526 



SCIENCE 



[N. S. Vol. XXXII: No. 824 



with an odd number of ovnles,' say seven, must 

 have a different number on the two margins, 

 while one with an even number of ovules quite 

 generally, though not invariably, has them 

 equally divided between the two sides. In 

 short, a locule with an even number of ovules 

 more generally represents a bilaterally sym- 

 metrical plate of carpellary tissue. 



Let us hope that there is no mistake in 

 regard to these two suggestions. They are 

 not advanced as proofs of, or arguments for, 

 a selective elimination, but merely as the rea- 

 sons which led me to look for elimination on 

 the basis of this character. 



The actual results secured are now to be 

 considered. 



With respect to the character of their loc- 

 ules, the fruits may be classified: 



3 " even " 



2 " even " + 1 " odd " 



1 " even " + 2 " odd " 



3 "odd" 



One test may be made in either of two 

 ways : we may either compare the percentage 

 of " odd " locules in eliminated ovaries with 

 the percentage in matured ovaries, or by 

 studying the percentage of each of the four 

 classes in the two series. 



Using the first method we find for the 1908 

 series : 

 " Odd " loeiilea in eliminated series 



:= 37.407 ± .412 per cent. 

 " Odd " loeules in matured series 



= 25.185 ± .325 per cent. 

 Decrease in " odd " locules by selective elimination 



= 12.222 ± .524 per cent. 

 And for the 1906 collections : 

 " Odd " locules in youngest series 



= 33.91 ± 1.40 percent. 

 " Odd " locules in oldest series 



= 27.23 ± 1.05 per cent. 



Decrease in "odd" locules by selective elimination 



= 6.68 ± 1.75 percent. 



There can be no reasonable doubt that these 



differences are significant, and that there is 



some biological reason why locules with " odd " 



numbers of ovules are less capable of com- 



■* I see no reason to doubt it, but of course a 



working out of the problem on Staphylea itself 



by some embryologist is highly desirable. 



plating their development than those with 

 " even " numbers. 



Naturally one locule can not fail to develop 

 without two others falling at the same time. 

 The final test of our theory is to determine 

 whether ovaries with one to three " odd " 

 locules are more likely to be eliminated than 

 those with all three locules with " even " num- 

 bers. The results are obtained at once by 

 reducing the frequencies for the 2,095 elim- 

 inated ovaries and the 2,Y07 matured fruits 

 of the 1908 series and the frequencies for the 

 174 youngest and the 273 oldest ovaries in the 

 1906 collections to percentages. Tables I. and 

 II. show the figures. 



TABLE I 



1&08 Series. Elimination of Ovaries with a 

 Preponderance of " Odd " Locules 



TABLE n 



1906 Series. Elimination of Ovaries with a 

 Preponderance of " Odd " Locules 



Two things are very prominent in these 

 tables. The first is the fact that in all four 

 series the formulae with a preponderance of 

 " even " are greatly in excess of those with a 

 preponderance of " odd " locules. For 1908 

 these are 66 per cent, of the eliminated and 

 81 per cent, of the matured ovaries. In 1906 

 they are 72 jjer cent, of the youngest and 87 

 per cent, of the most mature ovaries. 



No less marked is the fact that aU classes 

 of ovaries except those with only " even " 

 locules have become relatively less frequent 



