44 lULLETIX 1164, r. S. DEPARTMENT of ageicultuee. 
The various manifestations of sterility in the second and third 
generations of tins hybrid may be summed up as folk ws 
ttt **m « the .seeds to genmnate or of the plants to grow beyond the seedling 
flowed at^nS '"'"^ "' V " io,M 8tagCT °*' d -*»P™ent, so that few or nl 
lute'"Sf"t'e C nHtv t ' l0Pmem °'' *» "*** a " d P° Uen - ^^ to the point of abso- 
The frequency of partial or complete sterility, together with the 
appearance of numerous characters not expressed i„°e her pa rental 
race, some of which were of an abnormal or even pathological' nature 
are proof of the wideness of the cross between upland SZ 
Swh^^T tw i° varietles within tfce Egyptian^tvpe 
£2 «? f ' ha "}d showed no reduction of fertility and no expres- 
sion of extraparental characters. expres 
CORRELATION OF CHARACTERS IN P 2 OF THE HOLDON-PIMA HYBRID. 
The coefficients of correlation for all possible combinations of 38 
characters were determined on the 180 individuals of the F dodu 
latum on which all of the characters had been determined P The" 
correlations involved all measured or graded characters S Aspect to 
which the means of the parental populations differed by at amount 
equal to three and one-half or more times the probable erro of the 
difference, with the exception of leaf width «» 
formula CO f^ C nn nt llfi f J*™* 1 **™ ™! e calculated by the Jennings 
formula (S4, PP- 416. 4 7), using sorting and tabulating machines m 
the manner described by Collins and Kempton (.9. pp° 5 6 The 
whiVh Q?™' °l COefficl ?» ts of correlation determined was 703 of 
n n 1- >r ' e rega K ded f S ^cant, having been not smaller 
cient of 17' wl rl r ber t ° f mdlv ' dMls having been 180. a coeffi- 
cient ot 0.172 is 3.5 times its probable error. 
table 12 gives the significant coefficients of correlation, the prob- 
cats he^amf tM ^T the ni ^ bet ° f -dividual* I In all 
■'Expected W' I' P h 1? d m!nUS ? ,gns in the column he ^ed 
is to P be exnecfed if t "' b^ & P ° Sltlve or a ne S ative correlation 
bl ° expected if the characters are associated in the hybrid as 
cur ™™ 6d fl ' 0m / t I if P r ntS; m 0ther words > if coherence oc- 
";J h j f s | f ( l uence « f u the characters in both the first and second' 
P^7 to lll^n^Kl'T 6 ".^«ff«n«««l in the definitions on 
1x7s «„H I W 1 E . ' ,vh,c ' gi r s the statistical constants. 
Axis and branch characters are placed first, followed by the leaf 
bract, flower, boll and seed characters. Therefore, if it is desired to 
cWcter* wifl°[^ tl0n f ^t^' f °J CXam P le > bract Sfe 
character will be found as the second member of the pair in each 
and ^<r^&S% ia r £^^ti^t X Z^^' ii ' h r , ? V S- ^'^ P««^^h" 
were nevertheless included in the Mrrokl ions t ! £ ,""} I'}"'" of . 1! T 9 dl ", not lim ' t signiflcanfly, 
