KENT. — THE SPECTRA OF TITANIUM, ZINC, AND IRON. 289 



show no marked or consistent difference from those obtained in the 

 ordinary manner. Care was taken that the spark exposure be very 

 short. 



4. Introducing the self-induction coil into the discharge circuit of the 

 condenser (set 4) unquestionably conduces to coincidence of arc and spark 

 line (mean value of shift 0.005 t.-m.). 



5. Sets 5 and 6 show that the use of the alloy, 20 per cent titanium, 

 80 per ceut iron, give much less displacement (0.002 t.-m.) than that 

 given by titanium carbide, 85 per cent titanium, 15 per cent carbon 

 (about 0.02 t.-m.). 



6. The part of the spark image used is extremely important ; set 7, 

 given by that region of the image near the terminal, shows a shift of 0.018 

 t.-m., while the central portion of the spark (set 8) 4.5 mm. from the 

 terminal shows but about 0.006 t.-m. 



7. Sets 9 and 10 show that ohmic resistance (2.7 ohms) in the dis- 

 charge circuit of the condenser produces an effect exactly similar to that 

 of self-induction (see result No. 4), reducing the shift from about 0.017 

 t.-m. to practically zero. 



8. Sets 7 to 10, in which no fan was used, show that the same is not 

 a sine qua non in the matter of shift. 



9. Further, the developer used appears to have no effect upon the 

 shift. Therefore in sets 5, 7, 8, and 10 the measurements made on plates 

 developed with the Jewell solution are averaged in with the others. 

 Indeed it is difficult to see how there could be a difference, provided the 

 tip of the line be used in measurement. If, however, the setting be made 

 upon the body of the line a difference might very easily be introduced 

 (see below, result No. 16), a "softer" developer, such as Wallace's, 

 tending to displace the apparent centre of gravity of the line toward the 

 side of greater broadening. 



10. Sets 11 and 12, also 11 and 13, Table IV, show that the alloy and 

 self-induction effects of sets 6 and 7, also 3 and 4, Table III, respectively, 

 hold good with lines of the spectrum other than AA 3900 and 3913. 



11. Set 14 shows that large shifts exist in lines in other regions, e. g., 

 A 4827.74, 0.024 t.-m. ; A 4836.25, 0.036 t.-m. 



It is also noticeable that whenever both spark and arc lines are 

 reversed the central portions of the reversal are coincident. 

 Passing to zinc, Table V, we have : 



12. Sets 15, 16, and 17 give results comparable with those of 3, l,and 

 2 for titanium, and sets 18 and 19 confirm the alloy effect of 6 and 7 

 and also 1 1 and 1 2. 



VOL. XLI. — 19 



