286 



SCIENCE. 



[N. S. Vol. XI. No. 269. 



executed ; it is very desirable that he should 

 make a series of experiments, after adequate 

 study of sources of error and of means of 

 avoiding them. It is also desirable that, if 

 possible, the original process of complete 

 synthesis should be repeated with the little 

 modifications which time is sure to intro- 

 duce. These two would be enough, as far 

 as synthesis is concerned ; unless, indeed, 

 through the invention of another process 

 by a third experimenter, we could have 

 still more. Other syntheses of water than 

 by a complete synthesis seem less likely to 

 be of much service, except as a school of 

 experimentation . 



What further work is desirable on the 

 ratio of densities and of combining volumes 

 of hydrogen and oxygen ? Three constants 

 are involved : the density of oxygen, the 

 density of hydrogen, and the volumetric 

 ratio. 



The density of oxygen is known with a 

 probable error of about 1 part in 50,000. 

 It is very probable that no number what- 

 ever of further determinations would change 

 this value by 1 part in 10,000. JSTo further 

 work upon this density seems at present 

 desirable, except that w^hoever determines 

 the density of hydrogen cannot well fail to 

 determine that of oxygen also. 



The densitjr of hj^drogen demands further 

 experiment. It is possible to make, by 

 some one of three or four slightly different 

 processes, a series of experiments whose 

 average variation shall be less than 1 part 

 in 3000 or 5000, or even 10,000 ; but dif- 

 ferent series do not agree sufficiently with 

 each other. We are far from knomng the 

 density of hydrogen so well that more ob- 

 servations might not change our value by 1 

 part in 2000 or 3000. It is very desirable 

 that further observations should be under- 

 taken by at least two different methods. 

 In one method, hydrogen should be weighed 

 while absorbed in palladium, should be then 

 transferred to a measuring apparatus with- 



out the use of stop-cocks, and should be 

 there measured. This process should be re- 

 peated with measuring apparatus of varied 

 volumes. In another method, hydrogen 

 should be weighed after Regnault's method, 

 in a counterpoised globe, but with such pre- 

 cautions that leakage through a stop-cock, 

 and contamination with vapor of mercury, 

 should be excluded. The globe should be 

 exhausted till the remaining air is a small 

 fraction of a millionth, should be sealed off 

 from the pump, and should be connected 

 mth a condenser at the temjoerature of 

 liquid air, so as to remove mercurial vapor. 

 After this hydrogen is to be admitted with- 

 out the use of stop-cocks. The manipula- 

 tion is not difficult, and the method would 

 confirm the results of the previous method. 

 The ratio of the combining volumes 

 of hydrogen and oxygen is not known with 

 the degree of confidence which is desirable. 

 The history of the matter is not an uninter- 

 esting one. Further continuance of the two 

 series of experiments on which the present 

 value depends would be most unlikely to 

 change it by 1 part in 10,000, for its prob- 

 able error is 1 part in 40,000. But one of 

 the experimenters has obtained results dif- 

 fering from that finally adopted by as much 

 as 1 part in 220. The other experimenter 

 has entirely discarded the result of one 

 series and replaced it, not by a better series 

 of the same kind but by one of a quite dif- 

 ferent nature, not carried to its pi'oper com- 

 pletion, and accordingly reduced by the use 

 of the constants of van der Waals' equation. 

 It is desirable that experiments be made to 

 furnish means for a new reduction bj^ meas- 

 uring the change of volume when 2 volumes 

 of hydrogen and 1 volume of oxygen are 

 mixed, being at the same pressure before 

 and after mixing. This experiment has 

 lately been made by Berthelot, whether with 

 sufficient precision for the purpose is not 

 known at this moment. It is also desirable 

 that the ratio of the combining volumes of 



