ACCURACY OF COUNTS 123 



trouble experienced with spreaders. Observations for spreaders 

 indicated that 128 were found with (a) medium, 21 with (b) 

 medium and 23 with (c) medium. On the whole, it would 

 appear that the (b) mediimi was the most satisfactory. 



Uniformity of Technique. The several series of compara- 

 tive examinations produced some interesting data on the influ- 

 ence of technique. In the first series when each laboratory 

 used the technique as previously developed in that laboratory, 

 the results on duplicate samples showed a variation factor of 

 from 1.3 to 43.2 with an average of 6.2. The variation factor 

 was obtained by dividing the highest result by the lowest. 

 Duphcate analyses in each laboratory also showed variations, 

 the average factors varying from 2.1 to 4.8 with a general 

 average of 3.7. 



In a second series of tests the various laboratories all em- 

 ployed identical technique as to shaking of sample, diluting, 

 pipetting, inoculating, and counting of plates. As it was found 

 in the first series that one laboratory employed a magnifying 

 lens for counting plates and another the naked eye, it was 

 decided to use a standard lens in all laboratories and to deter- 

 mine the personal error in counting by an excha^pe of incubated 

 plates. The results showed that the personal error may be a 

 serious one, for, although the variation in duplicate counts of 

 identical plates was usually small, the extreme variation was 

 nearly 100 per cent. In this series the average variation in 

 each laboratory was from 1.6 to 2.2 with a general average 

 of 1.8. 



A five-day count was also compared with the two-day coimt 

 and, although the results were usually higher they were not 

 uniformly so. There seems to be no apparent advantage attain- 

 able by prolonging the incubation period beyond the usual 

 forty-eight hour period. 



In the third series the effect of agitation, amongst other 

 points, was determined, and although the results are not con- 

 clusive they indicate the importance of standardising this por- 

 tion of the technique. In the third and fourth series the plate 



