I5() IV CdM'KHKNCI'; INTKIINATIONALK HE liliNETIOlE. 



The cui'vi! is noi smoolli, Ijiil ?lio\vs modes whicli apjx'iir lo hc significanl; 

 Iheir signilicant-e is i'uiilier indicaled by comparison willi llie lolals i'roni 

 several F- familles growii in Ihe siibsequenl year (Fig. 7) wlien the same 

 modes tend lo reappeai', paiiicularly Ihat al U.08.") g.; Ihe modes in Iheheavier 

 pari ol' Iho ciirve are noI so ohviously coincidenl. since ahnosl ail llie F- 



150 gr 



Mnnn Si'ed-\Voi''lil. 



planls were grow ii rnuii iiglil-sceded parents. Thaï Uiese lieavy modes are 

 also genuine is show n liy dissecling for habit of branching; llie " unbranched " 

 planls fill up Ihe 0.120 g. mode, while Ihe " defmitely branched "' fill Ihe 

 O.lâog. mode (Fig. 7. d.) So far Ibere is évidence of correlalion only, Ind nol 

 of segregalion. If we nexl reslrict Ihe size of our groups. and lake eharaclers 

 whicb were shown by Ibe correlalion diagram lo hâve some influence on seed 

 weighl, we oblain invariably such graphs as Ihose of Fig. 7, a, b, and c: the 

 seed-weighl is .scallered over almosl Ihe whole range of llie graph, andalways 

 heaped up into hvo gronps, onc heavy and one lighl, the lalter being much 

 tbe smaller. The only diflicully lefl is lo décide how Ihe new " heavy seed 

 (•an havc made ils appearance, and this is overcome by carrying the principle 

 of autogenous corrélation back lo Ihe parenl slocUs. Considering the ell'ecl of 

 one characler only, for simplicily's sakc, Ihe matler slands thus : — Ihe l'^gyp- 

 lian parenl had a boll-diameler of 20 mm., while Ihal of Ihe U|)land was 

 Ti'i mm., and Ihe Fi boll was ralher larger than eilber: we ha(i aiready 

 foiiiid liiat llie niean seed weighl was nolal)ly increased in Ihe Ijig-boll Y, 

 planls, and hence il seems clear llial the Fgyptian parental seed-weighl was 

 gametically large. \\'lien plaeed, as Ihe dominant rliaracler (Fig. 5) by the 

 acl of Crossing, inside a boll wbose cubic capacity was nearly twice tliat of ils 

 ovvn boll, il " enlarged " lo a much greater size than Ihe Upland seed had been 

 capable of assuming in a boll of llie same diameter. Then. in F,, Ihe ligbl- 

 seed récessive .'^egregaled from il, lo lind ilself back again in Ihe '-d mm. boll. 



