10 AGRICULTURAL RESEARCH AND THE FARMER : 



groups be studied, but correlations must be made with such 

 factors as temperature, soil moisture, rainfall, organic matter, etc. 

 The continuation of the work may be summarised briefly. Each 

 of the nineteen kinds of protozoa is being bred separately, mainly 

 to discover reasons for the daily variations in number. Simi- 

 larly, mixed colonies of protozoa and bacteria are being studied 

 over prolonged periods. Soil samples from all parts of this and 

 other countries are being examined to see in what way their 

 living population differs from that found in the Rothamsted 

 soil. The whole bacterial and protozoal population of the soil 

 is being studied and classified. 



Some idea of the work involved in the daily counting of these 

 minute organisms may be gained by a reference to the laboratory 

 methods employed in connection with the counting of the bac- 

 teria. Samples of soil, each lo grammes in weight, were taken 

 daily. The soil was diluted with 250,000 times its volume of 

 salt solution, and a known quantity of this fluid placed in dishes 

 containing a substance which will provide suitable food for the 

 bacteria. On the seventh day, and again on the tenth, these 

 dishes were examined and the number of colonies of bacteria 

 counted. From these figures, by a simple calculation, the total 

 number of bacteria per gram of soil was estimated. Five dishes 

 were used in this manner, and the bacterial colonies counted in 

 each, the average of the five being taken as the number present. 

 The counting of bacteria again on the tenth day was to ensure 

 that slow-growing groups of bacteria v/ould not escape being 

 counted. This method was adopted for each of the 365 days in 

 the year. 



Here it may be useful to interpose an example of the great 

 assistance which may be rendered by an expert in statistics in 

 connection with research of this character. It will be understood 

 that in dealing with such a subject as bacterial counts, every- 

 thing depends on the accuracy of the method of counting 

 adopted in the laboratory. The final working up of the results 

 is largely a matter for the statistical expert, but his usefulness 

 does not end here ; he can also check the accuracy of the labora- 

 tory methods employed. The Rothamsted statistician has devised 

 a formula for dealing with the mass of figures obtained in the 

 bacteria investigation, and amongst other things this formula 

 shows that in more than ninety per cent, of the counts made, 

 the laboratory method was technically perfect, and that it is 

 possible to pick out the few days in which, through some disturb- 

 ance or other, the counts were made in an irregular manner and 

 must be discarded when drawing deductions from the results. 



