BIPARENTAL INHERITANCE IN PARAMECIUM 443 



EXPERIMENT 17: 239 SPLIT PAIRS PROPAGATED SEPARATELY FOR 



TWENTY DAYS 



The only possible explanation for the positive correlation 

 between the members of pairs other than that it is the result of 

 conjugation, is that based on assortative mating, as already set 

 forth on page 432. The two conjugants which united in a single 

 pair are known to be similar in size before union ; they might also 

 be similar in other respects, as in fission rate. The coefficient 

 of correlation between the two members might then exist before 

 conjugation, not being due to the latter. 



To properly test this matter, it became necessary to therefore 

 carry on an extensive experiment with split pairs, comparable to 

 Experiment 16 with pairs. It would be desirable if the pairs and 

 split pairs could come from the same culture and be propagated 

 side by side. With the large numbers we wished to employ, 

 this was however impraticable. 



We therefore collected material from the same ditch from which 

 came the pairs of Experiment 16; put this under the conditions 

 for producing conjugation, and in this way obtained on May 

 9, 1912, 239 split pairs (pairs of which the individuals were separ- 

 ated before conjugation was consummated). These were propa- 

 gated in the same manner as were the conjugants of Experiment 

 16, so that we may refer to page 434 for an account of the methods 

 employed. 



Unfortunately the mistake was made in this case of collecting 

 the material from near the mouth of a sewer, where the animals 

 were living in a dense mass, in company with bacteria of putre- 

 faction. Paramecia which come from such conditions do not 

 live well under the conditions necessaiy for slide cultures, so that 

 the mortality was high. Both members of 190 split pairs however 

 lived through the first ten-da}^ period, and both members of 155 

 split pairs through a period of twenty days. This gives ample 

 material for making the test we desire. 



The data obtained, so far as the}- bear upon correlation in 

 fission rate between the two members of split pairs, are gi\'en in 

 tables 46, 47 and 48. It may be desirable in a later paper on 



