TffE THEO_RY AND APPLICATION OF TKE ~BACKCP~OSS 52tO(,KNIQ~ E IN COTTON BREEDING

Br R, L, K• !!i.Sc.(Acmzc.), P.E.D., A.I.O.T.A.

CoN:rNNTS

76 Theory of backcvossing 77 Applicalffou of baokorosaing in co~to~ breeding 78 1, Police parent 78 2. _b'em~le parent , 78 .3. Progeny size 80 ,i, Basis of seleoMon in hybrids 81 5. The eud-puint . 81. C. AddiMve factors.. 8.3 - 7. Linked factors . 8. 7gleudh~g inherit~mee . 8t 9. JJuR~ propagatlon 8'1- ii). Pm'il.y of b~uekcross parent 85 Seminary 85 lleferences . 86

LWTI~ 013 UOTI ON The essential v.'-~lueof is that it, provides a means of limiting the hetero- geneity which we uld resuIt from 'strMght' crosses between two ~ypes, making it possible to produce a simila~ to whichev'er of the two varieties has the more valuable genetic constitution, yet containing desirable characters transferred from the other parent. Backerossing obviates the necessity for rigid selection generation after generation, in _~'=, Fa, F~, etc., by progJ:essivdy and automatieatl.y, rendering the hybrid more and more homogeneous. In many ways it is, to the plant , the equivalent of line breeding to the stock breeder, with the added advantage that many plants are nob harmed even by the closest izlbreeding. t{arland (1934) gives a list of some of the gone transferenees which have been effeeted" in i=~erspeeis crosses in cotton. I{e was the re'st {o realize the value of baekerossing as a tool for the cot,ton breeder, and his work remains a tallest, one in the development of the genetical approach to cotton breeding. The otass.[cal example (in cotton) of baekcrossing with a commercial aim is probably the transference, by Harland and Evelyn, of fed plant body-weak spot-from Trinidad Red Ki&my to the Sea Island strMn, V135, and, on the genetical side, the transference, by ~arlaud (1935), of the gene R.,as from the diploid coSton Go~s.ypiw~7, e~*5ore~.~..mb. to the allotetraploid American Uphmd ( g. h.irs~t.u.~ L.) Ud. The fact that these, and all other, a~tempts a.t gone transference failed to produce com~ mercially suceessfhl t~)es, in.diea~es fauRs in the technique employed, since it is u~likely {hat, in every case, the genes which R was desb:ed to transfer had unbreakable deleterious linkages. This la.ek of success had made cotton breeders sceptical as to the valu.e of bach crossing althoagh the technique hes been successf~d in other crops, and, in consequence, its possibilities have not been adequately ex]?Ioited. R. L. K~GxT 77 The foH0wi~g notes record the writer's experience of inter- and intrasped~ic back- crossing over a number of years, dm:ing which period four distinct genes fi'om tlu-ee different spades of cotton :'~ have been suocessfulJy ~ra,nsferred to two commercial strains of Q. barbade~se L., whilst several intraspecifie transferences have been made in ft. hirsu~um L.

THEOPoY OF ]SACKOI~OSSIlVC- The usual obiect in baekerossing is gene transference, but the technique has also been employed to add genetic variability to m~dsting cotton t3~pes in the 5ope that it might be possible, later, to isoh~te entirely new va.rietie# of commercial value. Notable examples of this latter use are the assortmelR of crosses sent out to the African cotton experiment stations in t932 by Hal'land and Evelyn ([Evelyn & Harland, 1934). These hybrids were very heterogeneous and were dm'ived from early backerosses of accepted American Upland varieties and Jamaica Xerophytic, Gambia-Na~ive, or Galapagos Native. the ba.ckcross parent~ being the American Upland type. The object ~ in this case., was to 'increase the genetic variability of Uzt so that selection for any particular environment wo~fld be facilitated' (Evelyn & Harland, i934), but, at the same time: ~o avoid the. complete heterogeneity which would have resuRed from straight crosses between these widely divergent types. Some of this material did not involve interspecffic hybridization, and Ducker & Miller (194t2) appear to be breeding successful varieties from (U~ x Cambodia) x U4 hybrids sent, out ])y ~-Iar]and and [Evelyn at the same time as their interspecifio crosses. Hutchinson (1938) suggests that 'an impm%ant reason for the lack ofsuccess of Harland's U4 hybrids' is that 'U~ itself was improved by selection so rapiclly that the products of selection in the hybrid material usually failed, to semi?ere ~ T]is use of inter- specific baokc.rossin~ to increase variability has yet t ~ prove, its value in. cotton breeding and it is not proposed to oleo] with it in greater detail here. In gone trs.~ference, the object is to move a single, dominant, or partially dominant, gene{: (or a small number of such genes) from one cotton type to a~].other, without deleteri- ously affecting the other qualitative or quantitati~;e characters of the backcross ]?areat varieb', A genei, al example would ]:,e the trausferen,ce of a genoa ~'om a donor parentw variety 1" to a l)ackcross parent variety Z. The proeedm'e would, be to cross 1:' • Z and cross the F I back to Z, to Droduce a baokoross proge.uy of Aa and aa ])tarts. An Aa plant selected from this first backcross progeny would agaifl be crossed with Z, giving the second backcross. This process should be repeated until the Aa ]?]ants in the baekcross progeny * TransfermJee of the blaekarm resistance genes B~ and B= fi'om G. ldrs~f.~f.m L., of the blaeksrm resisga.nee gone B:, from <~'. p~wgaftcm Sob, & Then. and of tire 6/. (~rborcl~#t.L. geue 767;8 from ]is.cloud's ~U~. (i~h'sld'~n~. x (~,r~csgen~c • Iu a.ddit,Jon, other gone transferences are head3: c

