Proposed: A Malting Sorghum Specification (GFB-MSS, draft v0.3)
Status: proposed, draft v0.3, never ratified by anyone. A discussion document in the same spirit as the Sorghum Malt Evaluation Standard. If you breed sorghum, grow it, sell it, or buy it for malting, and this draft is wrong for your work, we want to know exactly where.
The gap this fills
Four questions sit between a seed and a glass of beer, and they are different questions:
| Question | Answered by |
|---|---|
| What should I grow in the first place? | Seed Traits for Malting — the genetics |
| What grain is worth malting? | This draft — the lot, at the gate |
| What did this malt turn out to be? | GFB-SME — the finished malt |
| What may be called truly gluten-free beer? | GFB-TGF — the finished product |
GFB-SME evaluates malt after it exists. It does not tell a breeder what to select for, a grower what to deliver, or a buyer what to require on a purchase order.
This is not a theoretical gap for us, and the site already shows its shape twice. List Your Grain asks sellers to commit to grain standards, and for the actual numbers it reaches for a worked example on a guidance page — "one program's published gate." Cultivars for Malting sets out the evidence a brewer or maltster should require, and says of its own list that it "is not a perfect procurement standard." Both are right to hedge, because this site has not had a specification to point them at. This draft is the attempt to write it, and to be criticized for trying.
How we would use it
Three uses, and they are the reason this is a specification rather than an essay:
- Sellers declare against it. A lot listed with us arrives with Part 1's declarations attached, and its class stated. That is a paperwork commitment, not a laboratory one.
- Buyers require it. A brewer or maltster can put "GFB-MSS Malting Grade, Identified" on a purchase order and mean something specific by it, instead of renegotiating the same six numbers every time.
- Breeders and seed suppliers can aim at it. A specification that cannot express better than the floor gives a breeding program nothing to select toward. Part 3's premium band exists for that reason.
What it is not. It is not a certification, and we do not audit against it. It is not a pass/fail judgment on a seller — consistent with Lot Reporting Standard doctrine, a lot that misses a threshold is not a rejected lot. It is a lot in a different class, honestly described. An ordinary lot honestly reported conforms fully to Part 1. Silence is the only thing that fails.
Scope condition: non-tannin sorghum only
This specification applies to non-tannin sorghum. Condensed tannins bind the enzymes malting is meant to develop, and the class is set by pigmented testa, not by grain color — it is easy to get wrong by eye. See Tannin and Non-Tannin Sorghum.
A tannin-type lot is not a failing lot here. It is out of scope, which is a different statement, and it needs a specification this draft does not attempt to write.
Part 1 — Mandatory declarations
Following GFB-SME Principle 3: a number without its basis is not a result. Every lot offered as malting sorghum declares the following, whether or not it meets any threshold in Part 3.
No new testing is required by this part. Like the Lot Reporting Standard, it routes paperwork that already exists. A value not on the seller's certificates is declared absent rather than invented.
- Tannin class — non-tannin, tannin, or undetermined, with the basis stated (cultivar record, testa examination, or confirmatory assay).
- Cultivar identity — named, or explicitly declared anonymous. Anonymous grain is a different, lesser class of result.
- Crop year and growing region.
- Grade, with the grading system named.
- Moisture, with wet or dry basis stated.
- Test weight, with units.
- Germination, with the test named and the day count stated.
- Mycotoxin screening — DON and aflatoxin results with methods and limits of detection, or a declaration that no screening was run.
- Storage history — duration, conditions, whether the lot has been dried, and the drying temperature if known.
- Gluten-free chain of custody — the handling evidence for the lot, or its absence.
A lot that declares all ten and meets no threshold is still useful: it is characterized. An uncharacterized lot is not a cheaper version of a characterized one — it is a different product, and it cannot be compared to anything.
Why germination is declared with its method
This is the declaration most often collapsed, and the collapse is invisible once it happens.
