122 THE ENUMERATION OF BACTERIA IN MILK 



not only examined in four laboratories but each laboratory 

 was unknowingly checking the accuracy of its own work. 

 The various points investigated were as follows: 



1. Method of Inoculation. Three methods were employed : 

 (a) Measurement of the sample into plates and pouring the 

 agar from flasks, (6) measurement into plates but pouring the 

 agar from tubes, and (c) inoculation of the tubes and pouring 

 into plates after rolling. The results obtained show the slight 

 superiority of the tube inoculation method but the advantage 

 is so slight as to be of no real importance. In the few cases 

 where methods (a) and (6) were compared, (a) gave higher 

 results though there is no manifest reason why this should 

 occur. In laboratories where large numbers of samples are 

 examined the slight superiority of the tube inoculation method 

 is more than offset by the economy in material and labour 

 effected by the use of the flask method. The author's experi- 

 ence has been that, although the time required for plating sam- 

 ples was not very much reduced, the preparation of media was 

 greatly facilitated and the cost reduced. 



Composition of Media. In one series three different media 

 were used (a) standard agar (beef bouillon with the addition of 

 1 per cent agar and peptone and adjusted to +1.5 per cent 

 acidity), (6) standard agar with the substitution of Liebig's 

 extract for beef infusion, and (c) agar prepared with beef 

 extract but containing only one-twelfth the quantity in (6) and 

 having an acidity of +0.3 per cent. 



The results showed that 



In 30 samples (a) medium gave the highest count. 

 In 27 samples (c) medium gave the highest count. . 

 In 20 samples (6) medium gave the highest count. 



So far as the actual numbers were concerned the differences 

 were of no real significance so that, in this respect, the media 

 were of equal value. The size of the colonies on (c) medium 

 was generally small and rendered accurate counting more dif- 

 ficult. Against this disadvantage must be placed the decreased 