Table 1.. Rate of elim.Dl.atio~ of do'~or ye~.ot,y~ge by ~e~ol,;e'ross@,cd ('n.eg&et'i~N. the e.[/L,cts q[ sdectio~, c~,~d li.rUcage) Donor Baekcros~ Donor 13&ckcross paren~.... paren~% i) a,ren~o, par~n6% /0 /0 2?~ 50.0 50.0 6bh backcrosz 0-8 99,2 ls~ backcross 25'0 75-0 7~h backcross 0"~ 99.6 2ud bacJecrosa 12,5 87-5 8t'll baekcross 0-$ ,~J9.S 3rd backm'o.ss 6'3 93'7 9~h b~ckcrosu 0,1 99-fll :l:l,]t baekcross 3.t 91]-9 10gh baekoross 0,05 99-95 5~b, bt~ckot'o~s 1.6 .98-I,

in the endTroduo~ of f:kis hybridization ~ number of Aa plants would be soiled and ~he balk self-bred seed sown. This would give a ]?rogeny consisting of 25 % AA : 50 c~j As: 25 % aa pla~ts. Assuming the gone A to sDow only partial , then the three genotypes would be phenotypieally dis{inc~. The Aa an<[ aa plants would be pulled up before s and the remaining plan% (AA) would bg bulked t.oge~her to form the new strain, tIad die gone A been fn!ly domb nan% instead of only partially dominan% gte Aa and AA plants would have been pheno- typieahy ide.ntioa! and a short progeny row Dora each plant of the A phenotype wou]d have to be grown. All progemes producing a proportion of aa plants would be eliminated and the remaining rows would be balked as the new strain. The foregoing account gives the broad basis of tee theory of baekdrossing, bn~ success or failm'e is largely a matter of attention to the finer points of ~eohnique drab with below.

APPLICATION OF ]3ACI~CR.OSSIN~ I2r COTTON BREEDING

If possible, the hybrid should be the mate parent and the strain to which it is being back- crossed, the .female. The objects of Qis are threefold: (@ Any pollen from-ehe baekcross parent accidentally left on the stigma would produce self-bred plants which would automatically be eliminaeed from the backeross proge W since they wotfld not contain the gone being transferred. Their only effect would be to bias the genetic ra~io. (5) Each flower from the hybrid can be used to pollinate about ten prepared flowers on a ptu'e-line family of the baokcross parent. (~) I~ is possible to ~ow at least two generations per year because ~he hybrid plan~.s need only produce pollen which can be used on welLestablished

(B) Fem~le io~re~t Hybridization cannot replace selection and, during the time taken to carry ou~ a baek- c.rossh.na programme, the plant selection/st will prooa.ly have improved the. original variety used as She backcross parent. It is, ~b.eJ:eforc, essentia.t to keep dEe baekcrossing up-to-date, This is done by using, each season, the latest substrain of the parent varioW for backcrossing.

The parentage of one of the blackarm resistant Sakel st,rains synthesized by the writer illustraees Vhis point. The originaI objective was the production of a blaokarm resistan~ P~. L. K?czG,rr 79 X 1530 strain, :!"the resistance being derived from the American Upland tyq~e, Uganda ]3 31, which contains ~he blaekarm resistance factors B:,. and B~ (Knight & Cloust.on, 1939). A~ gke time this crossing was s~arted, the substrain X1530A wa.s replacing X1530 in ~he CxezLra~area of the Sudan, and X 1530B, the latest substrain, seemed a likely successor ~o X1530A. 213 31 was, accordingly, crossed with XIbg0B, .bu~ as this snbstrain failed, on gesr fie show superiority~ over X1530A, the first ba.ckcross was made ~o the latter s~rain. At this stage X1530E showed definite promise as a type for repladng X1530A, and the next two backerosses were, accordingly, made ~o Xlbg0E.. Daring this time Xl%0 and. i~s subs~rains had come fie the fore in the Gezira, and X15301~, though-{he eqaal of X1730, did not go int.o commercial cultivation. (The X1730 type derives from ~he same "blood" as X 1530 and the two strains are very similar in all charae{ers (Lambert, 1988).) The new two crosses were made to ~he latest subs[rain of X1730. viz. X 1780 @, and subsequent crosses were made to X 1730H. Thus the pa.ren.tage of ~he first blackarm- resistant X 1730 strain to go into com~eroial bulk was as follows:

{[((B 81 x X 1580B) x X 1530A) • X 1580E s] • X 1730 @?} • X I730H "~.

