182 



TEE AMERICAN NATURALIST [Vol. LVI 



able error, nor is there any certainly significant change in 

 variability having regard to the probable errors of the 

 differences involved. 



TABLE VII 



IxBRED Tests of the Several Lines 



104-100 — .86 ± 1.12 ^ ^77 " 



107-101 + 3.72 ± 1.37 2.72 



207- 201 — 2.06 ± 1.48 1.39 



208- 202 + 3.61 ± 2.19 1.65 



304-300 + .90 ± 1.65 .54 



309-303 + .98 ± 1.40 .70 



Eesults of Mass Cultuee Ee-tests for Constancy 



The point may well be made that in the re-tests of the 

 lines described in the preceding section an additional ele- 

 ment is introduced in the fact that the flies for the re-test 

 were the progeny of a second brother X sister mating. 

 Wliat one wishes to know is: what degree of constancy 

 in duration of life is exhibited by the general stocks in 

 each line, mating purely at random, after the initial se- 

 lection and inbreeding? "VVe wish now to present some 

 data on this point. Table VIII gives the biometric con- 

 stants for this material. Mass culture re-tests have been 

 made on two of the original lines, 100 and 101. These 

 mass culture re-tests were made in two ways as follows : 



{a) From the stock bottles of the line to be tested a 

 large sample of progeny was taken at random each day 

 as the flies emerged from the pupal stage, and these 

 progeny flies were put in small bottles for a duration of 

 life experiment in the usual way described in (27). 



(6) From the stock bottles of a jjarticular line to be 

 tested a number of virgin flies (usually 8 to 10 of each 

 sex) were taken at random immediately upon emergence, 

 and mated as a group in a mating bottle. The progeny 

 from this sample was then removed, upon emergence, to 

 small bottles and a regular duration of life test carried 

 as described in (27). 



