PHYSICS: H. 5. UHLER 
487 
4. The probable distribution of colors in the background of N. G. C. 
1647, and that actually observed for M 67, is similar to that for the 
North Pole and the region of S Cygni. 
5. In neither case is the limiting magnitude low enough to include all 
the cluster stars; but as far as these limits, the percentage distribution 
of the different colors in the cluster is notably different from that in the 
background; moreover, there is a marked difference for the two clusters. 
N, G. C. 1647 contains a considerable number of b and a stars, and prac- 
tically none of the k, and m classes, except as they may be fainter 
than the limiting magnitude. M 67, on the other hand, shows no h 
and only 3% of a stars. The maximum frequency of 71% is for g 
and, curiously enough, no m stars appear. 
^Mt. Wilson Conir., No. 100; Astrophys. /., 42, (1915). 
^Mt. Wilson Contr., No. 102; Astrophys. J., 42, (1915). 
3 A Notation for Use in the Discussion of Star Colors, these Proceedings, 1, 481 (1915). 
^Mt. Wilson Contr., No. 81; Astrophys. /., 39, 361 (1914). 
ON THIELFS 'PHASE* IN BAND SPECTRA 
By Horace Scudder Uhler 
SLOANE PHYSICAL LABORATORY. YALE UNIVERSITY 
Pre«ented to the Academy. July 24, 1 9 1 5 
In the emission spectra of all compounds and of many elements the 
lines start abruptly at a certain wave-length, near which they are very 
close together, and separate more and more as the distance from the 
beginning or ^head' is increased. Such spectra are called band or 
channeled spectra because, with relatively small dispersion, they have 
the general appearance of the chiaro-oscuro of a channeled column 
illuminated laterally. One of the problems of spectroscopy consists 
in finding the law governing the relative frequencies of the lines of 
one band or of a group of bands. In other words, the problem is to 
arrange the lines in series and to express their frequencies by a mathemat- 
ical formula, as has been done for a fairly large number of line spectra. 
The analysis of a line spectrum into series is greatly facilitated and 
made secure by the fact that lines of the same series show their functional 
relationship by the similarity of the changes which they undergo when 
the source is subjected to pressure, or is placed in a magnetic field, 
etc. On the other hand, the frequencies of the lines of the majority 
of well-developed band spectra do not exhibit the pressure and Zeeman 
effects. Consequently, since physical criteria are lacking, the group- 
ing into series of the fines of band spectra depends wholly upon the 
