34 MAINE AGRICULTUT^AI. EXPERIMENT STATION. I915. 



Inspection of these tables shows : 



1. That in each year the number of daughter rows is very 

 equally divided between the plus and minus columns. In no 

 case is the diiterence g'. eater than might be expected in random 

 sampling. 



2. The total deviations of the daughter rows are in most 

 cases also very equally divided between the plus and minus 

 directions. In four out of the six cases the excess of the total 

 deviation is in the direction of the selection. In only one in- 

 stance is this dev'.ation large enough to be significant. This is 

 in Table 13, the rows grown in 19 14. For the plus selection 

 in this year the theoretical total deviation is 119.43 centimeters. 

 The difference between the observed and expected ratios is 

 20.15 centimeters. As noted before it is difificult to determine 

 the standard error of these total deviations. However, it is 

 probable that this difference is not beyond the range of the 

 fluctuation of random sampling. In all of the other years the 

 difference is certainly insignificant. For this and other reasons 

 noted below, it seems probable that this one large deviation is 

 not significant in connection with the selection. 



3. Inspection also shows that the average deviation per 

 plant row is very ecjually distributed between the plus and minus 

 columns. The largest diff'erence between the observed and ex- 

 pected average deviation is only a little over 0.4 Cm. However, 

 in 9 out of 12. or in 75 per cent, of the possible cases the 

 excess is in the direction of the selection. This diff'erence is 

 not greater than might arise in random sampling. 



4. There is no evidence that the selection for height of plant 

 has produced a greater eff'ect the first year after the selection 

 than in later years. 



Since the above tables do not give any positive evidence of 

 selection, it has not been thought necessary to calculate all of 

 the indices of selection for this character. The indices for two 

 years have been calculated and they show essentially the. same 

 thing as the indices for weight of grain. The indices for height 

 tend to vary a little more widely from the average value of i 

 than those for grain weight. This is due to the fact that some 

 of the selections for height were much nearer the mean than 

 in the former case. It is not thought necessary to table these 

 indices. 