(The superscripts denote the number of times each sara.in was used as a parent. ~or the sake of clarity, the practice of writhag the female paren{ fn's~; has noi been followed.) Commerioal bulk propagation was s~ar~ed from the seventk baekeross progeny, and t.his is being replaced by ninth baokcross material of the following parentage :

{[((B 31 x X I530 B) • X ].530 A) x X 1530E "~ x X 1780 O:} x X 1730t.[~.

0mitbing seine{ion and linkage effects the percentage composition of these two com- mercial b~dks should be as follows:

8Lrain ,.Tth baokoroas 9~h b~ekeross Strain %h baekaross 9~h baokeross BS] 0"59 0"10 X] 53U:~ 4"69 1"17 XIgSOB 04"19 0"10 -I 17:~0 @ 18-7.5 ,_t.69 XlS,30A 0-78 0.~0 XlTa0~ 75.00 9a-75 The value of {]~e two :anal baokerosses lay, not so much in the elimination of 13 ,3] gone- type (which was reclueed from 0-39 ~o 0-10 %), as in bringin~ the type np-to-d.age by iacreasiag th e pro:courage of X I%0II blood from 75"00 to 93.75. A second example of keeping the ba.ckcross parent a~pZto-date is given by the b]aekarm- resistant NT2 strains bred at Shambat.$ ]Bulk pro]?aga,~ion was started from tLe fifth Sakel (fourth NT a) backeross. Ts :fifth baokeross z.aai,erial was completely sa.tisfaebory ia }bs lint q.ualit, y bu~ was still }~eterogeneous :for seed size and had ~oo low a ginning oul,- tn.rn. Seventh. baekoross ma.terial replaced the fif~a.backcross in balk prop~,gatioa~ an,.1 this, in turn, wi].l be auceeeded by o. final wave of zfinCh baokcross origin, o:t"the :following composition: [((.B 3I x X 1530B) x gT 2136 ~) • NT 2/37] x NT2/38 ~.

* X 1530 is a strain of Sakel orighl whiel~ was Bred in t;he Sudan. Subseleet,ions l:rom it wore distinguished alphabetically; thus X'ISg0A was the firsb a,dva.nne on X15:J0, Xlg30.B was a stit] laser se.lee~io~, a.nd ,so on, the later the iet~er of t,he Mpha.bet added to the strain number, ~he more modern ~he substrah~. "]~ ~[~he C4ezira is a. ~z'acb of hmd 1.5-hlg between the Blue Nile a'nd t,he White Nile. In ~his ~u'ea about 200,000 ~cres of Sakel eo~on are grown amnm.lly, irrigation being e~m:ied cab by gravity flow from the Sennar Dam. ,Sh~mbat is in ghe Northern Sudan hi the immedial~e vicinity of Khar~oum. 80 rBaCkC'~'O,~S techniqtce in cotto~ breeding Oraitb/ng selee~,i.ou and linkage effects, ~he percentage composigion of [hose three bnllcs should be: ,q~t~in 5t)h baekoross 7tli ba,ckcrOSS 9~,h backcro~s Bal t.d 0.4 0,1 ~].5aoB l..S O-~t o-t NT 2/36 9.4, 2.3 0.6 NT2/37 12.5 3.I O.S NT2/38 75,0 93-8 98--[ ('The NT:~/38 ps,i'eu~ or" ~hese crosses was reseieeted for leaf~eur] resistance each yat~r bet was nor given a new number.) This c{~ms~ion of ushlg, each year, the la~es~ selection of the b~ckcross paren~ sb'ain IXi.gyseen[ ~0 llaxre been unduly sfiressed. The point, however, is of pai'ainounb ilnportance, ail,-I it has not received the atb?.ntion it: nierits .[.tt hh/bridiza~ion .progra, m,nes. To sam rtp, t.he fina, l backorosses s.re relatively tulii.tipo~?eal.~.[ in eli.minafiing I;he gen.o. t)q?e of ~he donor pa.rent bur are essential in bringi.ng the s~rain up-to-date. Thus, thoug~ the eighth, ninth and l:.enth baekerosses toge[h.er remove only 0..35 ~ c~f ~he don.or p~:cen5 genotype, I;hey put_ in 87'5 o.,/o of the 'blood' of elle latest subselections.

