1912.] SCHLESINGER— RADIAL VELOCITIES. 179 



graphs. More than a thousand such stars are available at the present 

 time and in a few years this number will be at least doubled ; so that 

 there should never be any great difficulty in finding a suitable test 

 object close at hand. The process of observing would then perhaps 

 consist of an exposure on a star whose velocity we wish to determine, 

 immediately followed in each case by an exposure (made on the 

 same plate close to the first) on a neighboring star whose velocity is 

 already known. The difference of the lengths of these two spectra 

 is then to be measured, converted into units of kilometers per second 

 and applied to the known velocity. 



A quarter of a century ago the suggestion was made by Picker- 

 ing, in connection with his experimental work for the Draper Memo- 

 rial, that radial velocities could be determined from objective prism 

 spectra if some absorptive medium could be found that would pro- 

 duce one or more narrow and sharp absorption bands. If such a 

 substance were interposed at any point in the spectrograph, or in- 

 deed anywhere between the star and the plate, the resulting spectra 

 would also show these bands in positions not affected by the velocity 

 of the star, and would thus offer a beautifully simple method for 

 determining velocities. Pickering made a search for a substance 

 with this very desirable quality, but at that time did not succeed in 

 finding a satisfactory one. Recently, however, he suggested this 

 subject to Professor R. W. Wood, who after experimenting with 

 various compounds has proposed neodymium chloride for this pur- 

 pose. This substance introduces into the photographic region of the 

 spectrum a number of absorption bands. One of these, at A 4,272, 

 is sharp and fairly narrow, having a width of about three angstroms, 

 a quantity that corresponds to a velocity of about 200 kilometers 

 a second. This substance seems, therefore, to oft'er a method for 

 measuring the velocities of certain stars with a moderate degree of 

 precision. Just how accurately this can be done, we must wait for 

 actual experiments to tell us; and such experiments are under way 

 in at least two observatories. There can be little doubt that an 

 accuracy represented by a proba'ble error of not more than ten 

 kilometers can be attained in this way. This Pickering-Wood method 

 is hardly applicable to any but stars of the A and B types. In stars 



PROC. AMER. PHIL. SOC, LII. 209 L, PRINTED JUNE 6, I9I3. 



