October i6, 1891.J 



SCIENCE. 



213 



of flames under various pressures up to forty atmospheres, 

 Professors Liveing and Dewar liave come to the conclusion 

 that, though the prominent feature of the light emitted by 

 flames at high pressure appears to be a strong continuous 

 spectrum, there is not the slightest indication that this con- 

 tinuous spectrum is produced by the broadening of the lines 

 of the same gases at low pressure. On the contrary, photo- 

 metric observations of the brightness of the continuous spec- 

 trum, as the pressure is varied, show that it is mainly pro- 

 duced by the mutual action of the molecules of a gas. Ex- 

 periments on the sodium spectrum were carried up to a 

 pressure of forty atmospheres without producing any definite 

 effect on the width of the lines which could be ascribed to 

 the pressure. In a similar way the lines of the spectrum of 

 water showed no signs of expansion up to twelve atmos- 

 pheres; though more intense than at ordinary pressure, they 

 remained narrow and clearly defined. 



It follows, therefore, that a continuous spectrum cannot 

 be considered, when taken alone, as a sure indication of 

 matter in the liquid or the solid state. Not only, as in the 

 experiments already mentioned, such a spectrum may be due 

 to gas when under pressure, but, as Maxwell pointed out, if 

 the thickness of a medium, such as sodium vapor, which 

 radiates and absorbs different kinds of light, be very great, 

 and the temperature high, the light emitted will be of exactly 

 the same composition as that emitted by lamp-black at the 

 same temperature, for the radiations which are feebly emitted 

 will be also feebly absorbed, and can reach the surface from 

 immense depths. Shuster has shown that oxygen, even in a 

 partially exhausted tube, can give a continuous spectrum 

 when excited by a feeble electric discharge. 



Compound bodies are usually distinguished by a banded 

 spectrum ; but, on the other hand, such a spectrum does not 

 necessarily show the presence of compounds, — that is, of 

 molecules containing different kinds of atoms, — but simply 

 of a more complex molecule, which may be made up of 

 similar atoms, and be, therefore, an allotropic condition of 

 the same body. In some cases — for example, in the diffuse 

 bands of the absorption spectrum of oxygen — the bands may 

 have an intensity proportional to the square of the density 

 of the gas, and may be due either to the formation of more 

 complex molecules of the gas with increase of pressure, or it 

 may be to the constraint to which the molecules are subject 

 during their encounter with one another. 



It may be thought that at least in the coincidences of 

 bright lines we are on the solid ground of certainty, since 

 the length of the waves set up in the ether by a molecule, 

 say of hydrogen, is the most fixed and absolutely perma- 

 nent quantity in nature, and is so of physical necessity, for 

 with any alteration the molecule would cease to be hydro- 

 gen. 



Such would be the case if the coincidence were certain; 

 but an absolute coincidence can only be a matter of greater 

 or less probability, depending on the resolving power em- 

 ployed, on the' number of the lines which correspond, and on 

 their characters. When the coincidences are very numerous, 

 as in the case of iron and the solar spectrum, or the lines are 

 characteristically grouped, as in the case of hydrogen and 

 the solar spectrum, we may regard the coincidence as cer- 

 tain; but the progress of science has been greatly retarded 

 by resting important conclusions upon the apparent coinci- 

 dence of single lines in spectroscopes of very small resolving 

 power. In such cases, unless other reasons supporting the 

 coincidence are present, the probability of a real coincidence 

 is almost too small to be of any importance, especially in the 



case of a heavenly body which may have a motion of ap- 

 proach or of recession of unknown amount. 



But even here we are met by the confusion introduced by 

 multiple spectra, corresponding to different molecular group- 

 ings of the same substance; and, further, to the influence of 

 substances in vapor upon each other; for when several gases 

 are present together, the phenomena of radiation and re- 

 versal by absorption are by no means the same as if the 

 gases were free from each other's influence, and especially is 

 this the case when they are illuminated by an electric dis- 

 charge. 



I have said as much as time will permit, and I think in- 

 deed suflBcient, to show that it is only by the laborious and 

 slow process of most cautious observation that the founda- 

 tions of the science of celestial physics can be surely laid. 

 We are at present in a time of transition, when the earlier, 

 and, in the nature of things, less precise, observations are 

 giving place to work of an order of accuracy much greater 

 than was formerly considered attainable with objects of such 

 small brightness as the stars. 



The accuracy of the earlier determinations of the spectra 

 of the terrestrial elements are in most cases insulEcient for 

 modern work on the stars as well as on the sun. They fall 

 much below the scale adopted in Rowland's map of the sun, 

 as well as below the degree of accuracy attained at Potsdam 

 by photography in a part of the spectrum for the brighter 

 stars. Increase of resolving power very frequently breaks up 

 into groups, in the spectra of the sun and stars, the lines 

 which had been regarded as single, and their supposed coin- 

 cidences with terrestrial lines fall to the ground. For this 

 reason many of the early conclusions, based on observation 

 as good as it was possible to make at the time with the less 

 powerful spectroscopes then in use, may not be found to be 

 maintained under the much greater resolving power of mod- 

 ern instruments. 



The spectroscope has failed as yet to interpret for us the 

 remarkable spectrum of the Aurora Borealis. Undoubtedly 

 in this phenomenon portions of our atmosphere are lighted 

 up by electric discharges: we should expect, therefore, to 

 recognize the spectra of the gases known to be present in it. 

 As yet we have not been able to obtain similar spectra from 

 these gases artificially, and especially we do not kni^w the 

 origin of the principal line in the green, which often appears 

 alone, and may have, therefore, an origin independent of 

 that of the other lines Recently the suggestion has been 

 made that the aurora is a phenomenon produced by the dust 

 of meteors and falling stars, and that near positions of cer- 

 tain auroral lines or flutings of manganese, lead, barium, 

 thallium, iron, etc., are sufficient to justify us in regarding 

 meteoric dust in the atmosphere as the origin of the auroral 

 spectrum. Liveing and Dewar have made a conclusive re- 

 search on this point, by availing themselves of the dust of 

 excessive minuteness thrown off from the surface of elec- 

 trodes of various metals and meteoroites by a disruptive dis- 

 charge, and carried forward into the tube of observation by 

 a more or less rapid current of air or other gas. These ex- 

 periments prove that metallic dust, however fine, suspended 

 in a gas will not act like gaseous matter in becoming lumi- 

 nous with its characteristic spectrum in an electric discharge 

 similar to that of the aurora. Professor Schuster has sug- 

 gested that the principal line may be due to some very light 

 gas which is present in too small a proportion to be detected 

 by chemical analysis or even by the spectroscope in the pres- 

 ence of the other gases near the earth, but which at the height 

 of the auroral discharges is in a sufficiently greater relative 



