MKMoiuXXYlII.] DISTRIBUTION <>1 1 1 KAT IN THE SPECTRUM. 397 



tention must be paid to tbeir magnetic perturbations, 

 win-tiler arising from local or other causes; and, since 

 the vibrations are very slow, ample time must be given 

 before the reading is ascertained. 



The condition of the face of the pile is of importance. 

 It must be such as to extinguish as completely as possible 

 all the incident rays. To paint it with lamp-black, mix- 

 ed with gum, will not answer the surface so produced 

 is too glossy and reflecting. The plan I have found best 

 is to take a crlass tube half an inch in diameter and six 



o 



inches long, open at both ends, and use it as a chimney. 

 A. piece of camphor being set on fire at the lower end, 

 and the face of the pile to be blackened being held for a 

 moment at the upper, it is covered with a dense black 

 film, without any risk of injury to the pile. Even at the 

 best, when this has been done, there is an unavoidable 

 source of error in the want of perfect blackness of the 

 lamp-black. It is sufficient to inspect the face of the pile 

 when receiving rays from the concave mirror to be satis- 

 fied how large a portion of light is reflected. The ex- 

 periments of Dr. Tyridall show that this substance also 

 transmits a considerable percentage of the heat falling 

 on it. Its quality of transmitting light is well known 

 to every one who has looked at the sun through a 

 smoked glass. 



The galvanometer I have used is calibrated according 

 to the usual method. The numbers given in this Memoir 

 do not represent the angles of deflection, but their corre- 

 sponding forces. 



The proper position of the intercepting screens #, i, can 

 often be verified with precision by looking through blue 

 cobalt glass. This glass insulates a definite red, an 

 orange, and a yellow ray in the less refrangible regions, 

 and then, commencing with the green, gives a continuous 

 band to the end of the violet. Its red ray begins at the 



