Studies on Milk Analyses of Individual Cow Milk Samples

Abstract

Differences between infrared (Milko-Scan 104) and reference methods (Gerber and Kjeldahl) in fat and protein determinations were studied in individual cows of four breeds over a 7-month period. The overall accuracy of fat and protein determinations, expressed as the standard deviation (coefficient of variation) of the difference between the methods, was 0.11 percentage units (C.V.=2.3%) and 0.05 percentage units (C.V.= 1.5%), respectively. Fat analyses were significantly influenced by the biological factors breed, cow, stage of lactation and somatic cell count, and protein analyses by breed and cow. The factors studied accounted for about 50% of the total variation in analysis differences. The indirect IR principle for fat determination and its sensitivity to variations in fat composition and fat globule size explained the influence on the IR method of the biological factors tested. The non-correspondence between the IR method and the reference method as regards protein was not only an instrumental effect, but was largely explained by the indirect analysis principle of the Kjeldahl method. Stage of lactation had the most pronounced effect on fat analyses with the IR method. Analyses of Jersey milkfat had the lowest accuracy of the breeds studied. The accuracy achieved (inherent accuracy of the apparatus) was found to be sufficient for milk recording of individual cows, but the need to investigate all relevant factors in a routine analysis system is emphasized.