TECHNIQUE OF AUXIN DETERMINATIONS 39 



one or two values fall far off the distribution curve, they 

 are generally neglected. If the values are too scattered, the 

 experiment should be repeated with more uniform plants. 

 As a rule, the mean of 12 good plants is taken. 



As measure of the variation among the test plants the 

 standard deviation of the mean ^ is best used, and this 



I 2 3 4 5 6 7 8 9 10 II 12 13 14 15 16 17 degrees 



Fig. 18. Distribution of individual curvatures in the Avena test. Abscissa, 

 curvature groups; ordinate, number of plants in each group. Diagram A, 

 perfect test, agreeing with the variability curve A'; diagram B, imperfect test, 

 skew with relation to ideal variability curve B'. (After Went, 1928.) 



obviates the necessity of detailing the individual curva- 

 tures. 



The possible causes for this variation between individual 

 plants are many. Variations in thickness play no part 

 (Went, 1928). Differences in actual age of the plants are of 

 course excluded because all have been planted at the same 

 time. Variations in the length of the cut tip do occur, but 

 are of minor importance if they do not exceed a few milli- 

 meters. The following table shows the relative auxin curva- 

 tures of plants from which different lengths of tip have 

 been cut off (Table IV). 



/ '*' A" 



1 Standard deviation of the mean = A/ — , where A is the difference be- 



y n{n — 1) 

 tween the individual observation and the mean, 2 A^ the sum of the squares of 

 these, and n is the number of observations. 



