522 



SCIENCE 



[N. S. Vol. XXXII. No. 



11 12 13 14 15 16 



DiAQEAM 1. Showing 



19 20 21 22 23 24 25 26 28 29 30 31 32 33 



38 39 40 



:e in mean number of ovules due to selective elimination for each of 

 twenty-eight individuals. 



the matured ovaries. There are probably two 

 reasons for this. The first is that there is a 

 real difference between individuals in this re- 

 spect. The second is that a part of the varia- 

 tion is due to the probable errors of the con- 

 stants. 



Both constants, based as they are upon 

 random samples — -and in the case of the in- 

 dividuals, not very large samples — are subject 

 to the limitations of a probable error of ran- 

 dom sampling.' 



It is interesting to inquire how many of 

 the differences for individual trees may be re- 

 garded as trustworthy. Following the con- 

 ventional standard, we find one significantly 

 negative difference, two which are positive 

 and probably significant, and twenty-five which 



^ This is the plus or minus quantity attached 

 to the means given above. Conventionally a dif- 

 ference between constants is considered trust- 

 worthy if it is two and a half times its probable 

 error. The difference of the two general-series 

 means given above is over forty times its probable 

 error. 



are positive and unquestionably significant. 



From the 1906 series in which the ovaries 

 were taken in different stages of development,, 

 we find confirmatory evidence. We have for 

 average ovules per locule : 



Series A= 6-10 mm. ovaries, Av. = 7.232 ± .02D 

 Series C = 16-20 mm. ovaries, Av. = 7.821 ±: .016. 

 Difference = .589 ±333 



Here the difference is nearly eighteen times 

 its probable error, and represents an increase 

 in mean number of locules of over eight per 

 cent, of the mean for the youngest fruits. 



From the foregoing facts only one conclu- 

 sion can be drawn. In Staphylea the failure- 

 of ovaries to develop to maturity is not ran- 

 dom but selective. Ovaries with lower num- 

 ber of ovules have smaller chances of becom- 

 ing fruits than those with higher numbers.. 

 The intensity of selection is such that there is 

 a difference of about seven per cent, in the 

 mean number of eliminated and matured' 

 ovaries. 



Changes in Variability due to Selective 

 Elimination. — The comparisons for variabil- 