A seed-laboratory warm germination count and a germinative energy determination are different claims about the grain. Barley malting uses germinative energy (ASBC Barley-3C: 100 kernels, 9 cm dish, 4 mL water at 20 °C, counted at 24, 48 and 72 hours) and pairs it with a water sensitivity figure from the same test at 8 mL. Those ask how grain behaves under steep-like water conditions, which is a more useful question for a maltster than whether the seed is alive.
An "85% germination" from a lab warm test and an "85% germinative energy" are not the same number. A report that states the method with the number is worth more than either number alone.
Part 2 — Conformance classes
One pass/fail line would force every lot into a binary the grain does not occupy, and would push marginal lots into either silence or overclaiming. Three classes instead:
| Class | Requirement | What it is for |
|---|---|---|
| Declared | All ten Part 1 declarations present. No threshold requirement. | Research, trial malting, and honest trading of grain not selected for malting. Makes real variability publishable. |
| Malting Grade | Declared, and meets every threshold in Part 3. | Production purchase. The class a buyer can write on an order. |
| Malting Grade, Identified | Malting Grade, and named cultivar with traceable seed provenance. | Repeatable programs, breeding evaluation, and any result meant to be compared with another result. |
Identified is the class that makes research possible. A malting result on anonymous grain cannot be reproduced by anyone, including the person who produced it. It is also the class a seed supplier is uniquely able to support, because provenance is the one thing only the seed side can attest.
This implements existing site doctrine rather than inventing it: Cultivars for Malting concludes that the rule is to "preserve identity from grain through malt, wort, and beer, and don't promote an agronomic or laboratory winner directly into production." The Identified class is the first half of that sentence written as a purchasable term.
Part 3 — Thresholds
Adapted from Bard's Tale Beer Company's grain purchase specification — a specification that bought real grain from real growers across multiple crop years, not a committee's estimate. Offered as a starting proposal, and the first thing an adopter should argue with.
| Property | Threshold | Method to state |
|---|---|---|
| Grade | USA No. 1 sorghum | Grading system named |
| Moisture | ≤ 14% | Basis (wet/dry) and method |
| Test weight | ≥ 56 lb/bu | Units as measured |
| Germination | ≥ 85% within 3 days | Seed-laboratory warm germination count, day count stated |
| DON (vomitoxin) | ≤ 0.5 ppm | Method and LOD |
| Aflatoxin | ≤ 5.0 ppb | Method and LOD |
Two of these are deliberately tighter than food-grade minimums, and the reason is the grain's destination, not caution. The mycotoxin limits are strict because the lot is going into a warm, wet, multi-day process — the conditions under which a marginal lot becomes a dangerous one.
And the mycotoxin thresholds are the one pair of numbers on this table with no corrective action behind them. Every other line describes something a buyer can work around or a maltster can manage. A lot over the DON or aflatoxin limit cannot be brought back: kilning is not a validated kill step and cannot be relied on to eliminate mycotoxins, and blending is explicitly not corrective action — mixing a failing lot into passing lots produces a larger problem with worse traceability. Which means compliance with this threshold is not really decided at the gate at all. It is decided in the field, by grain-mould resistance and harvest timing, months earlier — which is why Seed Traits for Malting treats mould resistance as a malting trait rather than an agronomic one. This specification sets the limit; the seed decides whether it can be met. The germination floor is high because germination is the one property that cannot be added back: every other number on this table describes something a buyer can work around, and this one decides whether the grain can do the job at all. A lot at 85% within three days is live and vigorous, not merely technically viable.
Proposed premium band. The source specification targeted 90% and paid a premium at 92% or better. We propose keeping that as a published quality signal rather than a fourth class — a grower who delivers 92% has delivered something materially better, and that should be visible in the price rather than flattened into a pass.
