β-Casein from individual buffalo's milk was found to be homogeneous by starch-gel electrophoresis. β-Casein was separated from buffalo's milk by the method of Warner (1944) and purified by DEAE-cellulose chromatography.

Buffalo β-casein possesses identical end-groups to those of cow β-casein; namely N-terminal arginine and assuming a single polypeptide chain a possible C-terminal sequence of Ile-Ile-Val. However, the amino-acid composition and the tryptic peptide patterns of the 2 proteins are not the same.

Although casein micelles are disrupted by removal of Ca, individual caseins remain aggregated in sub-micellar casein aggregates or sub-units. These sub-units have been studied by: (1) the use of gel filtration on Sepharose 4B at 6, 20 and 37°C at pH 6·7 and 0·1 ionic strength, (2) ultracentrifugation and (3) electron microscopy. At 37°C the protein composition of the sub-units varied across the gel-filtration peak, with κ-casein being eluted towards the leading edge and the ratio of αs1- to β-casein being almost constant across the peak. Re-chromatography of the protein from the leading edge of this peak gave a new wide peak with the κ-casein again being eluted towards the leading edge. However, αs1-casein was eluted before β-casein in the leading edge of the new peak. Prior treatment of the casein micelles by dispersion with 6 m-urea solution, precipitation with acid or reduction with 2-mercaptoethanol did not alter the gel-filtration pattern. An examination of the purified casein components and their mixtures showed that a 1:1 ratio mixture of αs1- and β-casein had the same peak maximum elution volume as casein micelle sub-units. κ-Casein by itself eluted at the void volume of the gel-filtration column, but after admixture with a sample of small micelles it eluted at the leading edge of the sub-unit peak and was indistinguishable from the κ-casein normally present. These results suggest that the sub-units are in equilibrium with their component caseins and that their size distribution is determined by only those factors (such as protein concentration, pH, temperature and ionic strength) which determine the strength of association between the casein components. The results from electron microscopy and ultracentrifugation support these conclusions.

Autoxidation was studied in milk obtained from cows fed on formaldehyde-treated casein-safflower oil supplement. The fat of this milk contained high levels of linoleic acid. A rapid disappearance of dissolved oxygen (DO), measured by using an oxygen electrode, from milk samples stored at 0°C in tubes without headspace, coincided with the development of oxidized flavours detected organoleptically. A correlation coefficient of 0·9 with 69 degrees of freedom (P < 0·001) was obtained between the amount of DO disappearing and the taste panel scores for oxidized flavours.

Butylated hydroxyanisole, sesamol, nordihydroguaiaretic acid, ethoxyquin, or butylated hydroxyanisole with propylgallate or tocopherols, when added in emulsified form to the milk at the rate of 10–15 mg/l milk, checked the development of oxidized flavours and the rapid disappearance of DO. Other antioxidants tested were either ineffective or imparted off-flavours to milk.

Samples of mare's milk neither developed oxidized flavours nor showed rapid disappearance of DO over a test period of 8 d.

The oxygen electrode provides a convenient and sensitive method for studying autoxidation and evaluating antioxidants in milk.