340 J. A DETLEFSEN AND E. EGBERTS 
recourse to at least two methods of dealing with such unpro- 
ductive pairs. We can either include in our frequency distri- 
bution only those females on which we have ample data to give 
a somewhat reliable crossover value and ignore all pairs giving 
less offspring than a fixed minimum (fifty individuals for, 
example), or we can simply include all females and thus withhold 
no data. The latter course seemed preferable and we have 
followed it. There were five pairs showing lower crossover 
values than the one we selected, as follows: 10.0; 12.5; 16.0; 
16.6; 20.7. We did not always select the lowest absolute value, 
for in many cases this was based upon an insufficient number of 
offspring. It was also necessary to keep fertility in mind, in 
order to insure the perpetuation of our selected line. This 
explanation will make clear why we could not always choose 
the lowest absolute crossover value in the frequency distribution 
of any given generation. In table 1 the italicized frequency in 
each distribution shows the relative position and value of the 
selected pair. No dispersion can be given for the F2, F4, Fe, 
etc., since these represent en-masse matings. An x represents 
the point to which the progeny of the pair selected in the preced- 
ing generation regressed. The average number of offspring per 
pair shows how reliable the crossover values usually were in this 
experiment. The crossovers, total, and the crossover value for 
each selected pair are also given in the last column. Those 
generations which have any number of pairs entered under that 
heading are generations in which all matings consisted of pairs, 
while the other alternating generations were en-masse matings. 
Since the crossover value of a female may be based upon a small 
number of offspring in some cases, and thus give an apparently 
wide deviation which has little significance, we have not calcu- 
lated the variability of each generation in this paper. For 
example, a female showing a crossover value of 10 per cent based 
upon twenty offspring might well show 30 per cent if one hundied 
and fifty offspring had been secured, since age and changing 
temperature affect crossing over; or she might even show 10 per 
cent based upon twenty offspring as a sheer fluctuation of 
sampling. 
