292 
GENETICS: E. C. MacDOWELL 
denly, as conspicuous changes, or mutations. But the question is not 
settled as to how frequently, after their appearance, further changes may 
occur in these units. It is upon this special phase that the results here 
presented have their bearing. 
Most banana flies (Drosophila ampelophila) have four conspicuous 
bristles on their backs. A mutation occurred which permitted more 
than the normal four bristles to develop in this special region of the back. 
From one pair of flies produced by germ plasm bearing this mutation, an 
extra-bristled race was estabHshed (MacDowell)^. This race was distin- 
guished from the normal wild race by one Mendelian unit, as was shown by 
crosses which gave first generations with no extra bristles and second gen- 
erations (F2) in which one quarter of the flies had extra bristles. The 
number of extra bristles in this race was variable; experiments showed 
that this was largely due to the amount of food eaten during the develop- 
ment of the flies. Large flies, those from flourishing culture bottles, 
had numerous extra bristles, while small flies, those from mouldy, or 
old, dried up cultures, had few, or even no extra bristles. But even 
when no extra bristles developed, these flies, when given fair breeding 
conditions, produced offspring all of which had extra bristles. To dis- 
cover any changes in the inherited basis of this character (extra bristles) , 
selection for increased bristle-number was carried on for 49 generations. 
During this time every mating was made between brothers and sisters 
in pairs. Considerably over 100,000 bristle-counts form the basis of 
the following discussion. 
Two conditions must be met before selection can modify the means 
of a race. First there must be genetic differences between individuals, 
and second, these differences must be manifest to some extent in the 
somatic structures of the individuals. In other words there must be 
some tendency for extreme variates to bear extreme germ plasm. 
To measure this relation between soma and germ plasm, the coefficient 
of correlation is especially fitted. It is an expression of the degree of 
similarity existing between parents and offspring. All the data ob- 
tained from the inbred extra-bristled race were cast into tables corre- 
lating the grades of the parents and their offspring in each generation. 
The coefficients calculated from these tables show that there is an un- 
questionable positive correlation between the grades of the parents and 
offspring in the first six generations ; the coefficients for these generations 
are statistically significant. In the subsequent generations no further 
evidence of any positive correlation is to be found that has biological 
significance. The coefficients are sometimes plus and other times minus, 
but without any consistency or regularity. In most cases they are not 