Part 3b — Target profiles: the same grain, two different jobs
Part 3 is a floor. It describes grain that is sound, safe and alive. It does not tell a breeder what better looks like, and a specification that cannot express better gives a breeding program nothing to aim at.
There is plenty of data. The difficulty is that "good malting sorghum" has two different answers, and almost nobody states which one they mean.
- If conversion is bought in — this site's position, where external enzymes own conversion and the malt is a character ingredient — then the grain's job is starch, flavour and cleanliness. Its own enzyme activity is close to irrelevant.
- If conversion must come from the malt, the grain's job includes being its own enzyme engine, and the targets move in the opposite direction on the trait that pays for it: protein.
Every evaluation methodology in the field assumes the second. SABS 235, Novellie/CSIR, the barley-adapted methods, IoB/EBC — all built on malt converts. That premise is so universal it usually goes unstated, which is why cultivar screenings rank on diastatic power and why their conclusions do not transfer to an enzyme-added brewery.
This is not an exotic way to write a specification. The American Malting Barley Association already does it: two-row barley for all-malt brewing targets protein no higher than 11.8%, while two-row for adjunct brewing runs up to about 13%. Same crop, same association, two targets, because the brewing model differs. We are proposing the same structure on a different axis.
The two profiles
The same split, written for buyers. These profiles describe the grain; the same A/B division stated as a purchase decision is Seed Standard 1 — today's process, where external enzymes convert — and Seed Standard 2, a sorghum bred to convert itself. The overlap is deliberate: this page is the specification, those two are the brewer's and the researcher's view of the same choice. Where a number differs, the reason is the audience, not a disagreement.
Acceptable is the line we would buy against. Ideal is what we would most like to see — stated as a direction where the direction is the point, because a grower reading only a range has no way to know which end of it we actually want.
| Property | A — enzyme-added (this site's position) | B — native conversion | Basis for the number | ||
|---|---|---|---|---|---|
| Acceptable | Ideal | Acceptable | Ideal | ||
| Starch | ≥ 65% d.b. | As high as the cultivar will give. ≥ 72% | ≥ 65% d.b. | ≥ 70% | Species-typical range is 65–75% d.b. Profile A's ideal is the top of it, and this is the trait we would trade others for |
| Protein | 9–12% d.b. | 9–10.5% — every point of protein is a point not available as starch | 11–14% d.b. | 12–13% | Species-typical 8–12%; observed median 11.99% across 74 screened samples. AMBA's two-row all-malt ceiling of 11.8% is the nearest published analogue |
| Diastatic power (at 3-day germination) | No requirement | No requirement | ≥ 30 | ≥ 40 | The Madison screening's own working line was that anything over 30 was notable. At 3-day germination its 25 cultivars ran 19.2–45.7, median 31.4, with the ceiling at Pioneer 8282; at 4-day the ceiling was 40.1. Diastatic power barely moves with the extra day, so this target is stable to germination length — but it is stated at 3 days so it is never compared across lengths |
| α-amylase (at 3-day germination) | No requirement | No requirement | ≥ 62 DU | ≥ 78 DU | The 3-day median (62.5) and 75th percentile (77.6) of the Madison screening. Germination-length-bound as well as method-bound — see below |
| Germination | ≥ 85% / 3 days | ≥ 95%, and even | ≥ 90% / 3 days | ≥ 95%, and even | Part 3 floor; B is tighter because enzyme development depends on uniform, vigorous germination, not merely live seed |
| Fat | ≤ 3.5% d.b. | ≤ 3% | ≤ 3.5% d.b. | ≤ 3% | Species-typical 3–4% d.b.; lower favours storage and staling resistance |
| Tannin class | non-tannin | non-tannin, confirmed by assay | non-tannin | non-tannin, confirmed by assay | Scope condition. Colour is not the test |
Profile A's diastatic power row is the whole argument in one line. No requirement, in a field where every published standard leads with that measure. If the enzymes are bought, selecting grain on its native amylase discards good character grain for a property the brewhouse does not use. Every methodology reviewed on this site does the opposite, and leads with that measure.