(3) Progeny.!/size -\Vhere single gone gransference :is the objec,t, the question whe~,her to use small or large progenies arises. Each syst;era has its advantages in certain, circumstances, l~ut ~he small progeny method is of more universal application. (a,) ,SmaZ[pro 9..n~es. Where a single, easily distinguished gone is being ~ransferred, only one selection, for fur~he.r backcrossing, need be made in each baekcross progeny, and i.g is therefore only necessary to grow a f~nlily sulFleie~ltly large to guarantee bhe presence of one such plalit, with a margin of safety to allow for possible poor germination and insec~ damage. )"or determiaahion of family size Na~her's (1938) TaMe 3 is excellent and the following ~.gm'es for 1 : 1 and 3 : 1 expectations are taken from i~. Table 2. ~S"ise@'r .refu.isite to give at least or~e ph~nt co~..tain'ing the. t,'ra.~v~fe,rredge,ne Level of prObarbLli~y .Ra~io ~ .w x expected 0-900 0.950 0.980 0-990 0,995 0.998 0-999 I : I 3'3 -~-3 5-6 6.6 7'6 9-0 I0-0 3 : 1 S.1 10.4 I3-6 16.0 18,4 21-6 2-t.0 The numbers j.r~ ~he body of ]'able 2. show, for various levels o:[ probability, the size of progeny which should be grown, in order to predates a~ Ieas$ one-plan~ containing the gone or genes being transferred. It will be seen that a family of tenplan~S is adequate (p = 0.999) to ensure gee presence of the required type is. a backoross, where a single gone is being transferred ; in a baekcross involving a ~wo-fac~or {ransference, a proger~y of ~xgenty~four p2ants would be desirable. r~Ih..se f~ figures refer to p/.en~s, and suitable allowance nansS be made ae sowing time to cover risks of poor germination and Dest damage. (b) Large proyer~iss. Where research and oreedm~ mus~, lie co.ttettr throngh a lack of genetie.aI knowIedge of ~,he character ~o be transferred~ large progenies are essential, Wide basis backcrossing is also sui.~ed go int;erspecific crosses where ~he di:ffereuces in appearance are greab (e.g. in mesa/~i,rs'~tu.nz • barl.)~denss crosses), sii~.ce ig is possible, by s.sit~g large progenies, 5o speed u.p the elimination of-She d.o.uor geno~ype. For e.he firs~ ~wo or 6hrec backcrosses very large progenies are grown-~.a.~honsa~td plan~s, if possible. All planes no~ po.ssessing eh.e factor, or fact;ors, being ~re,nsfem:ed, are l?cdled up. The ]%. L. /{~mnT 81 rem,~ining plants are then rogued .severely on their vegetative similarity to the backcross parent type. The most off-type plants are removed first, t]~e process being repeated till only a few plants are left. If a single factor is being transferred, these remaining plants can be rogued down to ,some four or five, one of which is chosen for crossing and. the remainder kept as spares. This process of severe roguing (or selection) is of great import- ante in. eliminating the genom of the donor parent since the. of plants showing donor pa.ren~ characters automatically removes not only the genes concerned but also, presumably~ a relatively large segment of chromosome in the immediate vicinity, since cotton averages oNy 1.7-1-8 chiasmata per bivalent. After ~he second or third backcross, the materiat is usually so near the backcross parent in appearance that large prettifies possess little or no advantage. Small progenies become just as effective in eliminating any of the donor genotype which may remain, and mech. time and unnecessary labour at? saved. This system is valueless in intraspeeifie crosses because such hybrids, by ~he first back- cross, are normally very similar vegetatively to the baekcross parent. This is one of the dallgers of hybridization--the hybrid may look like the backeross parent bu~ it still contains a large proportion of the donor genotype and is unlikely to true {'or the various qualitative and quantitative characters desfl'ed. To sum up : the most profitable size of backcross progeny to grow depends on the number ~f visible character differences between ~he original parents. Where a large number of differences exist these give a valuabb basis for selection for the elimination of the donor genotype. Hence with many sloth visible differences the best.policy is to grow large early Mekcross progenies and to concentra,r on selection as a means of rapidly removing ~he donor genotype. Large progenies are best obtained by treating a pure line of some fifty plants of ~he baekeross paten5 type as one female parent.. With few visible differences between the parents it is advisable to grow small baekcross progenies and to concentrate on the olin_tinsdon of the donor genotype by making as many backcrosses per year as possible. (~i) .Bc~.sis of sdectio~ ~;~, hgN'~:gs Selection. in baekeross progenies should be limi~.ed to vegetative characters %oh lint except where the object, is liJ~t improvement and lint genes are being tra:asf_erred in- tentionally. The aim should[ be complete similarity to the backeross parent and the avoids.nee of all forms of hybrid vigour, such as increased )deld. or longer lint. Frequently F1 types have excellent lin.% and a too early use of li~lt as a basis for selection migb~ merely mean %hat She more heterozygous plants were bein.g select:ed. On the o~her hand, differences in plural, appearance are. presumably, due to numerous genes, and l~heir s would involve the ramoval of numerous unwanted chromosome segmem~s. Lin.t examination can be made when vegetative similarity is a.