Chemistry and Physics. 557 



and thus gave rise to the series lines of the gas in the tube 

 (helium, hydrogen, etc.). By this arrangement negatively charged 

 particles passed out of the electrostatic field through the perfora- 

 tions in the usual cathode. A suitably placed plane window was 

 affixed to the side of the tube with its normal at right angles to 

 the common axis of the condenser plates (the pair of cathodes). 

 In other words, the line of sight was perpendicular to the axis of 

 the electrostatic field and hence to the paths of the positive ions. 

 One typical illustration of the phenomena observed must suffice. 

 The line Hy was separated into seven components. Three of these 

 were found to be polarized (electric vector) at right angles to the 

 electric field and to have intermediate intensity. The central com- 

 ponent of this triplet was very slightly displaced from the position 

 of the unresolved line (electrostatic field unexcited). Outside of 

 the triplet two stronger components of the original line were ob- 

 served. Also between each pair of lines of the triplet two very 

 faint lines were discerned. The six lines were equally spaced, in 

 pairs, with respect to the central component of the triplet. The 

 four lines not members of the triplet were found to be polarized 

 parallel to the direction of the impressed electrostatic field. 



Longitudinal effect. In order to have the line of sight coinci- 

 dent with the axis of the electric field the vacuum tube had to be 

 remodeled as follows. Instead of a group of perforations in the 

 usual cathode a single slit was made in the geometrical axis of 

 revolution of the disc. The condenser plates of the electrostatic 

 field were fastened with their plane faces parallel to the axis of 

 the slit. The continuation of this axis passed half-way between 

 the condenser plates. The auxiliary anode was pierced by holes 

 so as to enable the light to reach the slit of the dispersing appara- 

 tus. It was thus possible to look along the direction of the elec- 

 tric field but at right angles to the paths of the canal rays. For 

 reasons which cannot be given in this place, it was found correct 

 to perforate the auxiliary anode but not the associated cathode. 

 Special experiments and considerations showed that all sources of 

 error, such as the curvature of the field, Doppler effect, etc., were 

 negligible under the experimental conditions. Reverting to the 

 line Hy it turned out that the original line was resolved into 

 a triplet whose components coincided in wave-lengths with the 

 triplet observed in the transverse field. In all cases, the lines 

 studied in longitudinal fields were unpolarized. This result is con- 

 sistent with the theory worked out some time ago by Voigt. 



Gradient effect. By varying the distances between the con- 

 denser plates and also the applied potential difference, it was a 

 relatively simple matter to investigate the dependence of the sepa- 

 ration of the outer components of any resolved line on the inten- 

 sity of the electric field. For the hydrogen lines A 4341 and 

 X 4861, and for the helium line X 4026, a linear dependence was 

 found both for the vibrations parallel and perpendicular to the 

 field. The separation of the outer components of H/3 amounted 

 to as much as 14*5 A. U. in afield of 50,000 volts per cm. On the 



