The aim of this research note is to evaluate the variability of the physical measurements obtained by a separator of feeds for dairy cows based on particle size. Fresh samples of total mixed ration (TMR) and corn silage were collected from four dairy units and were immediately fractionated using a particle separator (NASCO®, Pennsylvania State University) composed of two sieves (diameters of 19 and 8 mm) and a collector on the bottom. Repeatability expressed as standard deviation was similar between fractions (>19, 8-19 and <8 mm fractions: ±1.9, ±1.4 and ±2.2% for TMR samples; ±2.0, ±1.7 and ±1.5% for corn silages), but when expressed as coefficient of variation there were, for both samples, large differences between the values for fraction exceeding 19 mm (44.7 and 35.1%, respectively for TMR and corn silages) and the middle (3.9 and 2.5%, respectively for TMR and corn silages) and bottom fractions (3.7 and 5.6%, respectively for TMR and corn silages). The between operator reproducibility was very close to repeatability and this indicates that the operator's contribution to the overall variability is marginal in comparison with residual variability. In conclusion, different operators properly trained in the use of the NASCO® separator can produce satisfactory repeatable and reproducible values for the middle and the bottom fractions.