~taiaed but, even then, lint length and qaatity should be used more as a measure of success than as a basis :for selection., BNk seed fro:in un~elec~;ed pla.t~ts in the later bs.ckcrosses is used to sow tri.'s,] plo~,s from wt]ich ling samples can be obtained %r spinning and grading (see next section). (5) Tiw e~d-2o@,~ A major difficulty h~. ba, ckc:eossing is to know when to cease crossing and start bulking; '~eVeral promising hybridization programmes ihave failed for lack of a method of deter- ~aiaing the correct end-point. Usually an arbitrary point, such as the fourt]h baokeross. Journ. of GeneS/ca g7 a 82 Baclcc~'o~e tect~.,~ict,~e i'~, cot~o~ breed,i,~..g seems to be chosen. :for bulking, and the fu,ilure of I;he balked prodnot attributed to a deloe~erious linkage. Such a linkage might exist, bar, it woald be desirable to prove i.~a p~:esence, more conelusiveIy before dise~rd[rLg one's breedi.ng material. A~ arbitrarily chosen, end-poiN; would probably be eOml?letety satisfactory in fto~-r~? breeding or it~ hybridization work on many food crops. The cotton breeder, howe~ver, has. a much more complicated probiota: not only must he wal;c.h retch ~izgxieuI1~[ri[ qualil;i~s, as vigour, yield and ginning out-tin'n, but also liztt quality--spin.ni~g quality, stren.gt:h, Stal?le, lustre, COIoI|],r. fineness and_ : Nel' ; even tlii.e percezztage of otiia the seed ia of corm mereial i.napo~:tance. In the better qlu.Mity cott;ons (See Ishl,nd and[ Sakel) t]m market will only permit varlet.ion of lint quality within very narrow limits. It is for this reason that no arbitrary end-point :fol: backcrossing can be fixed--dhe breeder of long sta.pled cottons must ob~,aiiz as his I?roduet a strain entirely unaltered as to its qualit,y, ego., except %r the effect of the gone, or genes, inl~entionally I:ransferrcd. No question of balking ~hould arise unt;il the baekcr,Jss progeny it vegetatively homo- geneous, except .fur the presence or al~senee of the tz'a.nsferred gone. Whml homogene[U. is a~t.ained bc~clcerossi.~9 shotdd be co.~,t.i)v~ed, bu.t al~ the same lime bulk seed of all aa plants ilz the progeny shoMd be collected, A second bull< should be made consisting of seed o~ all the Aa plants. These two bulk seed lots should be put into a yield trial (or, if the seed is inadequate, into a replicated triaI for tint test) with the backoross parent type as controi, The seed from the Aa plam.s wotdd, of course, prodace a mixture, but tlfis sh.onk{ contain enough AA and Aa i~[ar~ts to show any m.a.jor faults due to ]inkage or t;o pbiotropio effects, If the lint (or other characters) of this AA, Aa, aa mixture As Below the backoros:s pa~en~ standard, the Niluze may either be due to linkage or to the portion of the donor pare~ genotype still remaining in the hybrid. The comparison between ehe aa plots and the. backcross parent strain will show whether or not the failm'e is due to linkage (and/o, pleiotropy). A scheme of baekcrossing should no5 be abandoned unless the homozygons recessive material (the aa ]}elk) is qualitatively and quantitatively indistinguishable fl'om the parent type and the heterozygous (Aa) bulk a Nih~re. Eve~ then. one more test ox a later baekeross would definitely be desh'able Lefore s discarding the material. If both types fail (as will most likely happen), the test should be repeated on m~ccessive backcrosses matil either the presence of deleterious linkage is proved or the heterozygous bulk is found to be the equal of the backcross parent strain. At this stage bnlking shc~zId. be started, and, since backcrossing will have gone on whilst the foregoing tests were being made, the type bulked will be one backeross ahead of the type which succeeded. Where it is proposed to use the produce of this baekorossing as a female parent in in- tegrating a still newer strain, it is advisabIe to take the backcrossing one or. more staten farther. Thus a SakeI strain to which A has beez~ transferred by five backcrosses may :prove to ]~e a commercial sneeess and be propagated in bulk. Later it may be decided to a<[d, say, a Tangnis gone for j assid resistance and a ,p~r gone to into'ease blackarm resistance using ~he AA Sakel as backeross ~parent. These additional crosses may well fail if the ffft]~ Sakel backcross AA material is used as backeross parent because the po.rcantage of do~0~ blood left in t,his suceessfld ~fth SakeI backcross may be fatal when. there is added ~o it th~ re,idu.a], t)ercentages :from the Ta~.xgtfis and 2t~r~c~et..~.~.. For this reason, it is advisable go carry out two or more apparendy unnecessary additional baolcerosses in order to remo~'~ ~s much as possible of the donor .blood h-ore the AA Sakel before using this as a b~ckerosS parent. ~. L. K~t~T 83 (6) ~ldditive fa, c:or~ In transferring two or more factors with additive effect, it may be necessary, or a,t I~ast desirable, to se])araCe them and deal with each individually to avoid l,he danger of the ,~0eidental loss of one of {~hem. Where the additive effect is marked= so that there is no possibility of con.fusing the AaB]~ phenotype with either the Aab5 or the aaBh, then obviously, sepaz-at~ion of ~he two genes would merely entail extra 1about. Where, however, ~he AaBb phenotype merges inl,o either, or both, of the others, separation is likely ~o save ~0nsiderable labour, since the plant breeder will only need to de~'d with two families per generation--a, baekeross progeny containing Aa and aa plants and a progeny consisting of BB and bl~ pla.nts. In each of these families he roll only require one selection for further backcrossing~an Aa plant and a 135 pla.~t. Had the genes not been separated, far more selections would have had to be backorossed to avoid the danger of losing a factor (Table 2). When the end-point, is reached the ~wo factors are recombined by crossing the A and B strains together, baekcrossing to the strain containing the weaker factor, salting, selecting and resetecting. Thus suppose A ,~nd I3 are dominant genes showing weak addi- tive effect with A the weaker factor of the two. Then AA and I313 strains are crossed and ~he E~ -backcrossed to the AA strain. This should give :

