H. Crew — Rotation of the Sun. 157 



The observations in latitudes 6 0- 64, 7° "37, and all those of 

 July 15th were made when the definition was marked " poor," 

 "very poor," "sky covered with thin clouds," etc.; I have 

 therefore marked these doubtful, but have not rejected them 

 entirely. The values of v'-v" above given are for the synodic 

 period. The weight given to each observation is proportional 

 to the " number of settings." 



Reducing by the method of least squares, we nave (since 

 a sin ^>= 0*086 miles per sec.) for the mean equatorial velocity, 



y'--y"=2*437±*024 miles per sec. 



which corresponds to a true period of 25*88 days. 



Errors. — When it is considered that the total displacement 

 amounts to only -^ of the distance between the D lines, it 

 will not be surprising if the largest error is that made in setting 

 the cross hairs on the lines. Two smaller errors may have been 

 introduced by the unequal heating of the jaws of the slit ; and 

 by a slight vertical displacement of the sun's image when shift- 

 ing from one limb to the other. But these must have been of 

 the second order, for immediately before making an observa- 

 tion, the adjustment of the instrument was tested by setting on 

 a sharp atmospheric line. Not the slightest motion could ever 

 be detected. Since we require only the difference of the read- 

 ings on the two limbs, and since these are taken in rapid suc- 

 cession, and under conditions practically identical, it will be 

 observed that all the ordinary errors of the spectrometer, ex- 

 cept that of setting the cross-hairs, are eliminated. Errors of 

 this kind are, however, such as would, to a great extent, 

 counterbalance each other in a large series of observations. On 

 the contrary, a single setting differed from the mean of the 

 series to which it belonged, on the average, by 11 per cent, 

 while 41 sets of observations (of eleven settings each) still differ 

 from their mean by as much as eight per cent. This leads one 

 to suspect the presence of irregular horizontal currents. For 

 the regular variation of angular velocity with latitude, described 

 below, is certainly not sufficient to account for an average error 

 of eight per cent. Currents having velocities, very moderate 

 indeed, compared with some already observed by Prof. Young,* 

 would more than account for all the irregularities of these ob- 

 servations even if they were perfect in other respects. 



Results. — When beginning this work, I expected to find the 

 angular velocity decreasing as the latitude of observation in- 

 creased, as in Carrington's curve for the motion of sun spots, 

 figured in Lockyer's Chemistry of the Sun, p. 425. On_the 



* See also Schellen: Spectrum Analysis, (London, 1885), pp. 378-388. 