Why these targets must name their method
Two of this site's own datasets disagree on α-amylase by roughly five to ten times.
| α-amylase | Source | |
|---|---|---|
| Bard's production malt | < 10 DU | Hartwick 2019 panel |
| 25-cultivar screening | 48–120 DU | Madison screening |
Both are real measurements of sorghum malt. They are not comparable, because the laboratories used different methods — and one of the 2002–2004 screening columns is recorded, honestly, as diastatic power in unlabeled units. SABS 235 exported the same problem to the whole field: sorghum DP exists in SDU, °Lintner, °Windisch-Kolbach and ASBC, and they do not interconvert cleanly.
So every enzyme figure in the table above is bound to the method that produced it, and is not portable without one.
And to the germination length that produced it. In the Madison screening, a fourth day of germination raised α-amylase in every one of the 25 cultivars, by 19.2 DU on average, while diastatic power barely moved. The effect on a target is large: the best 3-day result (86.2 DU) is almost exactly the 4-day median (86.5 DU). An earlier draft of this table set Profile B's α-amylase floor at ≥ 86 DU from a percentile pooled across both lengths, which meant a floor that only the single best cultivar reached at 3 days and half of all cultivars reached at 4 — a target that rewarded germinating longer rather than breeding better. Every enzyme target here is now stated at 3-day germination, matching the roughly 72-hour endpoint of the GFB-SME reference malting schedule, and a lot reported at any other length is not comparable to it. Reconciling them is an open ask. Until it closes, a Profile B target quoted without its method is not a specification — it is a number.
What Profile B's targets do and do not claim
They are percentile targets, not performance targets. For diastatic power, ≥ 30 selects roughly the upper half of the 3-day results and ≥ 40 approaches the best single result recorded. For α-amylase, ≥ 62 is the 3-day median and ≥ 78 the upper quarter. Neither is evidence that a lot meeting it will self-convert.
Nobody has established what diastatic power is actually sufficient for an all-malt sorghum mash without exogenous enzymes. Profile B says "the best available", because "enough" is not a known quantity. That is a gap in the field, not a hedge in this draft, and it is an open ask.
Part 3c — Stability: the properties barley's specification does not need
Every number above describes one lot. A breeding program needs to know what is being asked of the cultivar, across seasons — and this is where a sorghum specification has to carry weight that a barley specification can leave out.
Sorghum Protein and the Missing Specification sets out why: barley the crop varies as much as any cereal, and the reason a maltster rarely sees it is that the market filters the variance out before delivery — a century of selection for stability, official recommendation lists, and an elevator that discounts off-spec lots into feed. As that page puts it, the consistency a brewer observes is "the output of a functioning quality system… None of that's a property of the plant."
Gluten-free sorghum has no such system. No recommendation list, no malting premium at the elevator, no rejection pathway. So stability cannot be assumed into the background; it has to be specified.
| Property | Proposed | Status |
|---|---|---|
| Protein spread, same cultivar and region, across crop years | Report the spread, not the mean. Target ≤ 2 percentage points | Proposed target. Hybrid choice alone moved location-averaged protein 10.8→12.7% in US trials, so roughly two points is the variation a cultivar decision already controls |
| Lot-to-lot spread within a crop year | Report rows, not averages | Reporting rule, costs nothing |
| Germination retention through storage | Re-test after stated interval | No target — nobody records storage behaviour as a malting trait |
| Protein response to late-season cold | — | No data. An open ask. Nearest work is the nighttime-temperature and elevation study at UNR |
The last two are in this table without numbers on purpose. A property we want measured but cannot yet put a figure against is not an omission — it is the specification naming its own next piece of work.
Part 4 — Sampling and retention
A threshold is only as good as the sample it was measured on, so sampling is part of the specification rather than an implementation detail.
