146 DELL SYSTEM TECHNICAL JOURNAL 



doubt this assumption is too extreme, yet it leads to unexpectedly 

 good agreements with experiment. Thus when the thickness of the 

 Crookes dark space is altered (by altering the pressure of the gas) 

 lea\ing the voltage across it constant, the current-density varies in- 

 versely as the square of the thickness, as it should by (16). And when 

 Gunther-Schulze calculated the thickness of the dark space from 

 (10), using the observed values of cathode-fall and current for six 

 gases and two kinds of metal, and substituting the mass of the mole- 

 cule of the gas for the coelTicient m in that equation, the values he 

 obtained agreed fairly well (within 40^0 with the observed thick- 

 nesses. Long befoie, J. J. Thomson had proposed (17), and Aston 

 tested it by a scries of experiments on four gases, in the condition 

 of strong anomalous catho<le-fall. As k of that equation should be 

 inversely proportional to the pressure p of the gas, the product id^V-- 

 {V standing for the cathode-fall) should be constant at constant 

 pressure, and the prfxluct id'V'-p should be constant under all cir- 

 cumstances. These conclu>i<>ns were fairK- well confiniicr! for large 

 current-densities. 



Several attempts to test the theory by actuall\- dcterniiiiing the 

 potential-distribution in the Crookes dark space were made with 

 sounding-wires and by other methods; but the>' have all been super- 

 seded, wherever possible, by the beautiful method founded on the 

 discovery that certain spectrum lines are split into components when 

 the molecule emitting them is floating in an intense electric field, and 

 the separation of the components is proportional to the strength 

 of the field. This was established by Stark who applied a strong 

 controllable electric field to radiating atoms, and by LoSurdo who 

 examined the lines emitted by molecules rushing through the strong 

 field in the Crookes dark space, in the condition of anomalous cathode- 

 fall. Now that the effect has been thoroughly studied it is legitimate 

 to turn the experiments around and use the appearance of the split lines 

 as an index of the field strength in the i)lace where they arc emitted. 

 Brose in Cermanx' and l-'oster at \'ale did this. In the pholograjihs 

 (Fig. 10, 11) we see the components merged together at the top, wiiicii 

 is at the edge of the negati\'e glow, where the field is \-ery small; 

 thence they diverge to a maximum separation, and finally approach 

 one another very slightly Ix-fore reaching the bottom, which is at the 

 cathode surface." This shows that the net space-charge in the Crooke 



" The (lisplarcmcnts nf (crlnin coiniioiicnts arc not rigorously proportional to the 

 fii-ld, and sonictinicsi'ntiri-lv new lines make their appearance at hitherto unoccupiecl 

 places when a sironK fielil i> applied. Holh of these anomalies can lie delected in 

 Ihc pictures. l-"or the original pl.itc from which l-'ig. 11 was made 1 am iiideliled to 

 Ur. l"o8tcr. 



