OF ARTS AND SCIENCES. 149 



In the above historical sketch I have tried to represent, as fairly as 

 possible, what has been done in the way of measuring the distribution 

 of heat in the spectrum. Some of the results are undoubtedly quite 

 wrong, and nearly all of them contain very considerable errors ; but 

 the experiment is a delicate one, and the results are peculiarly liable 

 to modification from disturbing causes. The results often vary very 

 widely, even when obtained by eminent observers and with excellent 

 apparatus ; the reason being that the unknown conditions of the 

 problem have been so different. 



Though so large a number of researches have been undertaken, 

 they have all been entirely qualitative. The source of radiation, the 

 dispersing medium, and the means of measuring the heat in the spec- 

 trum, have been varied, and the qualitative effects noticed. 



The following research has been undertaken with a view to deter- 

 mining the quantitative relations that may exist between the nature 

 and the ternj)erature of the source of radiation, on the one hand, and 

 the geometrical form of the curve of distribution, on the other. A 

 long and laborious series of preliminary experiments was made, in 

 addition to a study of those of other experimenters, in order to deter- 

 mine the nature and magnitude of the disturbing causes. A few 

 of these may, perhaps, be best referred to here, but the larger num- 

 ber will appear in an examination of the apparatus, in which the 

 preventive means also af)pear. 



The first difficulty is, of course, the absorption exercised by the 

 rock-salt lens and prism. Although chemically pure and perfectly 

 cloudless rock-salt has so little selective absorption as to make it an 

 unimportant factor in these researches, the slightest deposition of 

 moisture, or the slightest cloudiness of the material, will exercise a 

 very decided influence. A deposition of moisture, not noticeable to 

 the eye, may form a slight coating of brine that will quite materially 

 alter the form of the curve. 



This was a serious cause of error in my preliminary experiments, 

 and undoubtedly has considerably interfered with other experiment- 

 ers' results. To prevent this, cloudless rock-salt was chosen in the 

 first place, and carefully polished with oxide of tin and alcohol before 

 using. It was afterwards always kept in a perfectly dry atmosphere, 

 secured by means described later. 



The selective absorption of the thermopile was guarded against by 

 covering the face with carefully deposited camphor-black, — the select- 

 ive absorption of which must be too small to enter as a factor here, — 



