Mr. J. N. Lockyer oil Spectrum-Analysis* 389 



at the red end of the spectrum — that the absorption spectra of 

 chlorine, iodine, bromine, &c. are columnar, and that these are 

 broken up by the spark just as the band spectra of compounds are 

 broken up — and that it is probable that no compounds exist in 

 the sun. The following facts, gathered from the work already 

 accomplished by Rutherford and JSecchi, are stated. 

 There are three classes of stars : — 



1. Those like Sirius, the brightest (and therefore hottest ?) star 

 in the northern sky, their spectra showing only hydrogen lines 

 very thick, and metallic lines exceedingly thin. 



2. A class of stars with a spectrum differing only in degree from 

 those of the class of Sirius ; and to this our sun belongs. 



3. A class of stars with columnar or banded spectra indicating 

 the formation of compounds. 



The question is asked whether all the above facts cannot be 

 grouped together in a working hypothesis, which assumes that in 

 the reversing layers of the sun and stars various degrees of " celes- 

 tial dissociation " are at work which prevents the coming together 

 of the atoms which, at the temperature of the earth, and at all arti- 

 ficial temperatures yet attained here, form the metals, the metal- 

 loids, and compounds. 



In other words, the metalloids are regarded as quasi compound 

 bodies when in the state in which we know them ; and it is sup- 

 posed that in the sun the temperature is too great to permit them 

 to exist in that state in the reversing layer, though they may be 

 found at the outer portions of the chromosphere or in the corona. 



It is suggested that, if this hypothesis should gain strength from 

 subsequent work, stony meteorites will represent the third class 

 of metalloidal or compound stars, and iron meteorites the other or 

 metallic stars. 



The paper concludes as follows : — 



" An interesting physical speculation connected with this work- 

 ing hypothesis is the effect on the period of duration of a star's 

 heat which would be brought about by assuming that the original 

 atoms of which a star is composed are possessed of the increased 

 potential energy of combination which this hypothesis endows them 

 with. From the earliest phase of a star's life the dissipation of 

 energy would, as it were, bring into play a new supply of heat 

 and so prolong the star's life. 



" May it not also be, if chemists take up this question, which 

 has arisen from the spectroscopic evidence of what I have before 

 termed the plasticity of the molecules of the metalloids taken as a 

 whole, that much of the power of variation which is at present 

 accorded to metals may be traced home to the metalloids? I need 

 only refer to the fact that, so far as I can learn, all so-called changes 

 of atomicity take place when metalloids are involved, and not when 

 the metals alone are in question. 



"As instances of these, I may refer to the triatomic combi- 

 nations formed with chlorine, oxygen, sulphur, &c. in the case of 



