320 
Francis B. Sumner 
As regards the second of the above stated conditions which should 
be fultilled by the experimenter, it is obvious that in the present in- 
stance I have produced readily measurable, quantitative changes in 
the parent generation. It would appear scarcely necessary to urge 
that a modification must first be manifested by the parents in order - 
that it may be transmitted to the offspring. Yet we read of one in- 
vestigator who watched for a reduction in the size of the wings of 
flies which had been prevented from flight for more than 40 gen- 
erations. It is difficult to understand how the length of an insect's 
wing could be increased or decreased by use during the life of the 
individual; and it is therefore scarcely surprising that such an effect 
was not found to be cumulative. 
If » acquired characters « are in reality transmitted, we should 
not in a given case expect that more than a small fraction of a par- 
ental modification would reappear in the first generation of ofispring. 
For this reason it is plain that we have more to hope for from a 
comparison of averages based upon the measurements of large masses 
of individuals than from a search for qualitative differences of a sort 
which are apparent to the eye. Yet with a few exceptions, past 
experimenters in this field have dealt with modifications of color, or 
of physiological reactions, such as do not lend themselves readily to 
quantitative treatment. Ever so slight measurable changes, if suffi- 
ciently constant, would be of far greater significance than occasional 
manifestations, however striking; for the latter are always open to 
the interpretation of being » mutations «, examples of » reversion « or 
the like, which are purely accidental as regards the conditions of the 
experiment. 
Having discovered suitable modifications in the parent genera- 
tion, there are theoretically two methods by which we may attack 
the problem in hand. We may either 1) raise the offspring of the 
experimental and control lots under identical conditions, or 2] we 
may raise the offspring of the modified parents under the same con- 
ditions as were employed to effect the original modifications. In the 
first case, we should compare the two sets of animals having differ- 
generation; or 2} that in any case the germ-plasm may be influenced during 
those early days of life before the animal has become strictly homoiothermic. 
1) For the sake of simplicity I have here assumed that one- lot was merely 
a *control« lot, i. e. one exposed to normal or indifferent conditions. In my 
own experiments, however, I have chosen conditions with a view to modifying 
both sets of individuals in opposite directions. 