Sampling follows the FGIS pattern: each delivery sampled representatively and split four ways — one portion graded, one tested, one retained by the buyer, one held for the seller.
The four-way split is the part that matters. A single consumed sample proves nothing once the truck has left, and a grain disagreement without retained material becomes a question of whose paperwork is louder. Retention is what makes it resolvable instead.
Part 5 — Lots that miss
A specification that only describes good grain is not usable at a receiving pit, because the decisions that need rules are the marginal ones.
- Below a Part 3 threshold, fully declared. The lot is Declared class. It may be perfectly good grain for another use, and may still be worth trial malting. It is not Malting Grade, and calling it so would make the class meaningless.
- Declarations incomplete. The lot is unclassified here. It can still be bought — but no claim made about it is comparable to any other lot.
- Germination below floor. The one result we would not work around.
- Unresolved mycotoxin or gluten-free chain question. Hold, not reclassify. This is a safety and integrity question, not a grading one, and moving a lot to a lower class does not answer it.
- Long or unknown storage since testing. Germination and moisture are the two figures that move in storage. A lot re-entering the chain after extended or undocumented storage should be re-tested on those two rather than carried on its original certificate.
What this tells a grower deciding what to plant
Every threshold in Part 3 is measured after harvest. A grower commits in spring and is graded in autumn, so a specification written only as an acceptance gate arrives months too late to be a planting decision.
The companion to this section is Seed Traits for Malting, which sets out twelve traits across composition, character and field resilience — what to select for, what to measure, where selection fails, and what the evidence does and does not establish for each. It also generates a seed selection brief a grower can hand to a seed supplier or breeder. Read together: that page is how to choose, this one is what the grain has to be when it comes back.
Four things here are decidable at planting, and one important one is not.
1. Plant non-tannin. This is the one hard directive. It is knowable at seed purchase, it is binary, and it is not recoverable later. Ask for a breeder declaration of tannin class and a confirmatory assay — grain colour does not establish it, and the tannin trait card names the test to request.
2. Keep the cultivar identified, and keep it separate. Identity is not a laboratory result; it is a handling decision made at planting, harvest, and bin. Commingling two cultivars costs the Malting Grade, Identified class permanently, and that class is the one a repeatable malting program can buy against. Grain that has lost its identity cannot get it back.
3. Protect germination, because it is the property that cannot be added back. Of the six thresholds, five describe things a buyer can work around. Germination decides whether the grain can do the job at all, and it is largely determined by decisions the grower controls: harvest timing and moisture, drying temperature, and storage condition. Heat damages viability, and a lot can look sound while failing a germination test.
4. Documentation is worth money here, and it costs paperwork rather than laboratory time. The declarations in Part 1 are values that already exist on grading and seed-lab certificates. Forwarding them is what separates a Declared lot from an unclassified one, and the premium band exists so that a genuinely better lot is visible in the price.
What we cannot tell a grower is which named cultivar to plant. Published work supports cultivar-level evaluation — Cultivars for Malting reviews three studies finding real differences in malting behaviour across 24, 22 and 8 cultivars respectively — but as that page puts it, the evidence "does not identify a universal best hybrid." No cultivar recommendation appears in this document or on the seed-traits page, and that is a deliberate position rather than an omission. Research Ask 9 states what would change it.
What we can do is make the choice a reasoned one rather than a guess. The seed-traits brief gives a grower the questions to put to a seed supplier, the measurements to ask for, and the specific way each selection goes wrong — without naming a winner. The grower still decides. They decide with a brief instead of a grain name.
The constraint is not land
It is worth stating plainly, because it is easy to assume otherwise: sorghum is not a scarce crop. US growers planted 6.64 million acres in 2025 and harvested 437 million bushels, third among US cereal harvests (Sorghum, Water, and Sustainability). Dedicated gluten-free contract acreage has existed too — a 2019 arrangement covered organic acres in south-central Illinois and near Lubbock, two 40–60 acre blocks in Kansas, and conventional acres in western Kansas (Sourcing Sorghum Farmer-Direct). One organic grower held 12,500 bushels of a named cultivar, more than a small malthouse could take or store.
