TABLE 40 (conlinu(d) 



94 PROBABLE VALUES OF THE GENERAL PHYSICAL CONSTANTS 



Using Houston's value of R H , and of Rnr-Rn, together with the values 

 of H He, etc., we obtain e/m= (1.7608 ±0.0008) X 10' abs. em units. This 

 value' of e/m thus agrees with that obtained by Babcock. Summarizing the 

 results we find e/m= 1.769*0.002 from deflection experiments = 1 761 

 ± ooo2 from Zeeman effect, =1.761-0.001 from H and He spectra The 

 discrepancy between the first result and the last two is four times the probable 

 error of the first. Theory gives only one chance in 143 of this occurring. The 

 discrepancy seems to be real. 



The last two results are measurements of e/m for electrons tnsideof an atom, 

 based upon the quantum theory of atomic structure. The first is the measure- 

 ment o e/m for electrons in free space. The figures point to the conclusion 

 that the e/m of an electron is less when it is inside an atom than when it is 

 outside. If this conclusion seems unacceptable, then it would appear that there 

 is some general error in the equations of the quantum theory of atomic struc- 

 ture or there is some unknown general error in all the deflection experiments 

 Under the circumstances two values may be assumed of e/m-^ne for where 

 atomic structure is involved, the other for free electrons. Hence 



e/m (spectroscopic) = (1.761±0.001)xl0 7 abs. em units per g, 

 = (5.279 ± 0.003) X 10 17 abs. es 



e/m (free electrons) = (1.769 ± 0.002) X 10' abs. em 

 = (5.303 ±0.006) X 10 17 abs. es 



Tetlus ;Ze « of S income. A satisfy de t e rmi na t io„ 

 of this constant is difficult. 



The first attempt to obtain a value of h, from the results of all seven meth 

 ods was made by Doctor Birge in 1919. The value found was (6.5543 

 I00L0 X to " erg ■ sec, the error being merely the least-squares probable 

 error terror has been criticized by Ladenburg as far too small. It t 

 rtbeLd error since, as clearly stated, one must add to .t - error somewh 

 greater than the proportional error in . This occurs with some po^ tve powe 

 . nnitv to twol in every known method for obtammg h. This makes me tot 

 (unity to two; me y Doctor Biree's IQIQ evaluation of h has 



probable error more nearly ±0.01. Doctor Birge s y y 

 been adopted by the I.C.T., but the probable error should be ±0.001. 



In Jo Ladenburg* wrote an article on the evaluation of h, in which 

 sevlrof Doctor ^^^ — ^own .suit m that 

 2&S ^i^J^**** which value he rounds 



1 Jahrb. Radioakt. und Electronik, 17, 93- 1920. 2 H.P., 23, 279 



Smithsonian Tables 



