84 SEVENTH REPORT. 



3. With the data in hand we may proceed to a discussion of the 

 results. The first thing one notes in looking over the tables and dia- 

 grams is that all the individuals reared under experimental conditions 

 average shorter than do the individuals in the "wild" culture (Table 

 II, and Diagram I). Furthermore the means for the 200 hour series 

 are lower than those for the 100 hour series, and those for the 300 

 hour series are still lower. In other words it appears that the condi- 

 tions under which the organisms were put in the experiments tended 

 to cause a decrease in the mean length of the body,' and the longer the 

 stay in the experimental vials the greater the decrease became. We 

 are not yet prepared to make any positive statements as to the cause 

 of this tendency for the organisms in the vials to decrease in length, 

 although we hope to present conclusive evidence on the matter in the 

 complete paper. The case here very obviously resembles superficially 

 the well known results of Semper and de Varigny on certain molluscs. 

 They found that these organisms (e. g., Physa, Lymnaea, etc.) decreased 

 in size when kept in relatively small amounts of water. The size of 

 the organism appeared to be very directly related to the size and shape 

 of the containing vessel. Objectively this is just what we have here. 

 It might be thought that in the present case the decrease in the mean 

 length of the individuals in a relatively small amount of water is due 

 to failure of the food supply to multiply as rapidly as do the Paramecia. 

 There would not be, then, a sufficient amount of food present to admit 

 of completing the growth after each division. This, however, seems 

 unlikely, for the reason that the vials were never as crowded with 

 Paramecia as thriving portions of "wild" cultures very frequently 

 are. There are other reasons which cannot be entered into now, which 

 also seem to indicate that the decrease in length in the experimental 

 animals is not due primarily to lack of adequate food supply. 



The addition of cane sugar to the culture medium seems in a measure 

 to check this tendencv towards a decrease in length. This comes out 

 clearly if we compare the means of the control and sugar series. Thus 

 from Tables IV and VI we have 



Mean length (100 hour sugar series) — mean length (100 hour control 

 series) =3.180. 



Mean length (200 hour sugar series) — mean length (200 hour control 

 series) =10.300. 



Mean length (300 hour sugar series) — mean length (300 hour control 

 series) =31.140. 



The rate of decrease of the means is evidently much slower in sugar 

 than in control series. The same fact is shown in the diagrams. 



Turning now to the variation results we note : 



(a) That in the control series the variability is, as we should expect, 

 considerably reduced below that of the general population of a "wild" 

 culture.^ The mean coefficient of variation for the three control series 

 is 8.G00 (Table IV), while that for the individuals from the "wild" 

 culture (Table II) is 9.363. Or, in other words, the variation in the 

 samples known to have originated from a single ancestor is 91.85 per 

 cent of that shown by the general population, when in addition to the 



iln comparing variabilities we shall have to make use of the coefficients of variation throughout, for 

 the reason tliat such considerable changes in mean size have been induced in the experiments that com- 

 parisons of standard deviations will obviously be unfair. 



