28 



SMITHSONIAN MISCEIiLANEOUS COLLECTIONS 



VOL. 54 



Fifth Series. Essentially like series four, except that the two gases 

 were prepared by electrolysis of bari^^m hydroxide. 



Since these series of measurements represent different methods, and 

 are evidently of unequal value, it is best to combine them mathematically, 

 giving each mean a weight inversely proportional to the square of its 

 probable error. JSToyes computed all of the experiments on the basis of 

 the oxygen standard, giving for each one its expression as the atomic 

 weight of hydrogen. I have chosen the present form for simplicity of 

 calculation, and for greater ease of combination with previous deter- 

 minations. In the following table I give ISToyes' deductions in an addi- 

 tional column ^ : 



Ratio H:0. Atomic weight H. 



Series 2. H:0 15.8750, ± .00052 1.00787 



H : H„0 15.8747. ± .00045 1.00789 



Series 3. H : O 15.8782, ± .00091 1.00767 



H:H,0 15.8772, ± .00097 1.00774 



Series 4. H : O 15.8716, ±: .00066 1.00809 



H : H„0 15.8705, ±: .00074 1.00815 



Series 5. H : O 15.8751, ± .00046 1.00786 



H:H,0 15.8749, ± .00051 1.00788 



General mean.. 15.8745, ± .00021 



1.00783, ± .000013 



' See Noyes' memoir for the results of combining his data in ilifferent ways. The 48 experi- 

 ments, taken as one series, give H = 1.00793, + .00002. 



Referring to his determinations published in 1890, Noyes points out a constant error in them, 

 the elimination of which reduces the value of the ratio to 15.879, in agreement with the measure- 

 ments by Morley. 



