Individual Strength and Loujth of Fruit. 533 



iired the mean length of the lower five and of the luwer 

 twenty fruits on them. 



The mean length of the fruits is found by dividing 

 the sum of their lengths by the number of fruits meas- 

 ured. For this purpose the fruits were cut tlirough just 

 at their base (which is marked at its point of junction 

 with the bract by a constriction, so that the measure- 

 ments could always be taken from a fixed point), laid 

 one after another, end on end, in a row, great care being 

 taken in arranging them; and the length of the wliole 

 series was read off. In this way a greater exactitude of 

 the measurements is attained, whilst only one measure- 

 ment is necessary for each plant. 



Let us choose an example. On one plant the total of 

 the lengths of the five lower fruits was 167 mm., that 

 of the twenty lower fruits 688. The mean numbers were 

 therefore 33.4 and 34.4. The difference is 1.0 mm. 



In this way the differences for the 38 plants were 

 determined; some were positive, others negative. Neg- 

 lecting the sign the differences were now written in a 

 series in order of magnitude. The result was that in 

 half the individuals the difference was less than 1.25, 

 but in the other half greater. In one case only (hd it 

 reach as much as 4 mm. The probable error is therefore 

 1.25. 



In other words: In the highly improbable case of all 

 the differences being positive, or all negative, the figures 

 in our table would have been 1.25 mm. more accurate if 

 I had alwavs measured 20 instead of 5 fruits. Differ- 

 ences of 1.25 and less must therefore be regarded as 

 within the limits of errors of observation. The difi'cr- 

 ences occurring in the experiments to be described are, as 



