35JO 



reading on the scale attached to thi^i screw gave tlie iiwaa wave- 

 length of the light used with an accnracj of a few ANGSTROM-units. 

 By repeating the calibration during the experiments it was proved, 

 that this mean wavelength could always be reproduced with the 

 above mentioned accuracy. This now is more than sufficient, as for 

 instance in the green part of the spectrum a change of I = 5400 

 into P. :^ 5500 and at the greatest possible velocity of the water 

 the shift of the interference fringes becomes O.GGO instead of 0.675 

 of the distance between two fringes. Even a change of 10 A.U. 

 in the wavelength of the ligiit used corresponds to 0,0015 only of 

 the distance between the interference fringes, while the probable 

 error of the tinal result is of the order of magnitude of 0.005. 



In order to determine the place of the interference fringes 1 used 

 two or rather three ditferent methods and in a few experiments 

 onlj' eye observations were made. In one series of expei'iments a 

 wire-net. which could be turned and shifted was adjusted in the 

 focal plane *- (see Fig. 1) '). In the focal plane of the telescope/ we 

 took photos of the interference fringes, while care was taken that 

 one wire was parallel to the fringes and that the other passed 

 through the middle of the lield. 



An advantage of this method is, that the interchanging of the 

 photographic plates in the focal plane of ƒ does not disturb the 

 relative position of the interference_ fringes and the wires. With 

 this method however it is rather difficult to adjust the wire-net 

 accurately as it is so far away from the observer. Moreover the 

 net must be very fine because of the strong magnifying power of 

 the telescope. On the proposal of Prof. Wood I used in a second 

 method Rowi,.\nd's artifice ') for the comparison of spectra. Rowlan» 

 puts in front of the photographic plate a brass plate with longi- 

 tudinal aperture of the same width as the thickness of the plate, 

 which could turn round a horizontal axis in front of the photo- 

 graphic plate. The rotation could easily be limited to an angle of 

 90°. By means of two fine quartz wires adjusted perpendicular to 

 the plane of the brass plate the position of the plate could be 

 measured accurately and corrected if necessary. 



Two photos taken by this method are reproduced in the Plate 

 (Fig. 4 and 5). The outer system of interference fringes has been 

 obtained while the water was streaming in one direction ; the inner 

 system corresponds to a ciu'rent in the opposite direction. 



Fig. 5 shows also the shadow of the fine quartz wires. 



') See the first part. 



2) Ames, Phil Mag. (5) 27, 369. 1889. 



