Automated Determination of Fat in Milk: Preliminary and Collaborative Studies

Abstract

Abstract The principles and operation of an Auto-Analyzer system are described and the system was compared to the Babcock method. Initial calibration studies verified the manufacturer’s claim of linear response between 2 and 6% fat. Homogenization and heat treatment caused slightly higher and lower results by the AutoAnalyzer system and the Milko-tester, respectively, when raw milk was used for calibration. Use of formalin as a preservative resulted in a positive bias of less than 0.04% fat. Mercuric chloride and potassium dichromate caused no effect. Standard deviation of the difference from the Babcock method for the calibration studies was less than 0.05%. The standard deviations ranged from 0.031 to 0.077 for 345 raw milk samples from individual dairy farms. Collaborative studies encompassed 15 samples ranging from 3 to 6% fat, including 3 blind duplicates plus calibration standards. Samples were analyzed in triplicate by 6 laboratories, using 2 methods of AutoAnalyzer calibration. The standard deviation was less for the AutoAnalyzer than for the Babcock method (0.048 and 0.047% fat, respectively). Differences between 3 pairs of blind duplicates were 0.013, 0.021, and 0.012% fat by the Babcock and AutoAnalyzer analyses, respectively.