So the bottleneck this specification addresses is not the number of acres. It is that grain arrives uncharacterized, unidentified, and without a buyer who has said in advance what "good" means. A specification is what lets a grower know, before planting, what the grain has to be and what better is worth.
The boundary of this specification
This specification governs the grain, at the moment of acceptance. It is a property set, measured once, at a gate.
It deliberately stops there. What a maltings must then do to keep that lot gluten-free and turn it into usable malt is a control set applied continuously — modeled station by station in Where Gluten Enters in Malting, whose intake station is this specification's subject. A Malting Grade lot run through an uncontrolled plant does not produce Malting Grade malt, and nothing here claims otherwise. Conformance is a necessary input to a gluten-free malting program, not a sufficient one.
What this specification declines to specify
These are properties a grain specification arguably ought to carry. This draft does not set numbers for them, and says so rather than filling the fields — a threshold invented to complete a table is worse than an acknowledged blank.
1. Germinative energy and water sensitivity. The steep-behaviour methods described in Part 1 are barley methods. This draft specifies the warm germination count instead, because that is the test routinely run on sorghum and reported on its certificates. Adapting the barley methods to sorghum — and publishing where they fail — is the most bounded piece of work on this page.
2. Kernel hardness and size distribution. Both affect steep uptake, germination evenness, and milling. Cultivars for Malting already asks a buyer to record hardness, kernel size, colour and damage; this draft keeps them as declared observations rather than promoting them to thresholds, because we have no basis to propose where the line should sit.
3. Protein — as a universal threshold. The number that most changes what a brewer should do, already measured during grading, and already required by the Lot Reporting Standard. A single threshold would imply one known optimum band, and Sorghum Protein and the Missing Specification is the page about why that band is not settled. Protein stays a declared value in Part 1 and a gate in neither profile.
What changed in v0.2 is the diagnosis rather than the position. A universal band may not be missing because the work has not been done — it may not exist, because the optimum runs in opposite directions depending on whether the malt has to convert. Part 3b therefore proposes protein as a profile-conditional target, not a threshold: something to breed toward, which a lot is never failed against. Both bands are first proposals and are the most arguable numbers in this draft.
3a. Starch. Named in Part 3b as Profile A's first priority, on the basis of the species-typical 65–75% d.b. range, and deliberately not a Part 3 threshold. We have no raw-grain starch distribution across cultivars to set a floor from — our own starch figures are two malt samples. The target says which direction to breed; it does not gate a purchase.
3b. Flavour and aroma. The seed-traits page makes flavour a first-class selection trait, and the Blue Milo decision is the reason — a cultivar chosen over candidates that beat it on starch and extract. But flavour is not a property of grain measurable at a receiving pit. It is scored on malt under GFB-SME, against the vocabulary in GFB-SML — not here. A specification that pretended otherwise would be inviting a number nobody can produce.
3c. Ash, fibre and the rest of the composition panel. Declared where a certificate carries them. No target in either profile: the only figures this site holds are two 2019 malt samples, and two samples of malt cannot support a ceiling on grain.
4. Cyanogenic potential. Sorghum's dhurrin content, and its handling through sprouting and rootlet removal, is a genuine safety question that belongs in a grain specification. This draft does not attempt to write it, and a maltster should not read its absence as a judgement that the question is minor.
5. Cultivar-level malting performance. The largest one, and the only gap whose closing would change the grain rather than describe it. See below.
What would make this a real standard
Adoption, criticism, and iteration — in that order.
