792 



SCIENCE 



[N. S. Vol. XXXIX. No. 1013 



All these lists are doubtless incomplete at 

 the lower end. Later researches have shown 

 that oxygen, gallium, ruthenium and all the 

 rare earths given in Mitchell's list, should be 

 added to Rowland's table, raising the number 

 of elements represented to forty-five. The 

 presence of nitrogen is also indicated by the 

 appearance of the cyanogen bands, but not 

 by its ow];i lines. Photographs showing fainter 

 lines would probably considerably extend the 

 list of elements recognizable in the flash- 

 spectrum. The list of elements in the earth's 

 crust is certainly very incomplete below those 

 which form 0.01 per cent, of the total. If 

 carried out to the limit it should of course 

 include all the elements, and it is not yet 

 known what positions some of them, such as 



the rare earths, would occupy in a complete 

 scheme. The list for the stony meteorites is 

 probably far from exhaustive, and it is not 

 safe to draw conclusions from the failure of 

 some elements to appear in it. It may be 

 added that, according to Farringdon,* the 

 average percentage of nickel in meteoric irons 

 is about 7.5, and that of cobalt about one 

 tenth as much, while that of copper averages 

 about 0.02. It would therefore seem reason- 

 able to suppose that the amounts of " Ni ~\- 

 Co " given in the table should be divided be- 

 tween the two metals in about this ratio. 



Upon comparing the lists of Rowland and 

 Clarke, we meet at once the fact — one of the 

 commonplaces of astrophysics — that the non- 

 metallic elements, with the exception of car- 

 bon and silicon, are scarcely if at all repre- 

 sented in the solar spectrum. The only one 

 whose lines appear is oxygen — which is from 

 20 to 100 times more abundant in accessible 

 materials than all the others put together 

 (excepting C and Si, as above). If we simply 

 accept this fact (which is still without ade- 

 quate explanation), and exclude these non- 

 metallic elements from the comparison, the 

 similarity between the order of the remaining 

 elements in the two lists is remarkable. 



Of the eight metallic elements (including 

 carbon and silicon under this head for the 

 moment) which are most abundant in the 

 earth's crust, six are among the eight whose 

 lines are strongest in the solar spectrum, and 

 one of the other two comes ninth in the solar 

 list. Of the next eight metallic elements in 

 the terrestrial list, seven are found among the 

 second group of eight in the solar list, and 

 the other one (Ni) is among the first eight. 

 That is, fifteen of the sixteen leading metallic 

 elements are common to the two lists, and 

 there is a general similarity in their relative 

 order in the two. 



Beyond this point comparison becomes 

 hardly practicable, as the terrestrial list is 

 probably incomplete. Four of the next eight 

 elements in Rowland's list are rare earths, for 

 which there are as yet no sufficient analytical 



* Publications of the Field Columbian Museum, 

 Geological Series, Vol. 3, No. 5, p. 110. 



