266 



partition ; this is necessary during the exhaustion of the tjlobe, 

 and tlie re-admission of the gas, for tliese operations prodiue such 

 great oscillations in the barometer as to introduce small (luantities 

 of air into the instrument, and thus vitiate the vacuum. The ex- 

 liausted globe is weighed witli the precautions that have been 

 indicated. If P represent the weight of the gas when the barome- 

 ter stands at H, and /; that weight when the elastic force in the 

 globe corresponds to a barometric height //, the weight of the gas 



at 0', and under the normal pressure of 760 mil., is (V—p)-^. — r 



To obtain a new weighing of the same gas, the exhausted glolje, 

 enveloped in ice, is placed in connection witli the apparatus for 

 generating the gas, and the series of operations which liave been 

 ixiiiited out, repeated. The gas thus becomes purer at each opera- 

 tion. M. Ilegnault found that it is only from the fourtli tilling 

 that the gas presents rigorously the same weight. It is desirable 

 to satisfy ones self whether the gas upon wliicli we are operating 

 ioUows the law of Mariotte, at pressures below those of tlie atmo- 

 sphere : this verification is absolutely necessary if the density of 

 the gas is to serve for the determination of atomic weights. For 

 the law of the volumes of gases, and the simple ratios which exist 

 between their densities and atomic weights, exist i-igorously only 

 (it the limit — that is, in a state of extreme dilatation ; we must 

 therefore see whetlier tlie anomoly in these laws does not commence 

 already near the atmospheric pressure. 



This is done by measuring the weight of the gas, with great care, 

 at different degrees of elastic force, as marked by the comparison 

 of the manometer and barometer. 



Finally, by this means we may determine the weight of the gas 

 which fills tlie globe at the temperature of 100° and under atmo- 

 spheric pressure, and thus determine the density of the gas when 

 compared with air at 100°. This new density must be exactly the 

 same as that calculated for 0°, in order that it may serve in the 

 calculation of the atomic weights ; for it is necessary for this pur- 

 pose that the gas should have the same co-eflicient of dilatation as 

 tlie atmospheric air ; at all events the weight of the gas which fills 

 the vessel at 100° compared with that which fills it at 0°, permits 

 us to calculate the co-efficient of the dilatation of the gas ^^ 



Again, in order to determine whether the gas follows the law of 

 Mariotte, at the temperature of 100°, we have only to repeat the 

 former experiments, filling the globe at this temperature, instead 

 of at 0°. 



M. Regnault then recapitulates the advantages of this method, 

 which are, — that it gives the density of the gases with more preci- 

 sion, and far less trouble, than the methods formerly used ; it gives 

 these_ densities at identical temperatures at 0° and'l00°, that is, at 

 the fixed points of the thermometer, and consequently gives 

 immediately the co-efficient of dilatation of the gas ; and, finally, 

 it permits us to determine with great exactness, whether the gas 

 follows the law of Mariotte, at the temperatures of melting ice, 

 and boiling water. 



He then proceeds to give the detail of all his experiments, with- 

 out a single exception, in order to allow the reader to judge of the 

 degree of precision obtained by this method. It is not necessary 

 that we should give these details — or those of the processes by 

 w hich M. Regnault purified his gases ; they were such as might be 

 expected from one so familiar with all the minutiae of physical 

 science. 



He first determined by nine experiments the weight of pure 

 atmospheric air, freed from carbonic acid and watery vapour, which 

 tilled his globe at the temperature of 0°, and under'the barometric 

 pressure of 760 millim. (29-941 inches). The mean of these experi- 

 ments was 12-7781 gr. The minimum, 12-7714. The niaximuni, 

 12-7809. The ditference, 0-006.5 or ,^,'7^-, very nearly ^.„'„„ of the 

 mean; and he remarks that it is probable that a great p'art of this 

 error is due to the variations which occur in the composition of 

 the atmosphere. He regards it as unfortunate that men of science 

 should have selected the atmospheric air, whose constitution is 

 known to vary, as the standard of densities for gases, in place of 

 some gas which could always be obtained perfectly pure, such for 

 instance as oxygen, which would be the more convenient since this 

 gas is already chosen as the basis of the tables (adopted by con- 

 tinental chemists) of chemical ecpiivalents. 



Founded upon this determination of the weight of a given 

 volume of air, he proceeds to determine the densities of different 

 gases, and his results are as follows : — 



* This Is the only direct method which can be used for this determiDatioD, in gases 

 »n:ch attack mercury. 



