Small grain is a concern for wheat growers in water-limited environments worldwide. Following from our preliminary results that cultivars can differ for small grain screenings despite similarity of average grain weight in wheat, grain size distribution was investigated with the aim of identifying alternative parameters that could more reliably discriminate cultivars for their tendency to screenings without actually subjecting breeding material to harsh environments.

Pre-cleaned harvest samples (using screen with 1.5-mm-wide holes) of 5 cultivars from 2 field cultivar × time of sowing experiments were divided into 5 grain-width fractions using 4 sieves with holes from 2.5 to 3.4-mm wide. The experiments conducted in the Western Australian wheatbelt experienced a dry finish in 2000 and a relatively wetter year in 2003. The grain that was separated into each width fraction was weighed and the average weight of grains for each fraction (called the size-specific weight) was derived from a subsample.

The weight of grain (expressed as a percentage of the whole) collected between the 2.8 and 3.1-mm screens (fraction F3) was highly and negatively correlated with small grain screenings (r = –0.85; P < 0.001), thus strengthening the perception that screenings percentages may follow a certain form of grain size distribution. Among various grain size distributions, the skew-Laplace distribution fitted on grain weight was found to be the most suitable to characterise the cultivars. This provides a general approach in the examination of a grain size distribution, which is invariant to combinations of sieve sizes that different individuals may choose.

We conclude that: (i) parameters µ and 1/α of the skew-Laplace distribution based on grain weight separated into each fraction can be used as a selection tool in predicting propensity for small grain screenings in wheat without having to subject breeding material to harsh conditions; (ii) higher values for both µ and 1/α are desirable but if the cultivar has small µ, a high 1/α is required to reduce screenings; (iii) breeders could consider the grain size distribution in addition to grain weight as a breeding objective; and (iv) millers should be looking for higher average grain weight and higher weight per unit grain length especially at small width fractions, to maximise flour yield. Hence, we anticipate that future research into aspects of starch packaging such as grain shape and endosperm density will optimise grain size distribution, grain weight per unit grain length, and eventually the flour yield per unit land area.