472 A . E. Bostwick — Preliminary Note on the 



with plane parallel glass faces one-quarter inch apart were 

 therefore substituted and the spectrum found to be the same 

 whichever liquid was next the slit. The position of the bands 

 was measured by means of a tube bearing an illuminated slit 

 the image of which moved over that of the spectrum as the 

 tube was revolved around the same axis as the telescope and 

 collimator. The amount of revolution was measured on a 

 scale bearing a vernier reading directly to five minutes of arc, 

 and by estimation to one minute. The cells being both in 

 front of the spectroscope slit, a liquid in each, the image of 

 the micrometer slit was fixed in the middle of the band whose 

 deviation was to be measured. The liquids were then turned 

 from the cells into a test tube, mixed there, and the mixture 

 turned half into one cell and half into the other. The image 

 of the slit was then seen to be no longer in the middle of the 

 band, and to bring it there again the tube had to be revolved 

 through from V to IT, according to the ratio of the liquids. 

 But to avoid the effect of any unconscious bias on the part of 

 the observer, the first band was observed with unmixed liquids, 

 then the second, and then the two in the same order with 

 mixed liquids, so that the observer did not know whether 

 there were any movement, or in which direction it was, till he 

 had fixed the micrometer and taken the reading. 



The principal liquids used were those employed by Melde 

 in his first experiments ; namely, solutions in water of carmine 

 with ammonia, bichromate of potassium, and ammoniacal sul- 

 phate of copper. The carmine and bichromate were tried first, 

 and the former being kept at the same strength, the latter was 

 diluted successively to •£, I, |, T V, etc., of its first strength, 

 which was that which just enabled the carmine bands to be 

 seen. Dilution seemed to have no effect on the displacement 

 of the band till the bichromate was very weak, when it fell 

 off almost entirely. The mean of sixteen measurements, at 

 various strengths of the bichromate, was 10' '9 displacement 

 toward the blue end of the spectrum (absorbed by the bichro- 

 mate) for the least refrangible carmine band, and that of ten 

 measurements for the most refrangible was 8 /# 8. The meas- 

 urements, which were all in the same direction, ranged in the 

 first case from 19' to 6' and in the second from 14' to 5 ; , a 

 result due probably both to the difficulty of fixing the slit 

 exactly in the middle of the band and to the variation in 

 strength. An arc of 1' in this place corresponds to a difference 

 of wave-length of about 4 millionths of a millimeter. The 

 width of the bands was about 40' so that the average displace- 

 ment was about one-quarter of the width, an amount plainly 

 evident without measurement. 



Carmine and ammoniacal copper sulphate were now tried in 

 like manner, and the bands were displaced toward the red, 



