false as a general claim.
• Gluten and coeliac disease — flatly contradicted. Sourdough fermentation did not reduce total immunogenic peptide load, and reactive gluten measured higher in the sourdough than in the control at 130 minutes of simulated digestion — roughly 773,000 against 535,000 ppm. It is explicitly a case study, so read it as one good study rather than as a literature. The sub-20 ppm results in the literature required selected lactobacilli plus fungal proteases acting on pre-hydrolysed flour for 24–48 h — an industrial process, not a kitchen levain. Ordinary sourdough is not safe for coeliacs.
• Amylase-trypsin inhibitors. In-vitro tetramer dissociation after 17 h, with reduced inflammatory signalling — against a published comment reporting that in an IBS clinical trial sourdough showed no advantage over high-ATI yeast bread. In-vitro dissociation is not clinical benefit.
• Phenolics. Feruloyl-esterase-processed high-fibre bread improved phenolic bioavailability and vascular function in a randomised crossover — but rat data show bound ferulic acid from bran is more bioavailable overall than the free compound, because colonic esterases release it slowly and continuously. "Free is better" is too simple.
• GABA — an honest deflation. Selected lactobacilli raise bread GABA to about 27–39 mg/kg against under 10 for controls: roughly 4 mg in a 100 g portion. Oral GABA trials use 100–300 mg, and dietary GABA barely crosses the blood–brain barrier. The claim that GABA-enriched sourdough improves sleep or mood is not supported at these doses.
• Fibre diversity. A meta-analysis of butyrate-production potential found inulin-type fructans raised butyrate producers significantly, resistant starch only trended, and arabinoxylan oligosaccharides showed no overall effect — with baseline microbiota strongly predicting who responded to what. Diversity is a reasonable bet-hedge because responses are so individual, not because diversity itself has been shown causal. The popular "thirty plants a week" figure comes from a cross-sectional citizen-science association, not a trial.

Where this research could not close a question

Four gaps worth knowing about. Glycaemic index values for specific breads could not be verified against the international tables, so all GI numbers here should be read as indicative ranges with roughly ±10 units of between-laboratory variation. DIAAS values for wheat, lentil and chickpea could not be independently confirmed. Published flaxseed lignan content spans more than an order of magnitude depending on analytical method, so no single mg-per-serving figure is quoted. And Polish rye technique is documented in English almost entirely by recipe blogs — the manual therefore gives no Polish percentages, only the caraway tradition, which is well attested.

Eleven weeks, twenty bakes, and an author who has not yet baked them either. Here is what that means for you, and what would be genuinely useful to send back.

What is solid, and what is not

Your data stays in your browser

Your logs save in this browser and sync to the cohort server, so you can start on a laptop and finish on a phone, and so the pooled results mean something. Only Umesh sees your logs; nothing is shared with other participants, and the Export CSV button still gives you your own copy at any time. If you want everything deleted, there is a button at the bottom of this section and it takes effect immediately.

What is stored, and how to delete it

Stored against a random participant id: your three logs, your planner settings, and a record of which screens you used and when — that last one so it is possible to tell whether the schedule planner is genuinely useful or merely decorative, without having to ask you. No name, no email and no password is stored in the app; the invitation link is the only identifier, and the label attached to it is whatever Umesh typed when he created it.

Please do not enter health information — symptoms, diagnoses, how a bread affected you physically. Not in the notes fields, not anywhere. This programme makes no health claims and deliberately collects no health data, and a free-text field is the wrong place for it in any case.

Delete everything I have logged Immediate and irreversible. Your local copy is cleared too.

What would be most useful to send back

1. Where you stopped, and why. This matters more than anything else, and an honest "I lost interest at bake six" is worth more than a polite completion. Most people will not finish an eleven-week programme; the useful question is what broke — time, cost, a failed loaf, boredom, or a stage that felt pointless.
2. Which card's prediction was most wrong. Name the bake and what actually happened. This is the highest-value single piece of feedback in the whole beta.
3. Whether you used the planner or ignored it. If you worked off the static timeline in §18 and never touched Bake mode, say so — that tells me the tool is decoration.
4. Nothing — your logs arrive on their own. They sync as you enter them, so there is no export to remember and no deadline to miss. Room temperature against hours-to-peak across a dozen kitchens is a dataset that does not currently exist anywhere, and you are building it simply by logging. The Export CSV button is still there whenever you want your own copy.
5. What you would have paid, and for what shape. A book, an app, a live cohort, or nothing. "Nothing" is a real and useful answer.

Two things this programme deliberately will not do

It will not tell you bread is medicine. The nutritional reasoning is real and the mechanisms are well evidenced, but the manual is explicit about the gap between improved single-meal mineral absorption and improved long-term mineral status — a gap that eight human intervention studies have so far failed to close. If you are looking for a programme that promises health outcomes, this is the wrong one, and §1 explains why.

It will not chase an open crumb. Roughly half the bakes are tin loaves, several are batters, and two are flatbreads cooked on a griddle. If the goal is a photogenic boule, most of this will look like a detour.

If you only have three weeks

Run bakes 1, 2 and 3 and stop. Two identical reference loaves and a boule-against-tin comparison will teach you your own kitchen's fermentation curve, and that alone is worth more than most of what is sold as a sourdough course. Everything after it is optional.