14 EXPERIMENTS WITH DROSOPHILA AMPELOPHILA. 
The building up of an abnormal strain from a long-inbred normal one 
was also nearly completed when it was stopped by accident. I did not 
think it worth while to start it anew, as its accomplishment would prove 
little, since it might be said that the normal strain was a ‘‘ mixed general 
population ’’ due to normality (inhibiting) units masking all sorts of 
latent abnormality units. 
In nature a small percentage of flies have the fifth longitudinal vein 
somewhat shortened (see fig. 47). This variation also appears in the 
experimental strains. Rather as a matter of curiosity, I selected for 
shortened veins during a few generations and very quickly obtained such 
specimens as are illustrated in figs. 48 to 50. One can not go further 
in this direction without some special technique, because the wings, 
lacking the support of the veins, droop and catch in the fly’s food. 
Probably breeding could be continued by cutting off the parent wings 
when matings are made. I did not try it, as it was already very evident 
that selection was just as effective in the negative as in the positive 
direction. 
On the other hand, all attempts to fix, by selection, some particular 
type of abnormality utterly failed. It was thought possible that the 
great variety of forms which the extra veins showed was due to a mix- 
ture of a number of simple forms and that selection might isolate these 
simple types. The most hopeful was a simple forking of the second 
longitudinal vein (see fig. 10). Selection for this type was started sev- 
eral times, but never went beyond the fifth generation, because, although 
there were plenty of abnormal flies in each generation, there was no 
increase in the number showing this particular type, and sooner or later 
a generation would contain none of them from which to breed. The 
same was true in the experiments aimed to fix the abnormality on, for 
example, the third longitudinal vein, but to keep it off of the second. 
It is easy to have all the abnormal flies abnormal only on the second 
longitudinal vein, providing one be content with small abnormalities. 
However, as soon as one increases greatly, by selection, the abnormality 
on the second vein, the other veins begin to be abnormal. 
These are the facts: Starting with slight extra veins, either in wild 
material or in material selected and inbred for normal venation, we 
ean quickly get by selection 100 per cent abnormal offspring. In future 
generations this strain can be quickly brought back again to its normal 
condition by selection. Selection also quickly shortens the veins and 
would probably largely do away with them, provided some technique 
were adopted to keep the results of selection alive. But selection, 
accompanied by the strictest inbreeding (brother X sister and parent < 
child) failed to isolate any unit characterized by a given form or extent 
of abnormality. 
The interpretation of these facts would doubtless vary with varying 
opinions as to unit-characters. 
