Table 40 (continued) no 



PROBABLE VALUES OF THE GENERAL PHYSICAL CONSTANTS 



It appears that Coblentz' estimated error for his own work (5.722 ±0.012) 

 is too small, but that his final average of the work of all investigators up to 

 1922 (5.72 to 5.73) should be more trustworthy than any single value. We 

 will choose 5.725 and 0.02 for its probable error. This result is then to be 

 averaged with the more recent work whence 



<x = (5.735±o.oii) X I O -5 erg • cm -2 • deg." 4 • sec. -1 

 a = 4<r/c= (7.652 ±0.015) x io" 15 erg • cnr 3 • deg. -4 



There has appeared a further determination of this quantity, by Hoare. 1 He 

 used a Callendar radio balance ; the advantage of the method is that both source 

 and receiver are essentially " black-bodies." Hoare obtains a— 5.735, agreeing 

 exactly with the value adopted. He lists 38 separate results, average deviation 

 only 0.016. The inclusion of this new result leaves the average value un- 

 changed, and Doctor Birge leaves the probable error unchanged. Objection 

 might be made to this adopted error as too small ; such an objection can hardly 

 hold in the face of Hoare's work. This new work also speaks against Strum's 

 assumption of an inadequacy of Planck's formula. We have then 



h~ (6.539 ±0.010) x IO "" 27 er g " sec - 

 (g) Summary. — We have now six determinations of h: 



Doctor Birge adopts 



h = (6.547 ± 0.008) x IO 27 erg • sec. 



This value of h is identical with Ladenburg's most recent estimate. 2 This 

 identity is spurious, since Ladenburg assumes £ = 4.774 X io~ 10 . If this older 

 value of e had been used in the present work, we should have obtained 

 /; = 6.5535, in practically exact agreement with Doctor Birge's 1919 value 



(6-5543)- 



Another potentially accurate method is given by the Compton shift of X-ray 

 lines. The theoretical equation for this is AA= (h/mc) (1 — cos <f>), where m 

 is the mass of an electron, as deduced from the values of e and e/m. Since h 

 varies in value with e, this equation can better be used to evaluate e/m. We 

 can in fact write AA= (h/e) (e/m) ( 1 —cos cf>) in which e as usual is in es units, 

 and e/m in em units. Then 



e/m- (A\)/(h/e)(i-cos<f>) 

 The most accurate work on this subject has been done by Sharp, 3 who obtains 

 AA= (0.04825 ±0.00017) x io" 8 cm, for (1 —cos <£) = (1.984 ±0.001 ). With 

 the adopted values of h and e, we have h/e— (1.3725 ±0.0005 ) x I0 ~ 17 



1 Philos. Mag., 6, 828, 1928. 2 H.P., 23, 279. 3 Phys. Rev., 26, 691, 1925. 

 Smithsonian Tables 