(j) 1AA13b : 1Aat3b : 1AADD : 1Aab5 The last~ two gano~ypes in (i) wilI have the same appearance and., since they contain only ~he weak factor A, plants belonging to this group ca~x be discarded. Plan~s of A13 phenot)~?e should be selfed and ~heir progenies grown in rows. All rows containing a proportion of bet, tom recessive (aabb) plants should be discarded. The remaining rows should show a 3 : 1. rat,io of AB plaut, s te A plants as follows :

(ii) ]AAE13 : 2AABb : 1AAbD 2'his ratio will be clan.r-cut: and easily seen because, aIthongh B plants and A~ plants might overlap in phenogypic appearance, the AB phenotype will be distinct frnm the group containing only ~,he weak factor A. Ati AABD pl.angs in (ii) should be discarded and a nnmba.r of the .remaining plants selfed for sowing in progeny rows. All ro~o~ eoataJ.nJng the A phenotype should be pulled up; the remainder would be of AABB composition, and would, constitute the first bulk plot of the new variety. This problem of a.dditive factors is best ilh.~s~,rat~edby a concrete example. The bin.charm resistS.nee f~cgors 13:, and B~ (both. dom[na.n~0 are a ease in l?oin~. Factor B 1 imparts weak resistance to Sakel whilst t~, alone, confers considerable resistance. Although B~ resistance is sI.lghbly incJ:eased by the further addition, of B~, it is impossible, itx mat~eriaI derived born late baekcrosses, to distinguish, wi{i c~.rtah~ty between I3~I3z and /ozB~ pI.an?0s or to ]dentil)~ the progeny of a. B~B~B~B~ plant as distinct fr,-_,n] that of a Bll)~13~B~ Plant. Yet, ~he sliglii~ extra, degree of.resistance echo.farted by the addition of 13~ should be of economic importance when the cotton is grown on a commercial scale. The method of solving this problem has been to, baekeross Bj and. B~ saps.rarely to 8akel, to cross the end-products and backeross ~o the tlaB z type. Th;. progeny of tihis cross Consisted of B~B~B~B~, B~B1B,]).,, BiBibzb.,. al?d B~bab~b~ plants in equal numbers. The las~ two genotypes, being only weakly resistant, were uprooted. A ~mlnbe~: of plants 84 .Poactccrosa technige~e in cotto~ breeding showing strong resistance (B~EIB.,B ~ and BibzB~,b~) were selfed and sown J.Japrogeny rows. At[ rows containing a.l?orportion of fully s[lseeptible plants were pulled up. The remaining progenies displayed a ratio of three very resistant plants to o~.e with steak resis~a~.r The weakly resist~n.ti plants were removed and a mnnber of t.he markedly resist;ant plants selfed. Progeny rows were sown from these plants. All progenies con- bai~ing a proportion of weakIy resistan~ plan~s were uproo~,ed and r remaining progenies hulked as the new resistan~ strain, all ~he pla,~ts in them being of BIBiB~B e composition.

