402 MILK AND THE PUBLIC HEALTH chap. 



When the sample analysed falls below the legal limits the 

 milk producer is notified of the fact, and, with his permission, 

 a mixed milk sample is collected from his herd at the same 

 milking-time as when the deficient sample was obtained. If 

 the mixed milk of the cows sampled at the farm falls below 

 the legal standard, it is presumptive evidence that the 

 deficiency is due to the cows and not to outside human 

 sophistication. 



There are a number of possible fallacies and sources of 

 error which make this test less valuable than would at first 

 sight appear. It is, for example, very important that the 

 interval between the collection of the first sample and the 

 taking of the second from the cows should be as short as 

 possible. Otherwise the herd may have altered in composition 

 by removal of some animals (running dry, by sale, etc.) or by 

 the addition of fresh animals. This will be used as an argu- 

 ment that the samples are not comparable. Also, if the 

 interval between the samplings is considerable, it will be 

 urged that the feeding has been altered, the weather has 

 changed, etc., and so, consequently, the samples are not com- 

 parable. 



Further, there is room for deliberate fraudulent manipula- 

 tion. One or more of the cows giving the richest milk may 

 be temporarily removed and the fact suppressed, the interval 

 between the previous milking and the milking for the second 

 sample may be intentionally much prolonged ; while, in ^a,v- 

 ticular, at the second sampling, many of the cows may be 

 intentionally not milked dry, so that the strippings (very 

 rich in fat) are not added. From all these causes a specially 

 impoverished milk may be artificially produced and the cows 

 be saddled with the blame. The writer does not believe that 

 such practices are anything but rarities, but their possibility 

 must be kept in view. It is not usually to the farmer's 

 interest to proclaim his milk of poor quality. 



Another source of error is that the sampling must be 

 properly done, and that however carefully done the samples 

 are really not comparable. A farmer does not, in general 

 mix the milk of all his cows before sale. He fills one 

 churn with the milk of four or five cows, then the next churn, 

 and so on. The deficient sample will probably not represent 



