DisplayTitle

Seed size is an important quality trait for most grains with 1000 grain weight often used as a surrogate measure. However, individual seeds within a sample vary in size, the variation being more pronounced in plants that are indeterminate, such as pulses (grain legumes). A more uniform seed size is visually more attractive and leads to greater consistency in composition and in processing. The seed size distribution method is routinely used as a quality parameter of pulse samples, particularly in breeding programs. Using a sieve stack, percentage (by weight) of grain retained on each sieve is obtained and can be presented in a table or histogram. However comparing histograms across samples is problematic. A single value of within-sample variability would make comparison and analysis much easier. This study examined the seed size distributions of a large set of pulse samples and compared the fit of different methods capable of generating the two parameters of interest; a mean seed size and within-sample variability. The seeds within samples were found to be Normally distributed without skewness or kurtosis and our proposed new method had a good fit to the data. The new method allows calculation of a mean Seed Size (<i>SSnorm</i>) and within-sample Size Variability (<i>SVnorm</i>). The method is most reliable when sieves are chosen so that the sample is retained on at least 5 sieves and it can also be applied retrospectively on existing seed size distributions. In addition, <i>SSnorm</i> and <i>SVnorm</i> were found to be independent of each other; meaning breeding programs can aim to maximize size uniformity without affecting the seed size. This method is suitable for calculating the mean Seed Size and within-sample Seed Variability of any grains that are normally distributed.