SH 



NATURE 



\Feb. 2, I 



plus exact que I'analyse de I'eau, si Ton pouvait peser 

 rhydrogfene et peser I'eau qui proviendrait de sa com- 

 bustion. Mais I'exp^rience n'est pas possible sous cette 

 forme. Nous sommes obliges de peser I'eau formee, et 

 I'oxygene qui a servi a la produire, pour en ddduire, 

 par difference, le poids de I'hydrogene qui en fait 

 partie. Ainsi, une erreur de 1/900 sur le poids de 

 I'eau, ou de 1/800 sur le poids de I'oxygene, affecte 

 •d'une quantity dgale a 1/90 ou k 1/80 le poids de 

 I'hydrog&ne. Que ces erreurs dtant dans le meme 

 sens viennent k s'ajouter, et Ton aura des erreurs qui 

 iront k 1/40." 



Let us now turn our attention to the evidence afforded 

 by Regnault's determinations of the densities of 

 oxygen and hydrogen. Prof Le Conte has detected 

 some slight numerical errors in Regnault's reductions 

 {Phil. Mag. [4] 27-29), and when the necessary correc- 

 tions are made it follows that the density of oxygen 

 is 1-105612, and that of hydrogen o"o69269 ; whence, on 

 the assumption of Avogadro's law, we have the ratio 

 O : H = I5"96i 1 : i. This result is in such striking agree- 

 ment with Dumas' value that it is generally held to afford 

 the strongest corroboration of it. The number given for 

 oxygen is probably among the most accurate of Regnault's 

 determinations of gaseous densities ; the subsequent 

 results of Von Jolly, which are alone comparable in cha- 

 racter with those of Regnault, when reduced to the 

 geographical position of Regnault's laboratory have not 

 materially altered the value. The number given for 

 hydrogen is certainly not entitled to the same degree of 

 confidence. Indeed, it has been stated that Regnault 

 was himself of this opinion, on account of the great 

 difficulty of procuring hydrogen free from air. It is 

 hardly necessary to point out that even an extremely 

 minute admixture of air would tend to lower the relative 

 value of the atomic weight of oxygen. Moreover, the 

 hydrogen in the course of its preparation must have been 

 saturated with moisture ; and although, of course, all pre- 

 cautions at that time known were taken to dry the gas, it 

 is quite certain that it could not have been absolutely free 

 from traces of water. The experiments of Dixon have 

 shown how extremely difficult it is to dry a gas perfectly, 

 and it is now recognized that the ordinary methods of 

 desiccation still leave appreciable traces of moisture in it. 

 The effect of this moisture in the case of hydrogen would 

 be to increase its density, whereas in the case of the 

 oxygen it would tend to decrease it. On the other hand, 

 oxygen and hydrogen when measured under the standard 

 conditions of temperature and pressure are not, strictly 

 speaking, under exactly comparable conditions, and the 

 assumption of the validity of Avogadro's law is not 

 mathematically correct. 



Within recent years the question of the composition of 

 water has been again attacked, and with a fuller know- 

 ledge of the various sources of error which the progress 

 of science has shown to be present in the older methods. 

 Julius Thomsen found that i litre of dry hydrogen, mea- 

 sured under standard conditions of temperature and 

 pressure, when burnt with oxygen gave, as the mean of 

 eight concordant experiments, o'8o4i grammes of water. 

 Accordingly, 2 litres of hydrogen, on combining with 

 oxygen, would give 1-6082 grammes of water. Assuming 

 the validity of Gay-Lussac's law, and using Regnault's 



values for the weights of the gases at standard tempera- 

 ture and pressure, the calculated weight becomes — 



2 litres hydrogen ~ o'lygi grammes 

 I litre oxygen = 1-4298 ,, 



water = 1-6089 >> 



The difference is o"] milligramme. But the question 

 may be immediately asked, " Is Gay-Lussac's law actu- 

 ally valid?" The work of Regnault and Amagat on the 

 relation of volumes of gases to heat and pressure indi- 

 cates that, as ordinarily stated, it cannot be absolutely 

 valid. Dr. A. Scott has recently put the question to the 

 test of experiment, and, from a long series of trials in 

 which large volumes of gases were caused to combine, 

 he finds that the most probable ratio is i -994 : i (Proc. 

 Roy. Soc, 1887, 398). Taking Regnault's data as before, 

 we have — 



I -994 litres hydrogen = 0-1876 gramme 

 I litre oxygen = I '4298 ,, 



water = 1-6084 >> 



which differs only by 0-2 milligramme from Thomsen's 

 result. 



Now, from Regnault's densities of oxygen and hydro- 

 gen, as recalculated by Prof. Le Conte, it follows that 

 the weights of equal volumes of the gases are as 

 I : 15-961 1, which, on the basis of Dr. Scott's ratio for 

 the combining volumes, gives — 



O = 16-009. 



Prof. J. P. Cooke and Mr. T. W. Richards, of Harvard 

 College, have recently presented us with a further con- 

 tribution to the subject (Proc. Amer. Acad, of Arts and 

 Sciences, xxiii. 149), which merits very special attention, 

 not only on account of the intrinsic excellence of the experi- 

 mental work of which it is an account, but also because it is 

 here attempted to obviate certain of the sources of error 

 which have already been pointed out as inherent in 

 Dumas' method. The method adopted by the American 

 chemists was to pass a known weight of hydrogen over 

 heated copper oxide and to weigh the amount of water 

 formed. It will be seen that the essential feature in this 

 method is that the 'weight of the hydrogen is known 

 whilst that of the oxygen is obtained by difference ; in 

 contradistinction to the method of Dumas, where the 

 weight of the oxygen was known and that of the hydro- 

 gen found by difference. The preparation of this hydro- 

 gen and the determination of its weight were, however, 

 problems which required the highest manipulative skill. 

 Obviously, everything depends upon the purity of the 

 hydrogen. A glass globe of about 5 htres capacity and 

 weighing about 570 grammes was so provided with stop- 

 cocks that it could be evacutated by the air-pump. The 

 vacuous globe was weighed against a similar globe, in the 

 manner already adopted by Regnault, filled with hydro- 

 gen, and its weight again determined. The weight of 

 hydrogen taken was about 0-42 gramme. The hydrogen 

 was then driven over the heated copper oxide by a current 

 of dry air, and the water formed collected partly in a 

 weighed tube, and partly by means of sulphuric acid and 

 phosphoric oxide. The hydrogen was obtained by three 

 different methods : (i) by the action of sulphuric acid 



