and an Optical Method of Calibration. 429 



to obtain homogeneous light when white light is placed 

 before the collimators. If it is too narrow (less than *5 millim.) 

 diffraction takes place, and if too wide — it should not exceed 

 1 mm. — the light is no longer homogeneous over the field of 

 view. The objective of R is usually stopped to about 

 15 millim. The field should then be perfectly circular and 

 crossed in the centre by a horizontal band 5 millim. wide. 

 In the calibration of the bilateral slit of T v , the rotating disk 

 is placed immediately in front of the slit of T, its axis being 

 in the plane of the axes of T and T', so that its radii cross the 

 vertical slit at right angles. The disk may then be moved 

 forward or backward by only a small amount to bring 

 different sectors over it. The pitch of the screw should 

 preferably be '25 millim. and a total motion of 2 millim. will 

 be sufficient, the working width of both slits being usually 

 not greater than 1 millim. The radiant is, in general, an 

 incandescent lamp with a ground-glass plate before the slit. 

 With careful adjustment the instrument has given for the 

 mean colours of the spectrum readings differing by less than 

 "5 per cent, from a mean of ten readings, or a mean error of 

 less than "25 per cent, for one setting, or double the sensi- 

 bility heretofore obtained. This is largely brought about by 

 the fact that adjustment is easier, and there is only one 

 refraction instead of three as in the Lummer-Brodhun form ; 

 the boundary is sharper and vanishes more completely. The 

 question of a suitable source has, however, been a serious 

 difficulty, particularly one of sufficient intensity to make com- 

 parisons of large absorptions in the blue and violet. All the 

 various forms of gas-burners have been found to be too incon- 

 stant. Specially constructed lamps with flat wide filaments 

 or with close spirals, together with frosted globes and plates, 

 have given almost as great a constancy and sensibility in the 

 case of high absorption, as obtained above with unobstructed 

 rays and under more favourable circumstances. Large varia- 

 tions in the voltage do not seem to affect the slit calibration, 

 which remains constant under varying conditions. The 

 rotating disk can be easily constructed from bristol board and 

 mounted for the purpose of calibration, which is readily made; 

 or a pair of mounted nicols may serve the purpose equally 

 well. If these simple means are not at hand, the distance of 

 the radiant from a frosted glass before the unilateral slit may 

 be used *. 



* The firm of Schmidt and Haeusch of Berlin will be able to supply 

 this instrument with some modifications for about 500 marks. They are 

 also able to furnish one of their larger spectrometers fitted with an 

 additional collimator, which makes a spectrophotometer superior to the 



