June 23, 1898] 



NATURE 



= 83 



That this gas is a new one is sufficiently proved, not merely 

 by the novelty of its spectrum and by its low density, but also 

 by its behaviour in a vacuum-tube. Unlike helium, argon, and 

 krjpton, it is rapidly absorbed by the red-hot aluminium 

 electrodes of a vacuum-tube, and the appearance of the tube 

 changes, as pressure falls, from carmine red to a most brilliant 

 orange, which is seen in no other gas. 



We now come to the gas obtained by the volatilisation of 

 the white solid which remained after the liquid argon had boiled 

 away. 



When introduced into a vacuum-tube it showed a very com- 

 plex spectrum, totally differing from that of argon, while re- 

 sembling it in general character. With low dispersion it 

 appeared to be a banded spectrum, but with a grating, single 

 bright lines appear, about equidistant through the spectrum, the 

 intermediate space being filled with many dim, yet well-defined 

 lines. Mr. Baly has measured the bright lines, with the follow- 

 ing results. The nearest argon lines, ^as measured by Sir William 

 Crookes, are placed in brackets : — 



Reds very feeble, not 



measured. 

 First green band, first 



bright line 5632-5 (S651 = S619) 



First green band, second 



bright line 5583'° (5619 = 5567) 



First green band, third 



bright line 5537-0 (5557 = 5320) 



Second green band, first 



bright line Sl63'0 (5165) 



Second green band, 



second bright line ... 5126-5 (5165 : 5065) brilliant. 

 First blue band, first 



bright line 4733 '5 (4S79) 



First blue band, second 



bright line 47ii'5 (4701) 



Second blue band, first 



bright line 4604*5 (4629 : 4594) 



Third blue band (first 



order) 4314° (4333 = 430o) 



Fourth blue band (second 



order) 42i3'5 (4251 : 4201) 



Fifth blue band (first 



order), about 3878 (3904 : 3835) 



The red pairof argon lines was faintly visible in the spectrum. 



The density of this gas was determined with the following 

 results :— A globe of 32-35 c.c. capacity, filled at a pressure of 

 765-0 mm., and at the temperature 17-43°, weighed 0-05442 

 grams. The density is therefore 19-87. A second determination, 

 made after sparking, gave no different result. This density 

 does not sensibly differ from that of argon. 



Thinking that the gas might possibly prove to be diatomic, 

 we proceeded to determine the ratio of specific heats : — 



Wave-length of sound in air 

 . ., . .. gas 



Ratio for air 



gas 



34-18 

 31-68 

 1-408 

 1-660 



The gas is therefore monatomic. 



Inasmuch as this gas differs very markedly from argon in its 

 spectrum, and in its behaviour at low temperatures, it must be 

 regarded as a distinct elementary substance, and we therefore 

 propose for it the name " metargon." It would appear to hold 

 the position towards argon that nickel does to cobalt, having 

 approximately the same atomic weight, yet different properties. 



It must have been observed that krypton does not appear 

 during the investigation of the higher-boiling fraction of argon. 

 This is probably due to two causes. In the first place, in order 

 to prepare it, the manipulation of air, amounting to no less 

 than 60,000 times the volume of the impure sample which we 

 obtained was required ; and in the second place, while metargon 

 is a solid at the temperature of boiling air, krypton is probably 

 a liquid, and therefore more easily volatilised at that temperature. 

 It may also be noted that the air from which krypton has been 

 obtained had been filtered, and so freed from metargon. A 

 foil account of the spectra of those gases will be published in 

 due course by Mr. E. C. C. Baly. 



University College, London. 



NO. 1495, VOL. 58] 



ON THE STABILITY OF THE SOLAR 

 SYSTEM} 



A LL persons who interest themselves in the progress of 

 ■'*■ celestial mechanics, but can only follow it in a general 

 way, must feel surprised at the number of times demonstrations 

 of the stability of the solar system have been made. 



Lagrange was the first to establish it, Poisson then gave a new 

 proof; afterwards other demonstrations came, and others will 

 still come. Were the old demonstrations insufficient, or are the 

 new ones unnecessary? 



The astonishment of these persons would doubtless be in- 

 creased if they were told that perhaps some day a mathematician 

 would show by rigorous reasoning that the planetary system is 

 unstable. This may happen, however ; there would be nothing 

 contradictory in it, and the old demonstrations would still retain 

 their value. 



The demonstrations are really but successive approximations ; 

 they do not pretend to strictly confine the elements of the orbits 

 within narrow limits that they may never exceed, but they at 

 least teach us that certain causes, which seemed at first to compel 

 some of these elements to vary fairly rapidly, only produce in 

 reality much slower variations. 



The attraction of Jupiter, at an equal distance, is a thousand 

 times smaller than that of the sun ; the disturbing force is there- 

 fore small ; nevertheless, if it always acted in the same direction, 

 it would not fail to produce appreciable effects. But the direc- 

 tion is not constant, and this is the point that Lagrange 

 established. After a small number of years two planets, which 

 act on each other, have occupied all possible positions in their 

 orbits ; in these diverse positions their mutual action is directed 

 sometimes one way, sometimes in the opposite way, and that in 

 such a fashion that after a short time there is almost exact com- 

 pensation. The major axes of the orbits are not absolutely 

 invariable, but their variations are reduced to oscillations of 

 small amplitude about a mean value. 



This mean value, it is true, is not rigorously fixed, but the 

 changes which it undergoes are extremely slow, as if the force 

 which produces them was not a thousand times, but a million 

 times smaller than the solar attraction. One may, therefore, 

 neglect these changes, which are of the order of the square of 

 the masses. As to the other elements of the orbits, such as 

 the eccentricities and the inclinations, these may acquire round 

 their mean value wider and slower oscillations, to which, how- 

 ever, limits may easily be assigned. 



This is what Lagrange and Laplace pointed out, but Poisson 

 went further. He wished to study the slow changes experienced 

 by the mean values — changes to which I have already referred, 

 and which his predecessors had at first neglected. He showed 

 that these changes reduced themselves again to periodic oscilla- 

 tions round a mean value which is only liable to variations a 

 thousand times slower. 



This was a step further, but it was still only an approxima- 

 tion. Since then further advance has been made, but without 

 arriving at a complete definitive and rigorous demonstration. 

 There is a case which seemed to escape the analysis of 

 Lagrange and Poisson. If the two mean movements are com- 

 mensurable among themselves, at the end of a certain number 

 of revolutions, the two planets and the sun will be found in the 

 same relative situation, and the disturbing force will act in the 

 same direction as at first. The compensation, to which I have 

 referred, will not any more be produced, and it might be feared 

 that the effects of the disturbing forces will end by accumulating 

 and becoming very considerable. More recent works, amongst 

 others those of Delaunay, Tisserand, and Gylden, have shown 

 that this accumulation does not actually occur. The amplitude 

 of the oscillations is slightly increased, but remains, nevertheless, 

 very small. This particular case, therefore, does not escape the 

 general rule. 



The apparent exceptions have not only been dispensed with, 

 but the real reasons of these compensations, which the founders 

 of celestial mechanics had observed, have been better explained. 

 The approximation has been pushed further than was done by 

 Poisson, but it is still only an approximation. 



It can be shown, in certain particular cases, that the elements 

 of the orbit of one planet will return an infinite number of times 

 to very nearly the initial elements, and that is also probably true 

 in the general case ; but it does not suffice. It should be shown 



1 Translation of a paper, by M. H. Poincar(S, in the Annuaire du Bureau 

 det Longitudes, 1898. 



