28 INFRA-RED ABSORPTION SPECTRA. 



located as accurately as the aforesaid lines, by using the large spec- 

 trometer. This estimation is based upon the fact that the prism has 

 been remounted and reset for minimum deviation so often, upon two 

 different spectrometers, with the adjustments made in three different 

 ways, that the consistency in the location of this band can not be attrib- 

 uted to mere coincidence. If one were to set up similar apparatus, and 

 repeat the work, the other absorption bands would be found accurate to 

 about 0.02 fx. This is a liberal allowance for errors in computing, draw- 

 ing, and reading wave-lengths from the calibration curve, in adjusting 

 the apparatus, in drawing the curves, etc. As an example of the diffi- 

 culty in keeping the apparatus in adjustment for a long period, the 

 experiences of Donath^ may be cited, who found his apparatus, which 

 was of the finest construction, out of adjustment by the time he had 

 examined half a dozen compounds. In the present work the adjust- 

 ment was tested, by means of the sodium lines, at the beginning and 

 conclusion of each series of measurements. 



A source of error in determining the transmission is the possibility 

 of the source of radiation varying in intensity while reading the deflec- 

 tion of the vane when the absorption cell is before the slit, and the 

 deflection for the direct radiation. This is likely to occur with a gas 

 flame, but did not occur with the " heater " used on a storage-battery 

 circuit. 



The rigidity of the mounting of the cell in its carrier, which moved 

 to a fixed height in a plane parallel to that of the spectrometer slit, in 

 accurately fitting wooden ways upon the spectrometer arm, eliminated 

 errors due to variation in thickness of cell. 



A great deal of time was spent on the question of the use of a prism 

 kept at minimum and at non-minimum deviation. As is well known 

 the minimum of deviation is not sharp, well defined ; hence any vari- 

 ation from the real minimum in the setting for the sodium lines will be 

 magnified far out in the infra-red when using the non-minimum devi- 

 ation method. This appears to be the chief objection in this work. 

 Since in the Wadsworth minimum deviation method the prism-mirror 

 table is turned through only half the required angle, thus doubling any 

 error in making a spectrometer setting, the writer preferred the con- 

 stant emergence method, which requires the collimating telescope to 

 be turned through the whole angle. In this manner a smaller power 

 microscope could be used in making the spectrometer setting, the circle 

 readings could be made more quickly, while there was less likelihood 

 of losing one's place in the spectrum, which easily results in this work, 



^Donath, loc. cit. 



