64 BULLETIN 813, IT. S. DEPAETMENT OE AGKICULTUKE. 
proved to be true bud mutations usually appear more frequently and 
more abundantly in trees of some of the unproductive or otherwise 
undesirable strains than they do in trees of the standard strain of the 
variety. 
The presence of variable fruits on all the trees and the fact that 
some trees produce many more such fruits than other trees emphasize 
the need of having reliable individual-tree performance records for 
use as a basis in selecting trees from which bud wood can be taken for 
commercial propagations. It is not sufficient to know that certain 
trees produce heavy crops, but data must be available to show the 
number and kind of variations on the trees, so that only those will 
be chosen which bear the most uniform and desirable fruit. 
In Table X are presented the total number of the different variable 
forms which were recorded for three years on the trees listed in Table 
VIII. The typical fruits of the Pear-Shape strain produced on the 
trees listed in ranks 71 and 104 are recorded as bottle shaped, and 
other fruits on the same trees having a tendency toward that typical 
shape are classed as collared. These collared fruits are believed to 
be of a different character from the collared fruits commonly found 
on trees of other strains, but being like them in appearauce they 
were listed in the same class. Propagations have been made to 
determine this point. 
Wherever no record was made of any factor for one or more months 
it is thought that the most practicable method of computing the 
monthly averages in Table VIII is by dividing the monthly totals 
by the number of months represented in each total. For example, 
the averages for the production of Green-grade fruit by the tree in rank 
1 in Table VIII during July, August, September, and October were 
computed by dividing the monthly totals by the factor 5, as there 
are records for five years during those months. In November, 
December, and January the records for six years are available. In 
February the pickings were missed in 1914 and 1917 and the average 
for that month is obtained by using the factor 4. In March records 
were secured during five years but the crops harvested in that montl 
in 1914 and 1917 were in reality the yields for two months in each 
case, so the factor 7 has been used in determining the average for 
that month. It is recognized that this resulting average is not strictly 
accurate, but it is believed to represent more nearly the correct aver- 
age for the period than any other figure that can be presented in 
this table. Because of the use of different factors, as just mentioned, 
the totals of the monthly averages are not equivalent to the average 
for the yearly totals. Other instances in this and other tables will 
be found where the decimal portions of averages of total records do 
not exactly correspond with the totals of the averages of the corre- 
sponding numbers. This is due to the small errors occasioned by the 
practice of retaining only two decimal places in average figures. 
