NA TURE 



OM 



THURSDAY, FEBRUARY 2, i! 



THE COMPOSITION OF WATER. 



DUMAS, in his well-known memoir on the gravi- 

 metric composition of water, which every student 

 is taught, and rightly so, to regard as one of the classics 

 of chemistry, states that of all analyses presented to a 

 chemist that of water is the one which ofifers the greatest 

 uncertainty. Critics of a certain type may possibly take 

 exception to the literal accuracy of this remark. No one, 

 however, will gainsay the statement that, in view of the 

 momentous issues which depend upon our knowledge of 

 the composition of water, this knowledge is not by any 

 means so exact as the state of contemporary science 

 demands. It is, of course, not merely the question of the 

 quantitative composition of water, but the far more im- 

 portant matter of the relative values of the atomic weights 

 of hydrogen and oxygen. Of all stoichiometrical con- 

 stants required by chemists nowadays, those of hydrogen 

 and oxygen are infinitely the most important. Every 

 chemist knows what is dependent on these ratios, and he 

 knows too that the difficulties which their direct determi- 

 nation involves are well-nigh insuperable. 



All the discussions within recent years on the vahdity 

 of Prout's law have tended to show that so far as experi- 

 mental work is concerned, the question may now be said 

 to hang upon these particular values. It is not too much 

 to say that, if any chemist could succeed in showing by 

 irrefragable experimental evidence that the atomic weight 

 of oxygen was exactly sixteen times that of hydrogen, 

 he would in the present state of scientific opinion at 

 once succeed in inducing his brethren to accept Prout's 

 law and all its far-reaching consequences as articles of 

 their chemical creed. 



It may be worth while to examine very briefly the 

 nature of the ground upon which the present accepted 

 values for the relative atomic weights of hydrogen and 

 oxygen are based. It will be generally conceded that the 

 evidence upon which chemists have almost exclusively 

 relied is that afforded by Dumas' gravimetric analysis of 

 water, and by Regnault's determination of the specific 

 gravities of oxygen and hydrogen. 



Dumas' work was published in 1843. His method 

 was identical in principle with that employed by Dulong 

 and Berzelius for the same purpose, and consisted, as is 

 well known, in heating copper oxide with an unknown 

 weight of hydrogen, and determining (i) the loss of weight 

 suffered by the oxide, and (2) the weight of the water 

 formed. The decrease in weight of the copper oxide was 

 assumed to represent the weight of the oxygen evolved, 

 and the difference between this weight and that of the 

 water formed was held to be the amount of hydrogen 

 which had combined with that of the oxygen. 



In all, nineteen experiments were completed, in which 

 quantities of water varying from about fifteen to eighty-six 

 grammes were formed. Treating the results in the manner 

 adopted by Meyer and Seubert — that is, in accordance 

 with the equation 



X = ^1 + ^a + ^3+ ■ • ■ + ^« = [^^ 

 . + a„ {«] 



Vol. XXXVII — No. 953. 



4^ 



in which a = weight of oxygen used, and i = weight of 

 water formed— it follows that [a] = 840-16 grammes, and 

 [i>] = 945 '44 grammes, whence the ratios H : O = i ; I5"96. 

 When, however, we come to examine more nearly the 

 details of the method of determination, we find that these 

 ratios are certainly affected by errors of which the magni- 

 tude cannot be even approximately known. In the first 

 place, the sulphuric acid solution employed to generate 

 the hydrogen must have contained dissolved air, the 

 effect of which would be to lower the ratio of the oxygen. 

 This fact was not indeed unnoticed by Dumas, but its 

 effect could not be estimated with any certainty. More- 

 over, it seems almost impossible to prepare hydrogen 

 from zinc and sulphuric acid without the formation of 

 more or less sulphur dioxide, the last traces of which can 

 only be removed by prolonged exposure to potash solu- 

 tion. Copper is one of the few metals that have the 

 power of forming a hydride, and although this hydride, 

 like the palladium hydride, is more or less readily 

 decomposed by heat, the affinity of hydrogen for copper 

 may be still traceable even at moderately high tempera- 

 tures. Melsens, working in Dumas' laboratory, found 

 that the reduced copper did actually retain hydrogen, and 

 in amount varying with the temperature to which it had 

 been heated. The weight of the condensed water must 

 have been increased, as Berzelius pointed out, by the air 

 which it eventually dissolved. Now the effect of all these 

 errors would be to lower the value for the atomic weight 

 of oxygen. Of course there may have been errors working 

 in the opposite direction of which we know nothing, but 

 it is reasonably certain that the net result of the constant 

 errors, so far as these can be ascertained, is to give too 

 small a value for oxygen. Above all, there are the for- 

 tuitous errors, such as those caused by differences in the 

 power of surface-condensation of the vessels employed ; 

 errors of weighing and of reduction to a standard atmo- 

 sphere, &c. ; which, although theoretically allowed for 

 and eliminated by a sufficiently frequent repetition of the 

 experiments, may, on the whole, tend to operate in a given 

 direction. Lastly, there is a source of error of the same 

 order in a circumstance which, as there is a certain 

 touch of pathos in them, may be stated in Dumas' own 

 words : — 



" II faut meme ajouter que la durde ndcessaire de ces 

 operations, en m'obligeant k prolonger le travail fort avant 

 dans la nuit, en pla^ant les pesdes vers deux ou trois 

 heures du matin dans la plupart des cas, constitue une 

 cause d'erreur rdelle. Je n'oserais pas assurer que de 

 telles pesdes mdritent autant de confiance que si elles 

 avaient 6t6 exdcutees dans des circonstances plus favor- 

 ables et par une observateur moins accabld de la fatigue 

 inevitable apres quinze ou vingt heures d'attention 

 soutenue." 



There is, above and beyond all, a fundamental flaw 

 in the principle of the method, of which Dumas him- 

 self was fully conscious. After having declared that of 

 all analyses presented to a chemist water is the one 

 which offers the greatest uncertainty, he goes on to state 

 to what this uncertainty is due : — 



" En effet, i partie [d'hydrogene se combine avec 8 

 parties d'oxygene pour former de I'eau, et rien ne serait 



P 



