Milling Ops: Protein’s impact on gristing decisions

July 3, 2026 user

By Jeff Gwirtz

Millers across the world are impacted by wheat crop production issues, carryover, and market price impacts at both the time of harvest as well as futures pricing.

A mill in a wheat-producing country may be located in a wheat-producing area (an origin mill) or a destination mill located near the flour market it serves. Alternatively, wheat may be imported from a different country. In either case, wheat selection and its impact on flour quality is an important consideration.

Wheat and flour quality is subject to many factors impacting wheat availability, selection and cost. Wheat grade and protein content often drives wheat purchase decisions. It generally is accepted that both quantity and quality of protein are important protein considerations even though quantity alone is generally identified in purchase contracts.

Wheat development programs around the world focus on developing and releasing wheat varieties that offer improved agronomic traits as well as protein quality. Controlled research identifies wheat varieties for release and production, offering in addition to other positive quality attributes greater loaf volume for each incremental increase in protein content as having better protein quality. Adjusting grist protein level to improve performance does not always achieve the expected or desired result.

The goal of this article is to examine why commercial changes do not match more closely with what we learn from research.

Protein content and loaf volume

In the 1948 article “Loaf volume and protein content of hard winter and spring wheats,” (Cereal Chemistry 25:291) Finney and Barmore found that protein content and loaf volume were linear, and that the slope of protein content and loaf volume varied from one variety to another. The slope of loaf volume and protein quantity for a variety is a measure of protein quality. Obtaining wheat and resulting flour from the same variety required production of wheat in various locations under less-than-controlled conditions but nonetheless served to characterize the variety.

In 1992, He and Hoseney published “Effect of the Quantity of Wheat Flour Protein on Bread Loaf Volume,” which reported a method to produce flour samples at different protein levels using the same protein by isolating starch from a flour and returning it to the original flour to produce flour at different protein levels. The average and standard deviation of bread loaf volume was reported for flours at 7%, 8.5%, 10% and 11.5% protein. Their data was used to generate 1,000 values for each protein level as shown in Table 1.

Credit: ©JAG SERVICES INC.

Figure 1 shows the distributions of generated observations for each protein level using the 1,000 generated observations. The difference between protein level performance is obvious. Using the same data, a scatter diagram with best-fit line and regression equation are shown. The R2 values (coefficient of determination) are high and supporting the notion of increased protein content results in increased loaf volume.

Research in general seeks to conduct experiments where only the variables of interest are modified for study, holding all other potential variables as parameters of the study. In fields of study such as physical science where experiments are tightly controlled, R2 values are generally quite high (0.7 to 0.99) when strongly correlated. In economics and finance, R2 of 0.7 to 0.8 may be significant while in social science and psychology where parameters are even more varied and controllable, an R2 of 0.3 to 0.6 may be considered quite meaningful.

R2 shown in Figure 2 represents the proportion of variance in loaf volume explained by the predictor, flour protein level. The experiments identified earlier were conducted with great attention to preventing parameters from becoming unintended variables, thus increasing result variability making it more difficult to identify the linear best-fit, least squares correlation between flour protein level and loaf volume. Unfortunately, we forget controlled scientifically established relationships observed under controlled conditions do not always follow in industrial reality of flour milling.

Credit: ©JAG SERVICES INC.

The reality of the relationship between flour protein and loaf volume can be observed in Figure 3, which shows the impact of flour protein on loaf volume for hard red winter wheat harvest collected over four years in the same production area. The R2 value shown on the graph is far less than that shown in Figure 2.

A smaller proportion of loaf volume variability is explained by the best-fit, least square linear model shown. The good news is the relationship remains positive in that increased loaf volume generally occurs with increased flour protein. The scattering of observations in Figure 3 explains the fact that selection of a higher protein wheat followed by the production of a higher protein flour will not always result in an increase in loaf volume.

In some cases, loaf volume may decrease. However, the reason for the decrease is not explained by the simple model where loaf volume is a function of protein level. That the relationship observed and reported in research is not always reflected in the industrial reality of milling should not come as a complete surprise. The simple model may not be adequate and the reality of endemic variability in the wheat supply chain is not fully understood, documented or appreciated.

The 6 M’s

Identifying sources of variation that may impact flour quality, and in particular loaf volume, can be presented in an Ishikawa (Fishbone) diagram. Such a diagram identifies the 6 M’s: Mankind, Machines, Measurements, Mother Nature, Methods and Materials. Items listed under each of the M’s should be further expanded to provide enhanced appreciation for the sources of variation and the enormity challenge to exert control along the wheat supply chain. Undoubtedly, factors identified as well as those not identified in an Ishikawa diagram are responsible for the spread of observed loaf volume around any given protein level.

Finney and Barmore reported R2 values for the linear relationship between loaf volume and protein level were 0.57 to 0.90 for hard red winter wheat and 0.79 to 0.98 for hard spring wheat varieties tested over several years. A comparable relationship between protein level and loaf volume was observed by He and Hoseney when flour protein level was modified by using starch from the original flour.

Using data from He and Hoseney’s article, loaf volume observations at any given protein content ranged from 38 to 80 cc. The range of loaf volume observations from the crop survey data in Figure 3 at any given protein content is 240 cc. While useful, belief in a simple model linking wheat or flour protein to loaf volume or other flour quality attributes is not sufficient to assure expected performance of wheat protein changes in the grist is achieved.

Using wheat protein content and grade alone to make purchase or gristing decisions may lead to unexpected outcomes. Sources of wheat grist variation beyond these simple measures must be considered and managed. Managing the wheat draw area to limit environmental and varietal differences in wheat delivered to the mill may improve expected results as well as the response to protein level selection to achieve desired product outcomes. Identification and control of these sources of variation will come at a cost, requiring careful cost benefit analysis.

This article has been republished from The World Grain.