26 BULLETIN 1121, U. S. DEPARTMENT OF AGRICULTURE. 
virtually identical in the cases of inbred females mated respectively 
with brothers (young inbred) and with unrelated inbred males (CO). 
When the crossbred young themselves become parents there is 
a marked increase in the regularity with which litters appear. The 
result is practically the same whether these crossbred parents are 
unrelated (CC) or brother and sister (C1), again indicating that 
the vigor or weakness of the young is not a factor. A falling off 
appears when the parents are from the first generation of renewed 
inbreeding (C2). In these respects there is considerable similarity 
in the results shown in Figures 8 and 12, dealing with percentage 
born alive and birth weight, respectively, characters which we con- 
cluded were largely dependent on the dam. 
There is, however, a striking contrast with those cases in the 
standing of Experiments CA and AC. The great increase in fre- 
quency of litter when a crossbred male instead of an inbred is mated 
with an inbred female (CA) seems to mean that the sire is most apt 
to be responsible for irregularity in producing litters. The mating of 
inbred males with crossbred females (AC) however gives a better record 
than where both parents are inbred. This indicates that the female 
is also responsible to some extent. The still greater improvement 
where both parents are crossbred (CC, C1) substantiates further the 
responsibility of both parents. 
It should be recalled here, however, that Experiments CA and AC 
are compared with inbreds during a period when the latter were 
producing litters more frequently than the inbreds breeding simul- 
taneously with CC and Cl. The actual records of AC and CA 
were only slightly below CC and Cl and were so near the upper 
limit possible for a guinea pig (which is about 5.3 litters per year) 
that the superiority of CC and Cl over CA and AC is probably 
somewhat exaggerated in Figure 16. The same considerations 
apply to Experiments CG and CL, which would probably have made 
records more nearly like CC and Cl under strictly comparable 
conditions. The main conclusion that frequency of litter depends 
primarily on the sire and secondarily on the dam is not weakened. 
Another important result is the great superiority of crossbreds 
derived from only two inbred families over the random-bred stock. 
In the other characters with which we are dealing there is merely a 
recovery of the condition of the latter. It is probably not a coin- 
cidence that frequency of litter is the only one of our characters in 
which the miscellaneous inbred families (OI) are inferior to the five 
largest families (2, 13, 32, 35, and 39). It seems probable that 
regularity in producing litters has been the most important factor 
in the unconscious natural selection among the inbred families and 
that the crossbreds are in this case derived from a selected ancestry. 
The possibility of improvement in all characters through conscious 
