DR L. BECKER ON THE SOLAR SPECTRUM. 115 



permit of accurate interpolation. When this happened we introduced another table, 

 computed from the original one by altering all the first differences by the same quantity. 

 This quantity was selected so that the first and last standard lines of the series gave 

 the same correction. Of course "r" had again to be interpolated from this auxiliary 

 table for the standard lines, and the whole process repeated. The curve could then be 

 readily drawn. Geometrically the ordinates of the new curve are equal to the differences 

 of the ordinates of the first curve and of a straight line drawn through the first and last 

 points, or, in other words, the second table eliminates the progressive error in the motion 

 of the apparatus. 



The standard lines are those published by Professor H. A. Rowland, in his essay " On 

 the Relative Wave-lengths of the Lines of the Solar Spectrum." # To these were added 

 fifteen lines, principally between c and F, which were reduced from the memoir entitled 

 " Bestimmung der Wellenlangen von 300 Linien im Sonnenspectrum," t by Drs Muller 

 and Kempf, and from Angstrom's Recherches sur le Spectre solaire. Twenty-two 

 lines between b and F which were determined by Messrs Eowland and Muller-Kempf, 

 gave the constant difference — 0"19 A.U., the probable error of one difference being 

 ± 0*04. To the same region forty-seven comparisons between wave-lengths determined 

 by Eowland and Angstrom were made, and the differences combined by a curve. The 

 probable error of one difference amounted to ±0'07A.U. Angstrom's measures, 

 however, were only combined with those of Muller-Kempf with regard to their weights 

 when there was no doubt left that the wave-lengths referred to the same line. 



The reduction of the bulk of observations was carried on in three columns ; one 

 column contained the readings, the second the corrections of the zero point as taken from 

 the curve, and the third the wave-lengths, which were interpolated from the table with 

 the difference of the values of the two preceding columns. 



All the wave-lengths were then compiled, and the corresponding lines identified. This 

 turned out to be a most difficult task in those regions where many faint lines of the same 

 intensity stood close together and were occasionally wanting. In such cases special 

 drawings were prepared and the most probable identifications adopted. There was not 

 so much difficulty experienced with faint lines in the neighbourhood of darker ones. 

 Neighbouring lines of different intensities were invariably combined line for line with 

 others of corresponding blackness, although other combinations might have brought 

 the wave-lengths to a closer agreement. Very often close double lines were found not to 

 have been separated when the sun was close to the horizon or the faintness of the light 

 had necessitated a wider opening of the slit. 



The adopted values of the wave-lengths are the means of all the observations made, 

 as well at medium altitudes of the sun as near the horizon. In the case of double lines, 

 which had not always been resolved, the mean of the middle was first computed and then 

 corrected by the average distance between the lines. The fainter lines having, as a rule, 



* American Journal of Science, vol. xxxiii., March 1887. 



t Publicationen ales Astrophijsikalischen Observatoriums zu Potsdam, Fiinfter Band, Potsdam, 1886. 



VOL. XXXVI. PART I. (NO. 6). U 



