354 



8 



calculated to be r"'33 and the spectra of the first, second and third order respect- 

 ively 2'"'51, 6™'88 and 5™'05 fainter than the star image as taken without the 

 grating placed in front of the reflector. Owing to the oblong figure of the spectra 

 the loss in limiting magnitude is still somewhat greater especially for the spectra 

 of higher order. Here we are chiefly interested in the spectra of the first order, be- 

 cause thej' have the greatest intensity and the main point is to reach as faint stars 



♦♦# 



Figure 2. 



as possible. The oblongness of the spectra of the first order varies sensibly with 

 the seeing. The better the seeing is, the more oblong the spectra will appear. In 

 the main we may say, that with the same exposure time the spectra of the first 

 order reach about 3™ less far than the stars taken without grating. Generally speaking 

 we must choose the constant, d-\- 1, of the grating by weighing the different points 

 of view against each other. The smaller the constant of the grating is, the farther 

 the spectra will be separated from each other and the same accuracy in measuring 

 will consequently give a greater accuracy of the effective wavelength. But at the 

 same time the spectra will get more oblong and less fit for accurate pointing, and 

 possible systematic errors, caused by the unsymmetrical figure — varying with the 

 definition — of the spectra, will be of more importance. Besides, the spectrum 



