■iOâ IV CdM'I-nK.NCK INTKIINMKINAI i: DK l.K M-TKJl K. 



low ;is I iii 7)2 or, evcn perhaps, lowoi' slill. Il secms lianlly pùs>il>lc liiiil a 

 • Icliiiciicy of Ihis amounl can bc diio tu incrc incqualily of (listrihiilicn iii liic 

 (lilIVri-iil pdils. ,111(1 \v(» are led lo cDiirliiilc lli.il iii tlicsc cases we are ilcaliiiii' 

 w illi a secdiid pair ol' l'aclors (iiiiliraled l)\ \' ^ in adililion lo Ihe pair X. ^ pro- 

 tmnl in ail siiifrle.s. The members of lliis secund pair appear lo lune a inm- 

 plemeniary disiribiilion aniong certain of Ihe IrLie-breedinjS^ and of Ihe ever 

 s])oi'linfj; strains, one niember of tlie pair occurring in some of Ihe ])ure single 

 strains. and the olher member being présent in some of the double-lhrowing 

 siraiiis. It follows that where siich strains are sflf-hred we can liavc no 

 nieans of knowing whelher they contain one of thèse addilional faclors. or noi ; 

 only by niaUinglhe appropriale croN.scan we obtain a zygote in whii h the pré- 

 sence of holh addilional factors can be inferred from the abnorinally high 

 proportion of singles occurring in some of the familles in the F, génération. 



The practical oiileome of thèse considérations niay be brielly summcd up 

 Ihus. — In order to maintain a constant high percentage of douldes in succes- 

 sive générations it is ail importaid that the original stock should be composed 

 entirely of eversporling individuals. If this condition is fulfilletl. tlicn. w hellier 

 self-breeding or intertireeding occurs among thèse individuals, wlielhcr Ihe 

 sowingbe made from the seeds of one individual or from a sample of mixed 

 seed from many individuals, wemayexpect Ihe percentage of doubles lo remain 

 high. But if the original population was mixed, if it included some Irue-breed- 

 ing singles as well as eversporling planis, the avcragc proportion oi doubles 

 will be likely to diminish, for crossing will lead to Ihe production of a num- 

 ber of individuals yielding only I double in 4,oreven Icwcr. while an increasing 

 numljcr of individuals will yield noue al ail. 



As already stated above the proportion (d' <l()id)lcs ([iiolcd l'or strains of liie 

 'l'en week class, c/: 5r>-57 0/0 is rni averaye fur tlic plant as a ii'/ioli'. lliese 

 numbers agrée very (dosely with the results obtained by Clialé for tlie plant a.s 

 (I ii'liole, whicdi are given in bis Ireatise on tiie Cultivation of Stocks'. Cliaté 

 luiwevcr concluded from bis observations that a considerably higlier proportion 

 of doubles could be obtained from the seeds in the lower régions of the pods as 

 comparcd wilh thosc in the upper, andalso in the pods on Ihe main axis and the 

 chief latcrals as compared wilh the latéral branches of a higher order, — tlic 

 proportion in Ihe one case rising somelimes as high as 80 0/0, and failing in 

 tlu^ otlier as low as 20 0,0. Of siicli a ditïerence in distribution of the two 

 kinds of ovules — for now that we know tiial the pollen is uniforin in eversporl- 

 ing plants, and ail carries doublcness. su(di a resuit miist dépend iipon the 

 distribnlion of tlie nv/ih-s alnni'. - I bave noI iiowcvcr beeii able lo (dilain an\ 

 contirmalion-. 



Exiieriments dcsigned lo test this point sliowed somelimes a greater pré- 

 pondérance of doubles from the seeds of one région, somelimes from those of 

 the olher. tlic Kvrrajje for the two lots being almost idenlical. On (lie other 

 liand a certain amounl of évidence bas now becn obtained in support of the sta- 

 lemcnl whicli is ofleu to be met wilh, riz that a liighei- proportion of douldes 

 can bc obtained Irom seed whi(di bas becn kepi for a consideialdc lime aller il 

 was galbered, thaii from seed wliicli was more recently harxesled. It was no- 

 Liced, for example, Ibat if Ihe seeds were noI sow n until tlie bulk of Ihciii had 



I. CiilliiiT pr.iliiiue lies girollécs. 



•i. \ iiiore tli'hiilotl sliilcnienl ol' Ihe f.-icls is (lUcii eUculki'o. ^^ei-'- P- '<•>'■>, noio 'i.) 



