A Grain Batch Is Not One Quality

In a 2018 study, researchers from Campden BRI and the University of Nottingham measured the protein content of more than 2,100 individual wheat kernels, drawn from around 200 commercial and breeder batches. Across the kernels studied, protein content ranged from as low as 6% to as high as 20%.

That is not simply a comparison across different wheat varieties or harvests. It shows how much individual kernels can differ within grain populations that are ultimately represented by a single average protein figure.

An average is a real, useful number. It just is not the whole population.

What an average does, and what it cannot

An average protein figure, moisture reading or DON result does exactly what an average is supposed to do: it reduces a batch of millions of individual kernels to a single, useful and comparable number. Averages are fundamental to how grain is bought, sold, blended and specified. This is not an argument against them.

What an average cannot describe is the distribution underneath it. Two batches with the same average protein content can be made up of very different populations of kernels: one clustered tightly around that number, another spread across a much wider range. The certificate may show the same average. The grain population can still be different.

Where this shows up commercially

This matters because grain quality is tied to specifications and value. AHDB reported that, in the 2022 UK harvest, only 36% of Group 1 milling wheat samples reached the 13% protein specification. During that market period, milling wheat premiums over feed wheat reached around £74 per tonne, with delivered premiums later reported as high as £93 per tonne.

These figures do not mean that kernel-level sorting would automatically capture that premium. They do show that quality specifications have real commercial consequences, and that the quality delivered by a crop can vary substantially from what the market requires. When a batch is assessed and priced using an average value, understanding the population underneath that average can therefore become commercially relevant.

Beyond wheat protein

The principle is not limited to wheat. In malting barley, variation in protein and kernel plumpness within a lot can affect the uniformity of modification during malting. In durum wheat, the proportion of hard vitreous kernels is already used as a grading characteristic, recognizing that a lot is made up of kernels with different properties.

The question this leaves open

None of this means averages are the wrong tool, or that existing quality assessment is broken. An average was simply never designed to answer a different question: what does the population of kernels inside this particular batch actually look like, and does that matter for what happens to it next?

For most of the history of grain handling, answering that question at industrial scale has not been practical. What becomes possible when it can be answered is the next part of the story.

Sources

Caporaso, N., Whitworth, M. B. & Fisk, I. D. (2018). Protein content prediction in single wheat kernels using hyperspectral imaging. Food Chemistry. DOI: 10.1016/j.foodchem.2017.07.048.
AHDB. Grain Market Daily reporting on UK Group 1 milling wheat protein specification compliance and milling wheat premiums, 2022–2024. Market figures are season-specific and included as commercial context.
NDSU Extension. Guidance on malting barley quality, including protein and kernel plumpness. Used as applied industry context.
Durum wheat grading literature and CWAD grading practice on hard vitreous kernel percentage. Used to illustrate kernel-level grading in durum.