L891.] Visibility of the different Rays of the Spectrum. 511 



placed at different positions in the spectrum, and the rotating 



3tors closed till all light had vanished, when the aperture was 

 loted. The first plan is the more convenient of the two, and gives 

 rery accurate results ; though in some positions of the spectrum the 

 second method must be adopted, since the graphic curve formed 

 from the readings becomes almost a horizontal straight line at one 

 portion of the spectrum. As will be seen from the table, it is quite 

 evident that no one aperture of the slit of the collimator and of that 

 in the slide would suffice to give the entire range of disappearance of 

 the spectrum, and that at least three settings are necessary. At each 

 change the D light falling on the ground glass was measured, and the 

 necessary factors to make the readings on one scale were derived from 

 these measurements. 



Four sets of measures throughout the spectrum were made on 

 different days. No one differed to any appreciable extent from the 

 other. A mean of the four has been taken as representing the truth, 

 and the measures given in the first table are those of that which most 

 nearly approaches this mean. It may be stated that very rarely did 

 one curve differ more than 4 per cent, from another at any portion 

 of the spectrum. The readings were taken when the eye had 

 rested in darkness some time, and were often repeated a con- 

 siderable number of times, the first parts measured being re- 

 measured last. That the eye was equally sensitive throughout the 

 time may be judged from the fact that the two sets of readings 

 scarcely ever differed. The process of making these measures of ex- 

 tinction is very fatiguing, and probably rather detrimental to the 

 eyesight ; owing to the strain on the eyes, one set of readings is 

 usually as much as can be properly carried out on any one day, if 

 Accurate results are to be looked for. 



It is now three years ago since I began this research, and, after 

 trying various plans, I have come to the conclusion that the method 

 now described is the most easy, as it is the most simple. 



There is one point in the method which might be open to criticism, 

 .and that is that the cutting off the light by rotating sectors might 

 iuse some error in the results. This criticism, I may say, I raised 

 in my own mind at its very commencement, and found that it was 

 innecessary. Polarising the light entering the slit of the colli- 

 lator, and then dimming ib by means of a Nicol's prism placed in 



Dnt of the colour patch, proved an unsatisfactory method for an- 

 swering the criticism, as in no case could a total disappearance of a 

 aright light be brought about ; but by diminishing the area of the 

 Dlour patch by placing different apertures of diaphragms in front of 

 the last prism of the colour-patch apparatus (and thus throwing on 

 the ground glass discs of light of various areas), the truth of the results 

 was readily verified. The two sets of measures, one made in this way 



