Chemistry and Physics. 415 
roughly estimating the percentage of frequency with which the 
lines were seen during the six weeks of observation at Sherman 
in the summer of 1872. According to Young, 7066, 5876, 4472 
have the frequency number 100, while 6678, 5016, 4922 have the 
numbers 25, 30, 30, showing that one of the two constituents was 
_always present, while the other was only seen about once in every 
four cases. ; 
The lines of both constituents have been observed in the spectra 
of aconsiderable number of stars /, 6, ¢, 6, y, Orionis, « Virginis, 
fi Persei, 6 Tauri, 7 Urse majoris, @ Lyre. In the spectrum of 
fi Lyre, thirteen lines have been identitied with certainty. But 
the most interesting case in point is the spectrum of Nova Aurigz, 
that wonderful star whose sudden appearance was announced 
to astronomers in 1892 by an anonymous postcard. In the 
spectrum of Nova Aurigz the two lines 5016 and 4922 were 
very strong, while 4472 was weak and 5876 has only been 
seen by Dr. Huggins, we believe only on one occasion, and appears 
to have been very weak. Now 5016 and 4922 belong to the 
lighter constituent, and are together with 6678 the strongest lines 
in the visible part of the spectrum; while 5876 and 4472 are the 
strongest lines of the other constituent in the visible part of the 
spectrum. In Nova Aurige, therefore, the lighter constituent 
gave a much brighter spectrum than helium proper. But there 
may here be raised an objection, which indeed we do not know 
how to refute. Why has the line 6678 not been observed? It is 
a pity that the red part of the spectrum cannot be more easily 
photographed. Nova Aurigze has now become very weak, and 
besides the spectrum is quite altered, so that we shall never know 
whether the red line 6678 was really absent or has only escaped 
notice. 
From the fact that the second set of series is on the whole 
situated more to the refrangible part of the spectrum, one may, 
independently of the diffusion experiment, conclude that the ele- 
ment corresponding to the second set is the heavier of the two. 
In the spectra of chemically related elements like Li, Na, K, Rb, 
Cs, or Mg, Ca, Sr, or Zn, Cd, Hg, the series shift to the less 
refrangible side with increasing atomic weight. But it appears 
that in the spectra of elements following each other in the order 
of their atomic weights in a row of the periodic system like 
Na; Mp. Al: 
K, Ca; 
Cu, Zn; 
Rb, Sr; 
Ag, Cd, In; 
the series shift the opposite way, so that the spectrum of the 
element of greater atomic weight is as a whole situated further to 
the more refrangible side. Now in our case the density of the 
gas has been determined by Langlet (published by Cleve) and by 
Ramsay to be about double the density of hydrogen. Assuming 
