yy>ril s> 1877] 



NATURE 



499 



same gas, observed by means of the same apparatus, accorJing 

 to the amount of pressure, and the conformity in the various 

 series of experiments, prove that the results obtained do not 

 depend on any constant errors in the methods employed, but 

 that they are really caused by the nature and the essential 

 qualities of the gases investigated. 



5. The variations from Mariotte's Law under very weak 

 pressures being very small, it is necessary, in determining them, 

 to make the reading of the pressures, the volumes, and the tem- 

 peratures (absolute t = 273°) with a precision of two-thotsindths 

 of these total values ; thus, e.q., if /•^ = 0-200 m., /^ = O'loom., 

 and ?/„ = 2,500 gr., 7'^ = 5,000 gr. of mercury (^ - 20), it 

 will be necessary to determine the pressures with a precision of 

 o-oi mm., the volumes to o'l gr. of mercury, and the tempera- 

 tures to o'oi of a degree. 



The results will be doubtful if the precision is less.^ Thus it 

 is found that under a certain small pressure i^ases present positive 

 deviations from Mariotte's Law ; even gases like sulphurous acid 

 and carbonic acid, which under high pressures show consider- 

 able negative deviations. It is the same with air. M. Regnault 

 commenced his researches with pressures which exceeded that 

 of the atmosphere, and obtained negative deviations. 

 _ In 1S74 I effected with all the care possible the determina- 

 tion of the deviations for air under pressures of from 650 to 

 2,000 milimetres, and towards the end of 1875 and in the begin- 

 ning of 1876, in a special apparatus provided with compound 

 manomettrs, I repeated the same experiments with M. Bogusski 

 for pressures from 700 to 3,oco millimetres with air, hydrogen, 

 and carbonic acid. These researches proved the rigorous accu- 

 racy of M. Regnault's conclusions. Air and carbonic acid were 

 found to be subject under these pressures to negative deviations, 

 greater for carbonic acid than for air ; and hydrogen, for these 

 same pre>sures, was found to present positive deviations. At 

 present we are continuing the same kind of experiments for 

 pressures of more than three metres. 



Thus hydrogen, under all pressures, commencing with zero 

 and ending with a pressure infinitely great, presents throughout 

 positive deviations ; at no pressure does it follow Boyle's Law 

 rigorously, and it never presents negative deviations. Increased 

 pressures always give a greater volume than what might be 

 expected according to the variation of the pressures. Air under 

 pressures less than 600 millimetres also presents positive devi- 

 ations ; but under pressures greater than that of the atmosphere 

 its deviations become negative, and under pressures which exceed 

 100 atmospheres its compressibility again becomes positive. 

 Consequently for this gas there are two pressures at which it 

 follows Boyle's Law ; the one is very nearly that of the atmo- 

 sphere, the other lies between 30 and 100 atmospheres. These 

 pressures, under which the changes of the sign of compressibility 

 occur, will be different for carbonic acid ; viz., under pressures 

 less than that of the atmosphere the change of sign is found at 

 nearly 200 millimetres, and for higher pressures it commences 

 near that which corresponds to 70 metres of mercury, if we 

 base our researches on this point on the observations of Dr. 

 Andrews on the compressibility of carbonic acid gas for tempe- 

 ratures above 31°. For lower temperatures this point will probably 

 correspond to the passage of carbonic acid into the liquid state. 

 Consequently wiih a change of temperature the pressure at which 

 the change of sign of compressibility occurs, changes also. For 

 sulphurous acid the sign of compressibility under pressures lower 

 than that of the atmosphere changes at about forty millimetres of 

 pressure. But even this gas, so easily licjuefiable, under low 

 pressures, has always a positive compressibility. There is not 

 then, and there cannot be, a gas which is rigorously subject to 

 Mariotte's Law under small pressures. 



The idea of an absolute gas belongs, then, to the number of 

 fictions which find no confirmation in facts. We cannot, then, 

 suppose that with the decrease of density or with the increase of the 

 vis viva of gaseous molecules, gases approach a state in which 

 they follow Boyle's Law. Then (the density diminishing, the 

 velocity of the molecules increasing, that is to say, the pressure 

 diminishing, the temperature increasing, and the molecular weight 

 diminishing) they all tend towards another state characterised by 



the expression — LcU, > qj i,e.^ they are assimilated to solid 



dp 

 and liquid bodies, when the condensation reaches its limit. We 



' It is by these causes that the want of conformity in the experiments of 

 Siljestroni is sufficiently explained {^PosS- Aim., April and May, 1874 ; 

 see also the Bull, de V Acad, dc Sc. de SI. PHersbourg, t. xix , p. 466, and 

 Berkhte der deutschcn chcm. Gesell., t. viii.,p. 1,339; t- viii., pp. 576 and 

 749), and of M. Amagat (Comptes Rendus, April 17, 1876.) 



