2 BULLETIN 1164, U. S. DEPARTMENT OF AGRICULTURE. 
generation. The means for leaf index and boll index of the F 3 proge- 
nies which gave, respectively, the highest and the lowest mean for 
the character in question, differed by amounts equal to about 10 
times the probable error of the difference. There was a high parent- 
offspring correlation, for the character boll index, between the means 
of 11 F 3 progenies and the values for the corresponding F 2 parents, 
the coefficient having been 0.81 ±0.07. The coefficients of variation 
of the several F 3 progenies for the characters leaf index and boll 
index ranged from not greater than that of the less variable parental 
population to fully as great as that of F 2 . 
It is well known that segregation is much more pronounced in 
hybrids between different types of cotton, such as upland and Egyp- 
tian and upland and sea island, in which cases the magnitude of the 
parental difference is fully as great as ordinarily exists between very 
distinct botanical species. Cursory inspection of a second-generation 
population of such a hybrid shows the occurrence of characters not 
expressed by either parental type and the presence of individuals 
which bear little resemblance to any of the cultivated varieties of 
cotton. 
Systematic study of a hybrid between upland and Egyptian cottons 
was undertaken in order to obtain statistical evidence of the kind 
and degree of segregation occurring. It was sought also to deter- 
mine, by study of the correlations of characters, whether coherence 
occurs; in other words, whether there is a tendency for the char- 
acters to be associated in the hybrid as they are received from the 
parents. The results of this investigation have a practical bearing 
in throwing light upon the nature of the variants to be looked for 
in a field of Egyptian or of upland cotton which has been exposed 
to accidental cross-pollination by the other type. This knowledge 
should be useful both in determining the fact of whether such cross- 
pollination has occurred and in guiding the work of roguing to 
maintain supplies of pure planting seed. 
The cotton plant has both advantages and disadvantages as a sub- 
ject of genetic study. Certain advantages are the large size of the 
flowers (which facilitates cross-pollination), the readiness with which 
controlled self-fertilization may be effected, the large size of the seeds 
(which makes it easy to handle and count them), and the high visi- 
bility of many of the characters. On the other hand, the relatively 
small number of seeds produced by each flower is a serious disad- 
vantage, especially as the ordinary method of field sowing requires 
the planting of not less than four seeds to the hill, all but one of the 
resulting plants being finally removed. A good stand is not likely 
to be obtained when fewer seeds are planted unless a nurse crop 
is used or the seeds are covered with sand instead of with ordi- 
nary soil. Transplanting has not been uniformly successful with 
cotton, although this method deserves further trial. This waste of 
seed in field planting is a serious disadvantage from any point of 
view, but especially so when the computation of Mendelian ratios is 
to be undertaken. 
Another drawback is the long growth period of the plant, which 
results in the leaves, bolls, and fiber borne at different nodes being de- 
veloped under different weather conditions, thus unduly augmenting 
the variability of the characters. To obviate, as far as possible, the 
effects of this factor it was sought to measure on different individuals 
