Falling Number Sampling Variation Within Trucks at First Point of Sale

Abstract

Cereal Chem. 94(3):480-484Falling number (FN) is a test widely performed on raw samples of wheat and barley as a means to indicate the level of enzyme activity (a-amylase) associated with seed germination. In most circumstances of wheat, high activity levels are associated with decreased quality of the end products and, because of this, grain lots with low FN are discounted at points of sale, including the first point of delivery when wheat is brought in from the field. Confidence in the validity of an FN assessed on a grain lot is dependent on knowledge of the precision of the analytical method as well as the sampling protocol. The current study examined the latter component; that is, the variation caused by sampling. Soft red and white commercial wheats from two seasons and two regions (states of Washington and Ohio) were sampled upon arrival at country elevators and a flour mill. By catch-stream sampling or probe sampling, catches and probings were run in quadruplicate according to standard methodology for FN. Among-catches and among-probings variance estimates, inclusive of sampling error and method error, were typically less than 100 s 2 , corresponding to a coefficient of variation of 3% or less. Estimated variances from the sampling alone, as determined by a mixed-effects analysis of variance, were typically less than 50 s 2 . Based on the amongprobings variance estimates and a simple model that assumes normality of FN and linearity in mathematically combining individual probe readings into the equivalent composite reading for a truck, retesting a truck in borderline FN conditions (i.e., FN £ 294 s) at the same facility will not yield a new FN value of 300 s or greater more than 95% of the time.

Using big data to predict the likelihood of low falling numbers in wheat

“…The single FN test performed on an individual load was at the discretion of the sampler (who often was a seasonal employee). However, we have confidence that the FN data were sufficient for our purpose based on the perspective reported by Delwiche and Vinyard (2017) that harvest sampling did not impact FN accuracy. They suggested that retesting a truck with a borderline FN level of 294 s would, if retested, only achieve 300 s or higher, 5% of the time.…”

As the number falls, alternatives to the Hagberg–Perten falling number method: A review

“…A trained technician can run 100–200 samples per week, starting with the grinding process. Variation in FN data can arise from users’ testing methods, sampling, and FN machine bias, as well as environmental and biological causes (Delwiche, Higginbotham, et al., 2018; Delwiche, Rausch, et al., 2020; Delwiche, Tao, et al., 2020; Delwiche & Vinyard, 2017; Delwiche et al., 2015; Risius et al., 2015). Collectively, in the United States, these issues resulted in industry stakeholders requesting that the USDA Agricultural Marketing Service (AMS) reduce variation in the FN test (Coale, 2019).…”