KENT. — THE SPECTRA OF TITANIUM, ZINC, AND IRON. 299 



elapses before the metallic particles, advancing from the terminal, reach 

 the central region of the spark. When they have finally reached this 

 region the air has yielded and the pressure fallen. An unusual amount 

 of vapor is disengaged in the discharge between electrodes of titanium 

 and zinc in water (not so prominently the case with iron). That the 

 pressure developed is also great was noticed by the writer in his work 

 with Professor Hale on the spark under water.* The terminals were 

 badly shattered ; indeed there is good reason to expect that the lines of 

 the spectrum given by the discharge in air pass through a series of 

 changes which resembles that in water. Certain titanium lines appear 

 reversed on some of the present series of plates and not so on others. 

 The reversal comes up unsymmetrically on the violet edge of the centre of 

 gravity of the line, a common occurrence in the spectra of the spark 

 under water. The question of the exact parallelism of the two phe- 

 nomena must be reserved for future investigation. 



Possible Explanation of the Inconsistent Results of 

 Haschkk, Eder and Valenta, and Middlekauff. 



Haschek undoubtedly dealt with over-exposed lines and increased the 

 true shift by a photographic displacement of the region of maximum 

 intensity. 



Eder and Valenta probably used a vertical spark and set the central 

 region of the same on the slit of the spectrometer (the lines studied by 

 the writer, Zn AA 4680, 4722, and 4810, are among those investigated 

 by Eder and Valenta). "We are assuming that the same effect obtaius 

 in the case of zinc as with titanium (sets 7 and 8, deduction No. 6 above) 

 with reference to the central and end portions of the spark. That no 

 shift exists under the conditions employed between the absorption regions 

 of reversed arc and spark lines, as stated by Eder and Valenta for zinc, 

 was found to be true in the case of titanium. 



Lastly, that Middlekauff found no shift with iron may easily be 

 explained by the fact that few lines show a shift, and these generally 

 to but a small degree, under highly disruptive conditions and near the 

 region of the terminal. Middlekauff probably used a comparatively weak 

 spark, set its image vertically on the slit, and integrated with his spec- 

 trometer the spectra of the central regions. f Moreover Middlekauff and 



* Astrophysical Journal, 17, 154 (1903). 



t The statement is made that the spark was eight or nine mm. in length and its 

 image on the slit about G mm. 



