Cluster article
Wheat Classes and Specifications: How to Write a Wheat Contract for a Flour Mill
Wheat classes by protein and hardness, the parameters that actually matter on a purchase contract, and the tolerance clauses that decide who pays when a load fails.
Key takeaways
- Contract on five parameters: moisture, protein, test weight, falling number, impurity. Everything else is noise.
- Protein tolerance should be ±0.3% with pro-rata pricing outside it, not a pass/fail cliff.
- Falling number below 200 s means sprout damage — reject, because no milling adjustment fixes it.
The five classes that matter commercially
| Class | Protein | Hardness | Used for |
|---|---|---|---|
| Soft white / soft red | 8.0–10.0% | Soft | Biscuit, cake, pastry |
| Hard red winter | 11.0–12.5% | Hard | Pan bread, general bakery |
| Hard red spring | 12.5–14.5% | Very hard | Hearth bread, strong flour, blending stock |
| Durum | 12.5–14.5% | Very hard, vitreous | Semolina, pasta |
| Local feed / general | 9.0–11.5% | Mixed | Steamed bread, flatbread, noodles |
Local names vary by country — atta wheat, maida wheat, chakki wheat — but the underlying parameters are the same three: protein content, protein quality and hardness.
The five contract parameters
1. Moisture
| Level | Consequence |
|---|---|
| Below 11.5% | Wheat is brittle; more broken kernels, lower extraction |
| 12.0–13.0% | Ideal for milling and storage |
| 13.0–14.0% | Acceptable short term; tempering water reduced accordingly |
| Above 14.0% | Mould risk; price must be adjusted on dry matter |
Always price on dry matter, not wet weight. A 14% moisture load priced at the 12.5% rate gives the seller 1.7% extra wheat for free.
2. Protein
Measured by Kjeldahl (N × 5.7 for wheat) or NIR, expressed at 14% moisture basis.
Write the contract as target ±0.3% with pro-rata adjustment, not pass/fail. A cliff clause means a 0.1% shortfall rejects a whole truck, which costs everyone more than a price adjustment.
| Clause form | Effect |
|---|---|
| “12.0% min, reject below” | Supplier ships 12.05%; you never see 12.5% |
| “12.0% ±0.3%, pro-rata” | Supplier targets 12.0%; you get consistency |
| “12.0–12.6% band, price by formula” | Best for blending operations |
3. Test weight
76 kg/hl minimum for bread wheat, 74 for general purpose. Below 72 the wheat is likely shrivelled or heat-damaged and will underperform in the mill.
4. Falling number
250 s minimum. This is the one parameter with no recovery path.
| Falling number | Meaning | Action |
|---|---|---|
| Above 300 s | Sound, dormant | Normal |
| 250–300 s | Acceptable | Monitor |
| 200–250 s | Mild sprout damage | Blend up to 20% max |
| Below 200 s | Significant sprout damage | Reject |
Alpha-amylase from sprouting degrades starch into dextrins. Bread made from it has a sticky crumb that collapses. Milling cannot remove the enzyme because it is distributed throughout the endosperm.
5. Impurity
| Fraction | Limit |
|---|---|
| Total impurities | 1.5% max |
| Foreign matter (stones, metal, glass) | 0.5% max |
| Other grain | 1.0% max |
| Broken kernels | 3.0% max |
| Ergots and sclerotia | 0.05% max |
Impurity above specification is a cleaning cost, not a milling cost. Every extra 1% of impurity is roughly 1% of intake capacity spent moving material you did not buy.
What is usually omitted and should not be
| Parameter | Why it matters |
|---|---|
| Damaged kernels | Heat-damaged wheat produces dark specks in flour |
| Vitreousness | Affects roll settings and semolina yield |
| Pesticide residue | Regulatory compliance for export markets |
| Heavy metals | Cadmium limits in EU and Japan |
| Mycotoxins (DON) | 1,250 ppb EU limit for unprocessed wheat |
For export flour, DON and cadmium are the two that stop shipments.
Sampling: where contracts are won and lost
The contract means nothing without a sampling protocol.
| Practice | Standard |
|---|---|
| Sampling method | Mechanical probe, 5–8 cores per truck |
| Sample division | Quartering or riffling, never hand-scoop |
| Reference sample | Retained, sealed, 500 g per lot, 90 days |
| Dispute resolution | Independent laboratory, cost to the losing party |
Hand-scooping from the top of a truck samples the driest, cleanest wheat in the load. It is the single most common way buyers get cheated and the easiest thing to fix.