•i.-o IV' r,i)\i-Kiîi:\r,K intkkwtihwi.i: iii: (ii:\i:Tini k. 



liiiiN <il' llir r\]ii'riiiicnlcr. Ili(iui;ll si \ ;ilii;ililr liiciN. lu l|ii~ jiMpiT I wisli li> 



puiul oui, esi)(!ci;ill_\ . Imw closcly tlic (i:iu>si:iii ciirN imiI' ci rur iii;i\ l)i' siiiiuhilcil 

 by a genuine segregiUion wliidi doi's uol ajipi-ar lo he comparable in coniplcxily 

 to many problems ol' biiiiian inhcrilniico. The. discpiaMIicalions wliich I bave 

 already menlioneil prevenl ilic cases ciled IVom being conchisive: Ibeir inleresl 

 is, hovvever, in no way (bminished, liowever much Iboir vainc may l>e depre- 

 cialed, and an cxaminalion ni' llieiu niay lead lo a rurllier slrenglhening of 

 scepticism as lo tbe « plasiirily » ni' Ihc organism. 'l'bere is a pecnliar appro- 

 prialcness in Ibcir cil.ilion IVoiii d:ihi dlilained ^villl (iussiiiiium . Un- iio phinl ol' 

 agriciilinrai imporlance liascM-r Ik-ch accused al' sm-b over-w hclniiiig « \aiia- 

 bility )), and gênerai unrelinbilily ; wbereas, in poini ol' r.nl. ils lluilii:ili(in 

 is in no way abnormal, and tlie original « variabilKy " iin[)nli'd tu il wns an 

 insnfficiently analysed complex, consisfing ol' flnctualion, segregalion. varielal 

 impnrily, reversion Ihrougb llie meeting of cry[)lonicres. gamelic ciinpling. 

 and frec inler-crossing. Il is not iinlikcly Ibat mulalion ilself migbl be addcd 

 lo this list, bul in view of onr ignorance of tiic otber iicms, \ve sball nol be 

 able even to discnss llie possibiJily of mulalion in llic rnlldn jiL-inl l'nr many 

 years lo come. 



A bricf noie im liic niilily of measurenicnl- mclhoiis sIkiuM lie iiere addcri. 

 Snrprise bas been expressed by persons ac([uainled with llie oullines of Ibis 

 researcb, Ihal one shonld lake Ihe trouble of measuring leaves, l'or example, in 

 order to obtain resulls whicb are no bélier Iban Uiose gol by eye-judgments. 

 Tbe obvions reply is, of course, thaï Ibe callipersare free IVom subjective error, 

 while tbe eye is nol. Furlher, subjeclivily being eliminaled, resulls oblained 

 in varions localities become comparable, vvhich is a point vvorlhy of économie 

 considération in sucb a widely cultivated plant as (iuifxypium. Beyond ail 

 this, ho\vc\('r, tbere is one suprême argument is favourof atlempls al accurate 

 determinalion of form-cbaraclers: nnmciy. Ihaf wben many factorsare involved 

 in sucb a resuit as the shape of a leaf. wc llien llnd Ibat tbe varions gronps, 

 — or modes, — formed between the extrêmes are so closely similar eacb to ils 

 neighbours, Ibat Ihc eye is nol capable of difl'erenlialion. Tbe rcsult of eye- 

 estimales in sucb cases, — vvbicb espccially comprise species-bybrids, — leads 

 Ihe sludent lo an opinion thaï variation is continuons between tbe two extrê- 

 mes, and furlher analysis of tbe component pbenomena is al once arresled. 



This view assumes thaï tbe probable error of the linal expi'ession or measu- 

 remenl ean be négligeable. In pi-actical reseai'ch. a compromise bas In lie 

 driven between tbe ideally small probable error, and Ibe laljonr and lime w liirb 

 ean be expended in niaking each 'JelerTninalion. In this work wc bave bad lo 

 be content witb a fairly large P. E., in order to obtain d;il;i Irom suflicient 

 individuals of eacb family. 



■/'/)(■ cxpi'riinenlal i-rror due lo vici»/'.s//ï. — The history of Ibis serions bin- 

 drance to the collon breeder is peculiai-. and much yet reniains lo be invesli- 

 galed. 



Till l'.H).") Ihe exislcm-c^ of nal iiral crossing bclwcen collon-jdanls was alinosl 

 ilisregarded. In tbal year the wriler showeil Ihal nol less than ."i percent of 

 crossing look place in Ihe Egyplian lield crop. This conclusion was indepen- 

 denlly reacbed by otber workers, notably Lcake in India. and. al llie prest-nt 



