166 Plant Tissue Culture 



number which can be practicably handled in a 

 laboratory with restricted space and facilities and 

 limited time. This laboratory has chosen 20 as 

 the standard number of cultures used in setting 

 up averages, and each experiment is generally re- 

 peated several times. The second point can be 

 settled by a statistical analysis of actual results 

 using this standard number. Some examples are 

 given in Table 3. 



In this group of experiments, in each of which 

 20 cultures grown under one set of conditions are 

 compared with 20 cultures grown either under a 

 different set of conditions or else under the same 

 conditions but on different dates, differences up 

 to 12 per cent between the two series (Nos. 1, 2, 

 3, 4, and 10) are without statistical significance; 

 differences of 24 and 25 per cent (Nos. 5 and 8) 

 are of doubtful significance (odds of less than 

 20 : 1 against the results being due to chance) ; 

 while differences of 31 per cent and over (Nos. 6, 

 7, and 9) are highly significant (odds of more than 

 20:1 against the results being due to chance). 

 This analysis indicates that in experiments car- 

 ried out in this way and using 20 replications in 

 each experimental complex differences of 30 per 

 cent and over can, in general, be relied on to indi- 

 cate a true difference in behavior and not merely 

 fortuitous differences due to the variability of the 



