1024 
Journal of Agricultural Research 
Vol. XXVIII, No 10 
Only one F 3 family bred true for the emmer type of keel, while 13 families 
were homozygous for the vulgare type of keel. Most of the individuals were 
apparently heterozygous and there is no indication of a simple ratio. 
Grain color is a very difficult character to classify in F 2 , due to different degrees 
of hardness of the grain. A very hard endosperm will give the grain an amber 
appearance, which may sometimes be mistaken for red. However, the F 3 behavior 
indicates the true genetic composition of the F 2 individuals and the classification 
of F 3 families into homozygous and heterozygous types is less difficult. All 
individuals classed as light amber in the original code have been considered as 
white in the present classification. The segregation approximates very closely 
that expected for a two-factor difference, the number of homozygous whites being 
practically 1 in 16 F 2 segregates. If two factors for red grain are involved in this 
cross, however, there should be a much larger proportion of the F 3 families 
breeding true for red. There should be forty-two F 3 families breeding true for 
red and forty-eight families segregating either into 15 : 1 or 3:1. Also due to the 
paucity of the individuals i» F 3 families, the twenty-four families which should 
show a 15 : 1 segregation would in many cases be classed as homozygous for at 
least one factor for red, so that the proportion of red should be in excess of 42, 
which does not agree with the actual data. 
The F 2 segregates were also classified in respect to type of head, whether they 
resembled vulgare, compactum, Spelta, etc. The types which bred true in the 
third generation included 4 vulgare, 1 Spelta, 2 durum , 2 dicoccum , and 1 classed as 
abnormal. There were also two compactum segregates in F 2 which bred true to 
the type in F 3 , but it is probable that these plants were mixtures, as they were 
completely fertile in all cases in both F 2 and F 3 and were extremely uniform in 
all characters. 
The date of ripening in F 3 ranged from the 10 th of August to the 9th of Septem¬ 
ber. Only eighteen F 3 families were apparently homozygous for date of ripening. 
Fourteen families which were homozygous ripened between the 10th and 13th of 
August while most of the heterozygous segregates had a mean date of ripening 
ranging from August 13 to September 9. There was some correlation in F 3 
between type of spike and date ripe. The Spelta segregates ripened earlier than 
the abnormal types which could not be classified. 
A number of correlations between various factors were obtained in F 3 , and 
between the F 2 individuals and F 3 families of the cross Marquis X Alaska. 
Segregates in F 2 which were comparatively sterile were also relatively sterile in 
F 3 while the more fertile F 2 individuals resulted in comparatively fertile progeny 
in F 3 . The correlation between grains per spikelet in F 2 and grains per spikelet 
in F 3 was .54±.05. There was also considerable correlation between the various 
characters of F 2 and F 3 segregates. For instance, culm type in F 2 correlated 
with culm diameter in F 3 gave a correlation ratio of .59±.04. Type of keel in 
F 2 correlated with keel type in F 3 , using the coefficient of contingency, gave a 
value of .52+.06. Height of plants in F 3 correlated with date ripe in F 3 gave a 
value of r=— .18±.06, indicating that vegetative vigor had little relation to 
the date of maturity. There was some tendency for the more fertile F 3 segregates 
to ripen earlier, as indicated by the value of r= —.25±.06. In these various 
cases the mean value of the F 3 families was taken. Using all of the 1536 F 3 
segregates a correlation of .62±.01 was found between grains per spikelet and 
height, indicating that fertility increased as the vegetative vigor increased and 
that much of the sterility was associated with poor vegetative development. 
It might be expected that F 3 segregates resembling the parent would be some¬ 
what more fertile than intermediate or abnormal types. In Table IV the various 
types of segregates are shown, the number of such segregates, and the average 
