11 



357 



order were of measurable intensitj'. For these effective wavelengths from the spectra 

 of the P' order new corrections to r/ = '1 mm were derived. They are given in 

 Table 4. It is seen, that for these usually very strong images no sensible variation 

 of the effective wavelength with d was found. For these images I have therefore 

 adopted the constant correction — 12/(. This correction is somewhat different from 

 the one ( — Tp,) derived above for exposures of the same intensity. But it is evident, 

 that there will be a systematic difference between the qualities of two images, 

 showing the same diameter d of the central star, but in one case the spectra of 



Table 3. 



Corrections to normal strength of the image, spectra of P' order, adopted values. 



diameter of central image, d -05 -06 -07 -08 -09 -10 -11 -12 -13 m 



adopted correction of the effective | ^ _2 _3. _^ 



■wavelength to d = -10 mm j i i i i i 



diameter of central image, (Z -14 -15 -16 -17 -18 -19 -20 •24 mm 



adopted correction of the effective ) ^ „ „ „ „ 



, ,^ , . ,„ r —5- —6 —6- —7 —7 —7- —7- —T ft 



wavelength to d = -10 mm ) 



Table 4. 



Corrections to normal strength of the image, spectra of P' order, when spectra of 

 higher order were also measured, observed values. 



diameter of central image, d -13 -14 -15 -16 -17 -18 '19 mm 



mean correction of the effective wave- 1 _^^g.^ _^^.2 _^^.g _^_^ _^^.^ _^^,^^ _^^^_ 



length to d = -10 mm ) 



number of differences used 3 4 5 10 24 40 38 



diameter of central image, d -20 -21 -22 -23 -24 -25 -26 mm 



mean correction of the effective wave- ^ ^„ „„ ^„ „ ^, , 



;■— 11-0 —12-0 —12-8 —11-9 —12 2 —12-0 —15-4« 

 length to d := -10 mm f 



number of differences used 40 30 85 71 62 50 10 



higher order measurable, and in the other not. In the latter case the images will 

 be more nebulous (owing to unquietness of the air or great distance from the centre 

 of the field) and this may give rise to systematic ditTerence in the effective wave- 

 length. 



5. Precision of the Effective Wavelengths. 



The effective wavelengths found for the same star from different plates showing 

 exposures of the same intensity (same d) were compared, and the mean error of a 

 single effective wavelength derived from the differences between these values. The 

 results are given in Table 5 and the adopted weights in Table 6. 



As only one setting was made on each spectrum it is impossible to determine, 

 which fraction of the square of the mean error is due to errors of pointing and 

 which to defects of the image. In this respect my earlier paper (Potsdam Publ. 



47* 



