94 PROCEEDINGS OF THE AMERICAN ACADEMY. 



eluded that a 'high E. M. F., rapidly changing, is a probable conditio 

 sine qua non for the appearance of spark lines in arc spectra.' Both 

 might better have expressed their results in terms of potential gradi- 

 ent." * * * "The lowest gradients are obtained in heavy current 

 arcs and Pliicker tubes with wide capillary; in the former case the 

 low gradient is due to the heavy current, in the latter to low gas>- 

 pressure. Higher potential gradients are obtained in arcs with very 

 small current, Pliicker tubes with fine capillaries and sparks with 

 small capacity and large inductance. The highest potential-gradients 

 are found in sparks and other interrupted arcs, the gradient increasing 

 with the amount of capacity in circuit and with the impressed voltage. 

 Gradients vary from about 20 to SO volts per cm. in ordinary arcs and 

 tubes up to thousands of volts per cm. in condensed sparks." * * * 

 "Inductance reduces the gradient down to a minimum, beyond 

 which it is inoperative." * * * "In the condensed spark without 

 inductance, the front of the pilot discharge must have a potential- 

 gradient not much below the dielectric strength of the intervening 

 gas. The remainder of the discharge is probably at a very low 

 gradient, approaching that of a direct-current arc. Hence such a 

 spark gives both spark and arc lines. Inductance and resistance 

 lower maximum gradients by smoothing out the current wave. The 

 spectrum of a spark rendered dead beat by series resistance can 

 scarcely be distinguished from that of a low direct-current arc." 



In 1909 Janicki ^* writes on the structure of spectrum lines, giving 

 the results of a study made with the Lummer-Gehrcke plate, the source 

 being an arc at low pressure (0.1 mm. or less) in a special form of 

 apparatus having an anode of the desired metal. 



The three zinc lines in the blue are described as sharp and simple. 

 They appeared at 0.3 amp., were good at 0.4 amp., and at more than 

 0.7 amp. were reversed in part. 



In certain calcium lines the change of position of their satellites 

 with increase of current is noted, and attention called to an unsym- 

 metrical broadening and reversal. Somewhat later reference is 

 made to the^work of Exner and Haschek on the displacement of 

 spark lines. 



" They traced these displacements, directed mostly toward longer 

 wave-lengths, to the different density of the metallic vapor. With 

 good reason Eder and Valenta objected that these displacements were 

 only apparent. * * * They photographed arc and spark lines im- 



14 Annalen der Physik — Band 29 (1909). 



