KENT. — THE SPECTRA OF TITANIUM, ZINC, AND IRON. 287 



added an excess of bromide (roughly 0.5 gram of potassium bromide 

 to a 3 oz. solution). 



Results. 



The results are contained in the following tables, concerning the data 

 given in which a few general remarks may be made: — 



The time of the two arc exposures varied from 10 to 60 seconds, the 

 short exposures predominating. 



The length of the slit, unless otherwise stated, was either 5 or 6 mm., 

 its width between 0.025 and 0.050 mm. 



The arc spectrum was generally thrown on the centre of the plate. 



The values of current, power, and voltage given in the tables are the 

 means of from two to six (average three) readings taken at fairly regular 

 intervals during the exposures. The conditions of the circuit were quite 

 stable, the three elements seldom varying more than 1 ampere, 20 watts, 

 and 0.3 volts respectively. 



The metals used were: Titanium carbide or "cast titanium" (made 

 by Eimer & Amend of New York), 85 per cent titanium, 15 per cent 

 carbon ; a titanium-iron alloy, 20 per cent titanium, 80 per cent iron 

 (Eimer & Amend) ; zinc battery rods; brass wire, 30.2 percent zinc, 

 69.8 per cent copper ; ordinary iron rod. 



Unless otherwise stated the part of the spark image used was that near 

 one terminal and the fan was employed to steady the discharge. 



The secondary spark took place between brass rods 4 mm. in diameter. 



The focus, to an accuracy of 1 mm., was determined repeatedly from 

 week to week as the temperature of the room changed. This was a 

 precautionary measure : no change in focus was noticed. 



The given values of the shifts represent generally but one set of meas- 

 urements on each plate, never more than two. This course was pursued 

 because the different plates agreed so well that numerous measurements 

 upon the same plate were deemed unnecessary. The average deviation 

 from the mean of the measurements on the different plates in all the sets 

 was about 0.0027 tenth-meters for titanium and iron, and 0.0070 for 

 zinc. The average deviation from the mean of different measurements 

 of a line on any one plate was about 0.0025 tenth-meters for titanium, 

 0.0018 for iron, and 0.00G0 for zinc. 



To gain in another manner an idea of the degree of accuracy of 

 measurement, four settings, two red-right and two red-left, were made * 



* A suggestion due to Professor Frost. 



