534 



SCIENCE 



[N. S. Vol. XXXIV. No. 877 



silicon spark in air gave the wave-length of 

 X 4,567 ". . . quite a little larger than on the 

 plates talfen with tlie observatory grating." This 

 latter measure was not given, but it is in the 

 direction of the value for the line by Exner and 

 Ilasclielc. Frost (Aph. Jour., 1910) recognized 

 the fact that the use of his laboratory determina- 

 tions of the wavo-longtlis did not increase the 

 accordance, in stellar spectra, of the velocities 

 from the separate silicon lines. 



The large systematic differences in the labora- 

 tory determinations would indicate that the effec- 

 tive wave-lengths of these lines may be influenced, 

 in some way at present unknown, by the condi- 

 tions of the e.xporiment. As the lines are gen- 

 erally diffuse and are sensitive to the atmosphere 

 surrounding the spark, it seems likely that differ- 

 ences in the electrical conditions of the spark or 

 in tlie surrounding atmosphere produce an unsym- 

 metrical widening of the lines. 



Further laboratory investigation of these lines, 

 under a variety of different conditions, is of great 

 importance in the discussion of fundamental prob- 

 lems in astrophysics. 



On Fundamental Systems of Wave-lengths in Stel- 

 lar Spectra, and especially for the B Type 

 Stars: S. Albrecht. (Read by Mr. Parker.) 

 The object of this note is to point out a clew 

 to the elimination, in part at least, of systematic 

 errors in the wave-lengths of the B type stars. 



In the stellar spectral types A to M a consid- 

 erable number of the spectrum lines in two or 

 more neighboring main divisions of the spectral 

 classification have a common origin. This makes 

 possible a comparison of the wave-lengths in any 

 one type with the wave-lengths or system of wave- 

 lengths in any other type, from types A to M, 

 both inclusive. In consecutive spectral types the 

 systems of wave-lengths can be compared with com- 

 parative ease. In the B types, however, the lines 

 are, with a few exceptions, of entirely different 

 origin from the lines of even the A type, and it 

 is only by means of the few lines which the B 

 and A types have in common that the system of 

 wave-lengths in the B type can be connected with 

 the systems of wave-lengths in types A to M. 



A connection with each other of the systems of 

 wave-lengths in the different spectral types is 

 highly desirable in the solution of several prob- 

 lems, the more important of which are perhaps: 

 (o) the classification of stellar spectral types; 

 (b) the elimination from the system of wave- 

 lengths for each type of shifts, other than those 



duo to radial velocity, which are systematic for 

 the entire system of lines, or for groups of lines 

 in each type, and which may be due to such effects 

 as ' ' pressure " or to other causes. The shifts 

 which are shared by all lines are at present in- 

 cluded in the radial velocities of the stars. 



One of the best lines which is at present avail- 

 able for this purpose is H-gamma. The wave- 

 length in the A type was found to be 4,340.655, 

 and it diminishes progressively, slowly from types 

 A to G and more rapidly from types K to M. 

 If we make a plot with wave-lengths as ordinates 

 and spectral types as abscissas, a smootli curve can 

 be drawn through the points in the plot from types 

 A to M. If this curve were extended to the B 

 type, it would indicate a wave-length for that type 

 of 4,340.657. The vcave-length actually found for 

 the line in the B types is 4,340.627. 



As pointed out above this difference for the B 

 type, relatively to other spectral types, is at pres- 

 ent to be taken as merely an indication of a 

 method of approach for the solution of this 

 problem. The final solution will have to be based 

 upon a greater number of lines, extending over a 

 longer portion of the spectrum. For a proper dis- 

 tribution to each spectral type, of relative differ- 

 ences which may be found in the different types, 

 it will be absolutely essential to have an accurate 

 knowledge of the behavior of spectrum lines under 

 various different conditions. 



In conclusion I wish to refer briefly to the im- 

 portance which the solution of this question has in 

 connection with our conception of the structure of 

 the sidereal universe. If the results found for 

 Il-gamma should be confirmed by a more extensive 

 investigation, such systematic differences in the 

 radial velocities of stars as a function of the 

 spectral type, as were found, I believe, by Kap- 

 teyn, Frost and Campbell, may find a simple 

 explanation. 



Photographic Determination of the Position of the 

 Moon: Henry Norris Eussell. (Eead in con- 

 nection witli the Symposium on Photographic 

 Astrometry.) 



The photographs discussed below were taken at 

 Harvard by Mr. King, and, by the kindness of 

 Professor Pickering, were sent to the writer for 

 discussion. They were measured and reduced at 

 Princeton by Professor A. H. Joy, of the Syrean 

 Protestant College, Beirut, and the writer, accord- 

 ing to plans prepared by the latter. An account 

 of the results will soon appear in the Harvard 

 Annals. 