must believe that there is a limit of condensation and a lim't of 

 larefaction. If we take, in fact, a mass of non-volatile liquid, 

 and if we submit it to pressures infinitely great and infinitely 

 small, we shall see it change volume ; but in the two cases, we 

 shall have finite volumes, capable of measurement, and even 

 differing little for one and the same body. It is the same with 

 gases, if we admit that for pressures approaching zero, gases con- 

 tract according to the same law as that which we can deduce 

 from our compression experiments under pressures less than 

 that of the atmosphere, or as hydrogen contracts. Under great 

 pressures, or under pressures excessively small, every gas 

 resembles a solid or liquid body, and possesses two limits of 

 compressibility. The volumes which coi respond to these limits 

 are very different, but there is always reason for believing that 

 they exist. 



Without launching into hypotheses to explain these limit 

 volumes (such, e.g., as the supposition that molecules in them- 

 selves possess volume), I will confine myself to the question ot 

 the matter of celestial space. What is the luminous ether ? One 

 of two things— either an elastic independent matter, siii generis, 

 or the gas of the atmospheres of celestial bodies, considerably 

 rarefied. In the latter case it is necessary to admit the absence 

 of limits in the atmospheres and a condensation of the ether 

 greater and greater in proportion as we approach a celestial body 

 (sun or planet). There are many arguments for and against 

 both hypotheses. On the one hand, spectrum analysis leads us 

 to conclude that the material of all heavenly bodies is identical ; 

 on the other hand, it proves the diversity of the composition of 

 atmospheres. This is why we abstain from solving the ques- 

 tion in its essence. But spectrum analysis does not speak less 

 in favour of the former hypothesis, because it shows the diversity 

 of composition of our terrestrial atmosphere from that of many 

 of the other celestial bodies. In the researches on the resistance 

 of celestial matter to the movement of the planets, there 

 appears also to be a confirmation of the former of these two hy- 

 potheses, for neither planets nor comets show any diminution in 

 the excentricity of their orbits, which would be an inevitable 

 consequence of motion in a rarefied medium, as has been ob- 

 served in the case of Encke's comet. Exact investigations on the 

 movement of that comet, repeated in recent times by M. von Asten, 

 the PulUowa astronomer, show clearly the advances towards the 

 sun at perihelion, although in the beginning M. von Asten had n-jt 

 noticed them. But that comet at perdiclii.n was found only at 

 one-third ot the distance which separates the sun from the earth, 

 i.e., it was nearer to the sun than Mercury. It is possible 

 that it passed near to the limits of the solar atmosphere. 

 Faye's comet, as is known, does not present these same diversi- 

 ties, but its perihelic distance is about i"68, that of Encke's 

 comet being only about 0*33 ; it exceeds it then so much 

 that their comparison would only serve to confirm the hypo- 

 thesis of a solar atmosphere. If we admit a limit for the atmo- 

 spheres, we must expect in gases, for small pressures, exactly 

 that kind of variation from Boyle's Law which I observed in 

 rarefied gases. 



To prove that gases under very small pressures, as well as under 

 very considerable pressures, vary from the Boyle-Marriotte Law is 

 by no means the same as to deny the truth ot that law ; I feci that I 

 ought to state this most explicitly. For a long time the law of 

 gravitation could not be made to accord with the pertuibations ; 

 latterly these perturbations have proved the best c 'nfirmation of 

 the la^s of gravitation. In the present case it may be the same. 

 There are three laws for gases : that of Boyle and Marriotte, 

 pv = const. ; that of Gay Lussac, vt = z/q (i +a /) ; and thatot 



Ampere and Gerland — = const {a being the molecular 

 tn 



weight, and m the mass). Their ensemble is expressei for all 

 gases in general by the equation — 



apv = 84s (273 -h t) m, 

 where a is the atomic weight (// = I), p the pres-ure in kilo- 

 grams per square metre, v the volume in cubic metres, m the 

 weight in kilograms, t the centigrade temperature. This is, 

 however, only a first approximation. In the second member of the 

 equation there must be additional terms which express a function of 

 /and of a, very small for the ordinary mean values of />, and which 

 become of a sensible magnitude only when p is very small or very 

 great. To find this function is a question of the future, and 

 demands the labours of a great number of investigators. My 

 aim is to be able to furnish some experimental data which will 

 permit of judging of the form of that function. This work 

 requires many new processes, measuring apparatus of a high 



