250 MULTIPLE FACTORS 
and there was little increased variability, but in 
F, the variability was increased greatly: the ex- 
treme grades of the selected stocks again appeared, 
together with all intermediate values. The F, curve 
was not, however, one that could be reconstructed 
by simply compounding the P, and F, curves in 
1:2:1 ratio as in Morgan and Bridges’ cross of 
thorax pattern, for the proportion of intermediates 
was much greater than in a distribution so calcu- 
Fig. 61.—Series of arbitrary grades of hooded rats used in classifying 
results of selection experiment. Above the figures the numbers assigned 
to the grades are given (see text). (After Castle and Phillips.) 
lated; this again showed that more than one pair of 
modifying factors was involved. 
One of the most exhaustively studied cases of the 
effect of selection on a mixed population is that 
carried out on hooded rats by Castle and his co- 
workers, particularly Phillips. The pattern of 
hooded rats is shown in Fig. 61. The dark pigment 
covers the head and extends as a stripe down the 
back. The extent of the hood and the breadth of 
the dorsal band are so variable that in one direction, 
