142. Van Oijen Modification of the Frost Little Plate Method. A Critical Investigation of van Oijen's Test for the Bacterial Content of Milk Samples

Abstract

1. Van Oijen's modification of the Frost Little Plate method for estimating the bacterial content of milk has been examined and compared with the plate count. 2. The colony counting error of the Van Oijen test is about the same as that of the plate count. 3. From the results of quintuplicate tests it is recommended that 24 hours' incubation at 28°C. be used for the Van Oijen test and 72 hours at 37° C. for the plate test. 4. Thirty-one low-count samples (3000–30,000 per ml.) and thirty-one high-count samples (30,000–300,000 per ml.) were tested in quintuplicate by both methods, and statistical examination of the results shows that at both levels of count the Van Oijen test is significantly more accurate. The Van Oijen results are on the average 1–3 per cent log lower than the plate count and this difference is found to be significant. 5. In the Van Oijen test the results are based on a larger amount of milk than the plate test, and a method is given for testing whether the increased accuracy exceeds the expectation owing to increased size of sample. 6. It is noted that failure of replicate tests to reflect random distribution may be due to the variability of the technique as much as to irregularities in the actual distribution of the bacteria in the sample. 7. The effect of size of sample upon reproducibility is discussed. A true comparison of the accuracy of various methods for assessing the bacteria content of milk samples cannot be obtained unless allowance is made for variation in size of sample as between different techniques.