96 BULLETIN 754, U. S. DEPARTMENT OF AGRICULTURE. 
With a new series of hybrids between these same varieties compris- 
ing some 20,000 seeds, a correlation has been obtained which more 
nearly approximates a 4-1-1-4 than a 3-1-1-3 gametic series. The 
correlation for these last hybrids differs from that obtained in the 
original crosses by more than eight times the probable error. From 
this we must conclude that the correlation between the same char- 
acters differs widely in different families, though the individuals 
within a family seem to approximate the same mean. 
We must further conclude that the association between these two 
characters in maize does not always fit the arbitrary series of 3 to 1, 
7 to 1, etc., proposed by Bateson, but that in some instances the 
couplings more nearly approximate the intermediate points of this 
series, such as 4 to 1, 5 to 1, etc. 
It is not intended to imply that the reduplication found in the 
gametes is the direct result of unequal cell division, although the 
observed results can be explained on such a hypothesis. The assump- 
tion that for the higher coupling ratios segregation must take place 
before synapsis in order to permit of the proper number of cell 
divisions seems thus far unwarranted by any cytological evidence. 
A study of the Mendelian behavior of the aleurone color has led 
to the realization that since characters are the result of many factors 
more or less independent the correlations must also be between these 
invisible factors rather than between the visible characters. 
The fact that correlations are between factors which can be de- 
tected only when in combination with certain other factors permits 
of many different degrees of correlation and lessens the degree of 
confidence to be placed in conclusions regarding both the factorial 
composition and the correlations or couplings. There is danger of 
explaining aberrant Mendelian ratios by the predication of correla- 
tions and to explain aberrant coupling ratios by assuming changes 
in the factorial composition. Because of this fact, great care must be 
given to the analysis of correlated characters from the standpoint 
of their Mendelian behavior, since auch depends upon the number 
of factors involved. 
Reciprocal crosses between plants heterozygous for endosperm 
texture and aleurone color and plants homozygous for the recessive 
characters waxy endosperm and colorless aleurone have shown in 
many cases that certain character combinations are deficient in the 
male gametes. It is suggested that this deficiency may be due to a 
higher death rate, less vigor, or a failure of the plant to produce 
these odd combinations in the proper proportions. 
If we can assume that correlations are the result of unequal cell 
division, we can further assume that the reduplicated character com- 
binations are also the most vigorous. Thus the tubes from the pollen 
grains bearing the reduplicated characters may grow more rapidly 