(7) Li~..ked fizctors -Where awe linked factors are to be t)ransferred from one variety of co bton to another, it is advisable to select cross-overs in the first ba.ckcross progeny as parents for t\~rther baek- crossit~g. This enables the plan:g breeder to handle each ge~e separatel.y and the separat, ion of [he two genes fa,cilita~es the elimination of the segment of donor, parent chromosome between ~hem.

(8) Ble.ndi.ng i,,dzerit~,nce Blending inheritance is common in hybridization work with cotton. Its t~resenoe is apt to suggest ~hat inheigahee of the character concerned is based, on a complex of cu.mnlative genes with small individual effect. I~ may be, however, that the character is, in the main, governed by one or two major genes and that the genetic ratios are masked by numerous minor genes and modifying factors. Such minor genes may have a stronger effect on the bottom recessive in a cross than on plants containing the major genes governing thB character, and, for l,his reason, although they are strong enough to mask the scheme of inheritance, such genes are often of slight economic wlue. As there is no known ~eohnique whereby a host of minor genes can be transferred from variety {o variety, such genes are. best eliminated since their presence will only confnse the issue. Blending inheritance may mean that, on our p_resent state of knowledge, the character concerned is no~ transferable from one variety ~o another. The point is that, because no clearocut segregation appears in the e~rly stages of a hybridization pro~amme, it does no~ .,~,ecessa'ri~y fotIow that the character cannot be transferred. It may prove im]?ossible to transfer it in the-full strength which it displayed in the donor parent, but at ]east some portion, possibly the major portion, of it can probably be transferred. A plant breeder, faced with blending inheritance in a first backeross, should select, for f~trther baekcrossing, plants showing the magma[ effect, of the character required: in the hope that, in la~er backcrosses, clear-out ratios wilt be obtained.

(9) Btdlc propagation (a) ~'u~l dornin(~nce. The firs~ deciion to be made is the size of propagation plot to be sown. Under irrigation conditions, an acre plot, sown at a spacing of 90 x 90 era. at one seed per hole, requires about 4200 seeds. At Shambat, an avergge Sakel plant sown at ~.ormal crvp time should yield upwards of 250 self-bred seeds and, winter sow-n, it should give abou~ :~00 such seeds provided suitable precautions are is,ken. To produce el200 seeds of AA co.mposl~ion from a. winter sown backcross _F,,therefore needs a family containing at least 2t. AA plan~s, so that t,be ~otal F~ population mast not be less than 80-90 pla;lts, The F z is gJ:ow~t our-of-season (winter) and. recessive (aa) plants are eliminated. All plant~ belonging to the A phe!~otype are selfed and given serial nmnbers. As soon as two or 1%. L, KNm~ ~. 85 ~ore bolls have ope~)ed on all. plau~s, progeny rows are so~w, each progeny consisting of ~en holes sown with three seeds per hole. These rows are oxentined and all numbers con~ ~aining any recessive (aa) plants are noted. The parents of these particular rows are then p~Iled up from the F~, leaving only plants homozygous for A. Seed from th:ese plants sows the hulk propagation plot. In this plot the ilzdixddual progenies are kept separate as an additional safeguard. Using this technique, it is possible to produce, from a small baekcross progeny in one season, seed for several acres propagation area of homozygous material ie the-next season. Seed from the heterozygotes in the ~v~ can be usefully employed in a small yield or lint ~rial (see w (5), ~The end-point'). Where only small plots are required, the -~a progeny rows can, themselves, be used as {he propagation area after remo~dng ' splitting' progenies. (b) Pa~'~ial domi~,a,~me. Where the transferred gone though not recessive, lacks full dominance, hulk propagation is simplified. To avoid selfmg, the backcross iv= should be sown in an isolatect position. Aa and aa plants sho~dd be pulled np and the AA plan~s leg to produce bulk seed of the new strain for further propagation.

(10) ~urit~ of baoIccross ~)a,re~zt The foregoing account of baekeross technique suited to cotton breeding is based through- out. on the assumption that a pure breeding, homogeneous variety is being used as the ~aekeross parent. Baekorossing is only one of the tools of the plant breeder, it is not a ~omplete system of in itself and the plant selectionist must firs6 have produced plu'c strains before baekerossing can be used at all, k strain showing marked heterogeneity woul.d be usdess as a haokeross pa.rent bug it should be possible, by suitable modi~eations of ~echnictue , to transfer genes to a variety showing a certain amonn{ of lle{erogeneity. The technique would be to stgrt with several/~1 families, each arising from a cross taet~-een the donor'parertt and a different plant i~ the backcross strai~. These F~ families would then each 5e backo~'ossed separately, on the :small progeny' system pre- viously described, and[ the end~prodncts bullced, l?or a strain sho~dng only slight he{ero- geeeity, probably three or four separa.te lines of baekerossing wend be su.f~dent; obviously the number required would increase with the degree of variation shown by the ba.ekeross >~rent strain. In baokcrossing sac5 ' line' it would :be preferable ~o ass ' 1)ulk backerossing' rather g]aa.n to select a single Aa plant in each line. The aa plan.ts in each progeny would be eliminated and. the Aa plants treat, ed as a single nmle parent for use in pollinating the room:rent parent.

