H. G. Thornton 273 



Rothamsted soil plated on them, eight parallel platings being made of 

 each medium. Table VIII shows the colonies developing on each plate. 

 It will be seen that no significant differences are shown between the 

 different batches. 



Table VIII. 



Colony Develofmeni from a Single Soil Suspension, Plated 

 on five different Batches of Count Medium. 



The uniformity between parallel platings on the same batch of 

 medium has been studied from about 3000 platings made in the Proto- 

 zoology and Bacteriology Departments at Rothamsted. A statistical 

 analysis of this mass of data has been made by Mr R. A. Fisher (Fisher, 

 Thornton and Mackenzie (2)). The results, which are on the whole quite 

 satisfactory, are published separately. 



About 4000 platings have been made on the medium since its develop- 

 ment. In the majority Rothamsted soil was used, but 240 of the platings 

 were of a light ironstone soil from Kingsthorpe, Northamptonshire. 

 Although "spreading" organisms occur on about 40 per cent, of the 

 platings of Rothamsted soil, only some 3 per cent, of the platings were 

 lost owing to the development of spreading colonies over the surface. 



G. Summary. 



1. For bacterial count work the first essential in a medium is that it 

 should be uniform and reproducible in its results. 



2. In the medium here described, details of which are given on 

 p. 271, reproducibility has been achieved by the use of pure chemical 

 compounds in an agar medium and by selection of such constituents as 

 will not produce a significant change of reaction during sterilisation. 



3. On agar media, surface spreading colonies interfere with the 

 accuracy of the results. A special study was made of a common spreading 



Ann. Biol, ix 18 



