580 LORANDE LOSS WOODRUFF 



These data indicate that the P medium reduces the division 

 rate of both Paramaecium and the hypotrichs, while the H 

 medium reduces the rate of the hypotrichs and is without effect 

 on that of the paramaecia. 



At the end of the four days, Pol and Po2 were continued for 

 four days more on the same medium while Pp and Ph were trans- 

 ferred to the medium. Thus during this second phase of the 

 experiment all lines were on the uncontaminated medium. The 

 results were : 



Pol = 2.10 divisions per day 

 Po2 = 2.20 divisions per day 

 Ppo = 2.10 divisions per day 

 Pho = 2.00 divisions per day 



From these figures it is apparent that placing the Pp on the 

 medium brought about in the Ppo series a division rate (within 

 the hmit of error) the same as that of the lines (Pol and Po2) 

 on the medium from the start, thus proving that the reduction 

 of the rate in the Pp series was a result of the P medium. 



A comparison of the data from these two experiments shows 

 that the results are not in entire agreement, that is, in Experi- 

 ment 1 the H medium favored the division of Paramaecium, while 

 in Experiment 4 it was without influence. Such discrepancies 

 are not surprising when the large number of factors involved in 

 procuring heavy growth of the forms are taken into considera- 

 tion, but they make it apparent that results of value cannot be 

 secured without many repetitions of the experiment so that inci- 

 dental disturbing factors can be ehminated. Accordingly, a 

 series of experiments were made which involved the observation 

 of more than 8000 individuals and the isolation of over 2000 ani- 

 mals. The results of the entire series may be best interpreted 

 from the following brief table. In the first column (designated 

 'minus') the average division rate of the series was below that of 

 the controls and beyond the limits of error as indicated by the 

 differences between the two controls. In the second column 

 (designated 'neutral') the average division rate was the same as 

 the controls or within the limits of error as indicated by the differ- 

 ences between the two controls. In the third column (designated 



