194 SCIENCE PROGRESS 



radiation. Some evidence is brought forward by Fowle that 

 the smaller the average solar radiation, or sun-spot number, 

 the greater is the absorptive power of atmospheric water 

 vapour. 



The " moist haziness " associated with water vapour causes 

 losses from the direct solar beam of about 2 per cent. 



The Photography of Stellar Spectra in the Infra-Red. — Very- 

 little has been done so far in the investigation by photography 

 of the spectra of stellar bodies in the region of long wave- 

 lengths, and the chief need of stellar spectroscopy to-day is 

 an addition to our knowledge in this direction. An investiga- 

 tion recently carried out by P. W. Merrill of the U.S. Bureau 

 of Standards is therefore to be welcomed (Scientific Papers 

 of the Bureau of Standards, Washington, No. 318). 



He investigated the results obtainable from the application 

 of dicyanin to stellar photography. Dicyanin is one of the 

 dyes used for staining photographic plates to make them 

 sensitive to the red and infra-red ; but, to obtain the full advan- 

 tages from its use, special methods of staining are required 

 and the plates must be used very shortly after sensitising. By 

 arrangement with Prof. Pickering, the 24-inch reflector of the 

 Howard Observatory was used, in conjunction with an objec- 

 tive prism. Results of high accuracy were not sought for, 

 the investigation only seeking to determine the possibilities 

 of the method. Seventy-seven spectra of twenty-nine stars 

 were obtained. 



The spectra investigated were chiefly those of the earliest 

 types (0 and B) and of the latest types (M, N and R). It was 

 found possible to photograph Fraunhofer's A band (wave- 

 length 0*760//.) and a considerable region of the spectrum 

 beyond. The greatest wave-length observed was o'Sy/M, 

 A strong band at 760/i, nearly coincident with the A band, 

 was discovered in spectra of type M. This band is particularly 

 strong in Mb and Md type spectra and may prove of use in 

 the classification of late type stars. It is thought to be prob- 

 ably due to titanium oxide, thus extending the series found 

 by Fowler in this type of spectrum. In spectra of class N 

 bands were observed at 6920, 7080, 7230 which may be iden- 

 tical in origin with cyanogen bands. 



The spectra of class R were found to differ considerably 

 from those of class N as regards the spectral distribution of 



