^^o. 479] 



VARIATION IN LOTUS 



767 



From this table we see that, in round numbers, 1400 capsules 

 produce 35,000 seeds. Further, it is clear that the different classes 

 of capsules do not contribute in proportion to their frequency of 

 occurrence to the total seed number. Thus, for example, a ref- 

 erence to Table 1 shows that capsules with 21 seeds each and 

 capsules with 28 seeds each occur with ecjual frequencv in our 

 sample. But obviously the latter will contribute more to the total 

 numl)er of seeds. As a matter of fact the 2S-seed capsules contri- 

 bute SI per tliousand of the total number of seeds, as a^jainst (>() 

 j)er thousand of the 21 -seed capsules. Taking the data as a whole 

 I find l)y a very simple calculation that: 



(a ] ra])su]cs with Jrurr than the median niiml)er of seeds bear 

 altoo-cthcr 1 .")()( U).;^2r) .seeds, or 42.!)() percent of the total numi)cr. 



(h) ('a])sules with more than the median number of .seeds bear 

 alto<j:ethcr 2000().()7:) seeds, or oT.Ol percent of the total innnber. 

 In other words 50 percent of the capsules produc-e 57 percent of 

 the seeds, or, put in still another way, one half of the heads bears 

 14 percent more of the total number of seeds than does ilie (»tlier 

 half. This result is, of eoui'M.. an .,l>^ion^l^ neee.^ar;. aiilln.i.li- 

 cal con.setpience of the svinmetry of ih." eaosnle di>iril.ution, yet 

 it is a point ^^h^u■h is fre". |.iemb " ox eriooked'. .V .Nunneineai (lis- 

 tribution of the individuals of a population with respect to some 

 meaMU-e ..f feeuu<litv ,loes not mean that the eontril.ution. of these 

 individuals to the n<"xt ovueration cNtMi l>ef..re seleeiion uill be 

 rejiivsented by a synnnetri( al .listrihution. The veiy fact that the 



relation. 



The results with reference to the proportionate contributions 

 of the ditferent classes of heads to the total seed number show the 

 conditions before elimination bepns. Many of the 35,000 seeds 



