With 10-day composite milk samples the effect has been investigated of various preservatives, of variations in the method of preparing the composites and of the condition of the composites at sampling on Milko-tester results for fat content.

The results for composites preserved with mercuric chloride agreed much more closely with the Werner Schmid results than did those for corresponding composites preserved with potassium dichromate, formaldehyde solution or paraformaldehyde.

In composites preserved with mercuric chloride, prepared and held at ambient temperatures at 2 dairy factories, a small proportion contained destabilized fat when sampled at 98–103 °F and gave very low Milko-tester results in comparison with the Werner Schmid and Gerber results. After high-speed mixing of the composites at 115 °F the Milko-tester results were closer to, but still lower than, the Werner Schmid results.

The average Milko-tester result for 133 composites free from destabilized fat when sampled at 98–103 °F was 0–099 ± 0.055% lower than the average Werner Schmid result. Results for 133 duplicate samples, which were also free of destabilized fat at 98–103 °F but which had not been swirled after the daily addition of fresh milk, were on average — 0.158 ± 0.076% lower than the Werner Schmid result. The Werner Schmid and Gerber results agreed well. The lowering of the Milko-tester result in the ‘swirled daily’ samples was due (a) to the effect produced by dissolving mercuric chloride in milk, (b) to slight fat hydrolysis, and (c) to some other factor, possibly fat oxidation. The further lowering of the Milko-tester result in the ‘non-swirled’ samples was due mainly to greater hydrolysis of the fat with production of higher fat acidities. Other composites, which were swirled daily but which were held at 38–40 °F gave on the average higher fat acidities and lower Milko-tester results than the ‘swirled daily’ composites held at ambient temperatures.

A correction of +0.10% to the Milko-tester readings of all ‘swirled daily’ composites, preserved with mercuric chloride and in suitable condition for sampling at 98–103 °F, brought over 90% of the results to within ± 0.10% of the Werner Schmid results.

The Milko-tester (Foss Electric, Hillerød, Denmark) is already in use in a number of countries for the rapid estimation of fat in milk. The fat content is read directly by a photometric measurement of the turbidity due to the fat globules after heating, homogenizing and diluting the sample with a protein-dissolving versene solution. The instrument has been found satisfactory for the routine testing not only of fresh milks but also of milks to which preservative has been added (Černá & Písecký, 1966; Murphy & McGann, 1967). Grubb (1966), however, reported that Milko-tester readings for samples preserved with potassium dichromate were lower after 14 days' storage.

In the present paper the results are presented of an investigation into the suitability of the Milko-tester for fat testing of 10-day composite milk samples.