Weight of -Iqo / and with an 

 the gas at J 'I elastic force , 



Densities determined by Dumas and Boussingault. 



Nitrogen, from .. 970 to 974 



Hydrogen, „ .. IJG91 ,, 0'06U5 



0.\yge:i, „ .. 11055 „ 1-0158 



If we calculate the theoretic density of carbonic acid gas, ad- 

 mitting for the atomic weight of carbon 7j, (oxygen = 100, or 6 if 

 hydrogen = 1,) lately found by M. Dumas, we get the number 

 1-52024, which approaches the density found for this gas under the 

 pressure of 224-17 millim. (less than nine inches.) 



The density found at the temperature 0° and normal atmospheric 

 pressure leads to an atomic weight for carbon 76-6 which ap- 

 proaches very nearly the number 76-44 (6-1152, if hydrogen ;= 1) 

 which chemists for a long time admitted from the experiments of 

 M. Berzelius.'"' 



We see, by this example, how much circumspection is necessary 

 in deducing the value of the atomic weight of a gas from its 

 density. 



Three expei-iments to determine the co-efficient of dilatation of 

 the air between 0° and 100°, gave as the result 0-03663, which 

 differs but little from the value obtained in the First iSIemoir. 



An attempt to verify the law of Mariotte showed slight differ- 

 ences, in which the weights by experiments were always a little 

 lower than those got by calculating the density by means of 

 Mariotte's law from the observed elastic force ; but these differences 

 were always within the limits of the errors of observation. 



The co-eflicient of dilatation of carbonic acid gas between 0° and 

 100° was determined to be 0-003719. (In the First Memoir, the 

 determination by the method V., in which the gas preserved the 

 same elastic force at 0° and 100°, as in the present case, was 

 0-0037099.) 



The experiments to determine whether carbonic acid gas obeys 

 the law of Mariotte at pressures less than that of the atmosphere, 

 gave the following results : — 



T,„ „„, , Calculaled by 



By experiment. ji^^i„„^,, ^.J,_ 



} 224 17 mil., 5-7:i45 gr., 5-7(;34 gr. 



.■t74-13 „ 9'5S45 „ il-0628 „ 



„ 10101° „ u;i8-39 „ 0-3049 „ (1-3545 „ 



SO that it appears that carbonic acid gas deviates notably from the 

 law of Mariotte at ordinary temperatures, but conforms to it with 

 the limits of experimental errors at 100°. 



Third MeiHOir. — DETERsnNATiON op the -weight of the LimEt 



OP AIR AND OP THE DENSITY OP MERCURY. 



In the preceding memoir, the densities of the different gases 

 were determined, referring them to that of air assumed as the unit ; 

 but in a great number of circumstances, it is required to know the 

 absolute weight of these gases : this is easily obtained when we 

 know the absolute weight of air under the normal conditions, that 

 is at a temperature of 0°, and under a pressure of 760 millimetres 

 of mercury. 



The weight of the litre of dry air under the normal conditions 

 was determined l)y MM. Biot and Arago, with all tlie care which 

 they could take — they found that at Paris this weight was 1-299541 

 gr. {Memoirs of the Academy <f Sciences fur 1808. liiot Traite de 

 Physique, torn. 1, ;;. 387.J This number has been generally adopted. 



But if we reflect upon the imperfections which the theory of 

 gases and vapours still presented at that time, and the great num- 

 ber of uncertain corrections which they were obliged to introduce 

 into their calculations ; and if we note that they operated upon air 

 charged with aqueous vapour, for which they endeavoured to allow 

 by a correction ; and that, in spite of the most minute precautions, 

 this circumstance must necessarily introduce great disturbances 

 into their experiments, we shall understand how absolutely neces- 

 sary it was to make new determinations of this important doctrine, 

 which will be frequently used in the following investigations. 



In the preceding memoir, the weight of dry air which filled the 

 globe at 0°, and under a pressure of 760 mil., was determined with 

 great care ; it will be enough then to find the capacity of this globe 



• Berzelius himself now aclinowledges the error of his former determination, and fixes 

 the atomic weight of carbon at 76-12 or (> 01 (H = 1), which amounts to an admission of 

 Uumas'a delerminntion. (Berzelius Traits de Chimie, Seconde Edition Fianvaise, Paris, 

 1845, torn. 1, p. 2(!;i.) 



-t The French litre ie equal to 0-22 of the imperial gallon. 



