24 ' Agricultural Journal of Victoria. 



THE VARIATION IN MILK TESTS. 



By B. T. Archfr. 



The manager of a butter factory often finds it a most difficult task 

 to convince some of his suppliers that the Babcock test is a correct 

 method of estimating the butter-fat in milk. When a test shows 

 any variation from the previous one the supplier often jumps to the 

 conclusion that it is wi^ong, losing sight of the fact that the Bab- 

 cock is a mere machine, while the cow is not, and that the machine, 

 if properly handled, is so accurate that chemists often rely on it 

 when analysing milk. 



We frequently hear of suppliers sending two bottles of milk to be 

 tested, representing them to be from two different cows. If the results 

 do not come out the same, the process is immediately condemned, 

 though nothing is more probable than that the contents of the two 

 bottles are quite different owing to faulty methods of taking the 

 samples. Many imagine, for instance, that a correct sample of the 

 milk of any cow is obtainable by merely pouring a portion from the 

 bucket immediately the milking is finished. This idea is entirely 

 erroneous, as the fat is rising all the time^, while the strippings, which 

 are the richest part of the milk, never reach the bottom of the bucket. 

 The milk must be poured at least three times from one bucket to 

 another, and the sample then taken if accuracy is desired. Again, 

 those in the habit of testing their cows sometimes test the bulk 

 as a check on the factory. This is a good plan, but it is of thk utmost 

 importance that the sample tested shall be a correct one. At the 

 factory the milk is all poured into a weighing vat and thoroughly 

 mixed and the sample then taken gives the true basis for the 

 accurate estimation of the butter-fat in the whole. It should always 

 be borne in mind that a correct sample cannot be obtained by merely 

 stirring the milk in the can, and that a can of night's milk will con- 

 tain a higher percentage of fat than a can of morning's milk, and since 

 there will be a lesser quantity of the former, the testing of a sample 

 containing equal proportions of the two will give a higher result than 

 obtained by the factory, though the factory will be correct. Suppose 

 there are three cans of morning's milk and two of night's, then three 

 parts of the former must be taken to every two of the latter, or the 

 whole properly mixed before sampling. 



I have known a supplier whose milk had been regular in quality 

 to suddenly obtain a very low test and on enquiry it was found that 

 the bottle containing the sample was knocked over while the plug was 

 out, so that some of the cream was lost. Of course, such a test would 

 not be recorded. Oases have occurred in which the samples have been 

 tampered with, cream taken out or water put in. This would cer- 

 tainly be shown in the test, but if the supplier's test had been consistent 



