No. 641] 



THE DURATION OF LIFE 



505 



on as steep aslant as possibJe, so that any iii()i>tui c Toi-med 

 would drain down and be absorbed by \\\v \)i\\)vv in the 

 bottom of the bottle. Flies of ration ') from four 

 different lines were used in the cxpci inicnt, two normal 

 wild type lines, one of Old l-'alinonih stock and one of 

 New Falmouth stock, and two lines of (L)uintuple stock. 

 Flies were taken out from stock bottles of these lines as 

 they emerged every day, beginning March 18, 1921, and 

 continuing through April 4, putting hatches of alternate 

 days in the specially ])re])are(l bottles and the other 

 hatches in ordinai x boi t lo. TIk- flies put into the specially 

 prepared bottles \vi ic of course kept throughout their 

 lives in such bottles, and the controls in ordinary l)ottles. 

 Table VI shows the L lines of the four uronps of Hies — 

 long- winged with paper and slant food, lonu-winucd with- 

 out paper and food surface horizontal. -Iiort-w inuvd with 

 paper and slant food, short-winge<l without pap..]- and 

 food surface horizontal. l)i^ti'lbution> liaNV Ix-cn made 

 foi- tho four lines separately. an<l for the s.'Xes >eparately. 

 l)Ut since they all show the >aine results the separate 

 .li>tribution^ an' not uivn. 



The .hata .>f Tabh' W ai'e >hown -Tapiiicallv in Kig. G. 



It i. evident tluit there is n.) delinite or marked .liffer- 

 ence between the slant food group and the oihei-. Such 

 differences as do appear between the /. line> in the two 

 cases are only of the order of magnitu«le which might 

 readily appear in random sampling. This is indicated in 

 another way by the data of Table 



In the case of the short-winged flies the difference in 

 the mean is plainly not significant. In the case of the 

 long-winged flies the difference is 2.96 times its pi'obable 

 error. One wwld expect a difference as gi'eat as this or 

 greater to occur from chance ah»ne only 4 to times in 

 every 100 trials, so that the ditference i> heiv -ctting on 

 towa"rds the ma-nitude where it nmst be ivuar.lcl a> cei- 

 tainlv M-nitieant on purely >tati>ti.-al gr<.un.l<. Ibit tin 

 difference is in favor of the horizontal food without 

 drainage, and (ni,rnisi the food with drainage. 