For breeding programs and seed suppliers. This draft describes grain rather than directing selection, because the property that matters most — how a cultivar actually malts, under one shared method — has no agreed measurement. Research Ask 9 states the gap precisely: cultivars malted under one standardized protocol, enzyme activities measured by one named method, reported with units and replicates. Beta-amylase and limit-dextrinase are the specific hole. A breeding program that ran GFB-SME's reference mash against its own material would produce the first data capable of turning Part 3 from a description of grain into a selection target — the point at which a specification starts changing what gets planted.
For laboratories. Adapting the barley steep-behaviour methods to sorghum and reporting where they fail would close gap 1. It is bounded work with a publishable result either way.
For growers and sellers. Adopting Part 1's declarations while rejecting every threshold in Part 3 would fix most of the field's comparability problem on its own. The declarations cost paperwork, not laboratory time.
For buyers. Writing a class onto a purchase order is what turns this from a document into a market signal. A specification nobody buys against is a position paper.
This specification is also a set of research asks
Publishing it is how we ask. Every property named without a number is a question put to whoever can answer it, and they are listed as such:
- What diastatic power is actually enough to self-convert? Profile B's targets are percentiles of published screenings, not evidence of sufficiency. Nobody has established the sufficiency figure.
- Can the α-amylase methods be reconciled? Our own two datasets differ by five to ten times. Until they agree, Profile B's enzyme targets are method-bound and not portable.
- Does late-season cold move sorghum grain protein, and by how much? A stability property we can name and cannot yet quantify.
These are filed with the rest of the open work at Research Asks. A specification that listed only what it knows would be a shorter and less honest document.
Version history
- v0.3 — 2026-09-13. Every Profile B enzyme target is now stated at a declared germination length (3 days, matching the GFB-SME reference schedule). The α-amylase targets were corrected: v0.2 set them at ≥ 86 and ≥ 100 DU from a percentile pooled across 3-day and 4-day germination, which made the floor the single best 3-day cultivar and the 4-day median at once. They are now ≥ 62 and ≥ 78 DU, the 3-day median and upper quarter. The error was exposed when GFB-SME v0.7 split the Madison screening by germination length. Diastatic power targets unchanged, with their germination length now stated.
- v0.2 — 2026-09-12. Adds Part 3b (target profiles for enzyme-added and native-conversion brewing, with acceptable and ideal columns), Part 3c (stability properties, and why barley's specification does not need them), the method-binding warning on all enzyme figures, and the research-ask framing. Protein's status is restated: still not a gate, now with a diagnosis for why a universal band may not exist. Starch, flavour and the composition panel added to the declines-to-specify section. The gap table gains its fourth question — what to grow — and the mycotoxin thresholds are connected to the field decision that determines whether they can be met.
- v0.1 — 2026-09-12. Initial draft. Thresholds adapted from the Bard's grain purchase specification as reconciled 2026-09-04; declarations modeled on GFB-SME Principle 3 and the Lot Reporting Standard; conformance classes, the premium band, Part 5, and the grower section are new to this draft.
Where this page comes from
This proposal synthesizes Bard's commercial grain purchasing practice and the site's published grain-acceptance doctrine. It is a draft authored to be criticized. Nothing here has been ratified by any standards body, and the thresholds in Part 3 are one company's numbers generalized — a claim about their origin, not about their universality.
Related
- Seed Traits for Malting — twelve traits, and the selection brief
- Cultivars for Malting — what named-cultivar evidence does and does not support
- What Makes Good Malting Sorghum — the same gates, as reader guidance
- Sorghum Protein and the Missing Specification — why protein is declared, not thresholded
- Lot Reporting Standard — the reporting discipline this builds on
- Grain Source and Acceptance — QA Gate 1
- List Your Grain · Find Grain — the two sides that would use it
- Tannin and Non-Tannin Sorghum — the scope condition
- Where Gluten Enters in Malting — the process model on the other side of the gate
- Sorghum Malting Protocol — what a conforming lot goes into
- Research Asks — the open work, including Ask 9