No. 641] 



THE DURATION OF LIFE 



489 



before, to insure always having on hand newly emerging 

 flies with which to start experiments. 



When any experiment is to be started tlies are taken 

 from the stock bottles and etherized within 4 hours of 

 emerging (usually sooner), before the wings have un- 

 folded, so that they are surely virgin. Matings are made 

 up as desired, putting the mated flies in half-pint milk 

 bottles in the incubator. The parents are taken from 

 the mating bottles in 8 or 9 days (before any young begin 

 to emerge), and removed to a second mating bottle if a 

 larger sample of progeny is desired than can be obtained 

 from one bottle. As the offspring begin emerging they 

 are shaken out every morning from the mating bottle 

 to a small shell vial. Thus all the flies in a small bottle 

 are the same age, and are propcrl.x- labelled with mating 

 number and date of emergence. 'I'lieii e\ ei y morning all 

 these small bottles are looked <.vei' an<l those with dead 

 flies separated out. After all have l>eeii looked o\er. the 

 live flies in the bottles whieli have .lead <ui.- are >liaken 

 across to fresh bottles, the dead tlie> taken .uit aii.l >exed, 

 and all the pertinent facts as to dnrat'KMi oL' life, etc., re- 

 corded on printed blanks, from which the records are later 

 (when all the flies of an experiment havi> died off) coded, 

 punched on Hollerith cards, and sorted and tabulated by 

 Tabulating Machine C^ompany electric sorting and tabulat- 

 ing machines. Flies from any small bottles which are not 

 changed (because of dead flies) within five days of the last 

 previous transfer are transferred on the fifth day to 

 fresh food. The physical manipulation is too great with 

 the numbers we desire to use to admit of changing all the 

 bottles everv dav, which would be the ideal way to keep 

 food conditions' absolutely identical for all the flies. 

 Changing every five days keeps them approximately so. 



We desire to record our indebtedness to Mr. James 

 Kriickv. technical assistant in this work, for his pains- 

 takiuu" eare and fi<lelity to the lii-hest ideal> of exact 

 expenmental work. 



In the work discussed in this paper no attempt has 



