308 



Prof. A. W. Eiicker on the Density and 



cent, and 79*5 per cent, are —290 and —210 respectively, at 

 80'27 per cent. 



(p-\ =-210-80x0-77= -272. 



\CLp J m 



,\ Since at 80 per cent./J^ = 11230, 



at 80-54 per cent. (^)= 11230-272 x 0-54 

 = 11230-147 = 11083. 

 Table III. 



Po- 



\dp/o 



Vm- 



U 2 L' 



p. 



ds 

 dp' 



80 



11230 



80-27 



-272 



80-54 



11083 





J5 



8002 



-252 



80-04 



11220 



79 



11440 



79-24 



-191 



79-48 



11349 







79-06 



-179 



79-12 



11419 



78 



11580 



78-31 



-125 



78-62 



11503 



70 



11645 



70-53 



+ 55 



71-06 



11703 



68 



11523 



68-55 



63 



69-10 



115V>2 



66 



11380 



66-56 



74 



67-12 



11463 



64 



11220 



64-75 



82 



65-51 



11343 



>> ' 





64-56 



83 



65-12 



11313 



t [ 



„ 



64-25 



85 



64-50 



11262 



62 



11033 



62-54 



96 



63-08 



13137 



60 



10830 



60-50 



103 



61-01 



10934 



58 



10602 



58-47 



102 



58-94 



10698 



57 



10497 



57-47 



105 



5794 



10596 



56 



10395 



56-44 



102 



56-89 



10488 



54 



10177 



54-44 



108 



54-89 



10273 



52 



9955 



52-45 



110 



52-91 



10055 



48 



9503 



48-45 



109 



48-91 



96U2 



46 



9293 



46-47 



103 



46-94 



9390 



It remains to compare the two sets of values given by my 

 single curve and Mr. Pickering's five curves respectively with 

 his experimental numbers. 



In doing so we must remember what the estimated experi- 

 mental error is. Mr. Pickering regards the possible error in 

 the determination of a density as =0*000008 = e say. Hence 

 the possible error of dsfdp is 2e/8, where 8 is the difference 

 between the percentages of the two points from which it is 

 determined. In general 8 = 2, so that the error of ds/dp is 

 -j-0'000008. The experiments at 18° (which are alone treated 

 of in this paper) are supposed to be less accurate than the 

 others, but Mr. Pickering does not think that the error of expe- 

 riment reaches twice the above value*. He admits that the 



* It must, however, be remembered that the 18° curve was chosen 

 as the abnormal errors appear to be fewer and less than in other cases. 