SxJh~Ma~.~- Baekcrossing, though snecess.Nl in other ero]?s, has, in the main, f~dled to produce economic results in c.ottot~. The n.se of the teehnictne discussed ~a this paper has produced seve~:al commercially successful interspecifie gone transferences. The following suggest.ions are made : 1. The hybrid should be the male pare~t. 2. The latese substrain of the backcross parent variety sitould be used each season as female ]?arent to kee l) the crossing programme up-to-date. ?,. Where a large number o:f visi})le differences exist bet~-een the original parem,s, these provide a valuable basis for seleetiom In such a case large early backeross jorogenies shoaid be grower and severe selection in the tield utilized to accelerate the remedial of the 36 .t~ac/cc~'o~s techniq,~e in cotton b,reeding donor parent ge~aotype.. With few visible differences between the parents i~ is advisable to grow small baekcross p::oge~fies and to concentrate on ~he elimim~tion of ~he donor ge.aotype by making as many baekerosses per year as possible. -1-. Selecgion of hybrid plants Nr :Nrther baekcrossing should be made solely on (a) presence of tr~nsferred gone and (b) vegetative similarity to the backcross parent. All oh.stutters likely to be due eo sho~fld be aroided, e.g. longer lin~:or h.igher yield than the baokeross parent. 5. An arbitrary end-point in backerossi~zg should be avoided. The criterio.tz should, be a repl.[ca~tcd test of bulk seed[ 5:u,u heterozygoges :from the backeross, ag,~inst bottom reeessives from the baekoross, agah~t tt~e 5ackeross parent as control. When the hetero- zygous bulk is qualitatively and quantitatively equal to the backcross parent, bulk propagation and large scale test.big should be sta.rted. 6. 0umal.ative facgors with only sligkl; additive effcet should be separated ia back. crossing.and recombhaed ]a~er. 7. Linked factors are best separated ~o facilitate the elimination of the donor parent chromosomal segment between them. 8. The ~ppeara,nee of blending inheritance in a ~rs~ backcross need not discourage a plang breeder since inheritance of the character may still be mainly drte to one or two major genes and clear-e~ ratios :m~y appear in later b~ckcrosses. 9. A method of bulk propag~tiott from a bsckeross progeny via an out of-season baeb cross F~ to an F~ komozygous propagy~tion plo~ in the foliowing season, is discussed. 10. A technique is suggested for transferring genes to a strain showing moderate heterogeneity.

I h~ve pleasure in ghanking 5![essrs J. B. Hutchinson and I1. A. Silow and Dr S. G. Stephens of ~he E.0.G.O., Cotton Research Station, Trinidad, for reading the manuscript of this article and making valuable criticisms and suggestions.

I~EFEP~ENOES D~C~r~B, 14. C~ & ~{!LL~,2., ~V, L. (19"~2). ~2z Ezp. ,~t~s Emp. d'ot~, (-+r.Corp. 1940-1, p. 159. ]~v~L~, S. H. & I:L_~L:txo, S. C. (1934). ~e~. Ezp, ,%~s Em~. Cott. Gr. Corp. 1932.-3. West Indies, 1930 3, p. 200. H_~_l~r,Az~, S. C. (1934). The va[tm of int.exspeciflohybrids in cothon from the standpoinl of genetics. Y~e2. Emp. Oct,. Gr. Cor~, Jtdy 1934. K_~r~,~D, S. C. (1935). The genetics of cotton. XIL J. Goner. 30, 4~5-76. HuT:cffn~-so~LJ. ]~. (1938). Note on a policy of intrcductbn of new v~rieties of cotton in Africa: Emb Corr. ~r. ~v. 15, 284. T(_Wj:OJKT,P~. L. & CLOUSTOI~-,T. ~V. (1939), The gene~ies of blackarm resis<~nce, I. J. Go.net. 38, 133-59, L:~EnT, A. P~. (1938). New Sakel sSra.ins in the Anglo-Egyp~ian Sudan. Emp. Cert. (4r. Roy. 15, 15. ih'L~,TH~, Is (1938). The A(easurement of Linkage in Heredity, Table tli, p. 1~7,