292 CARNEGIE INSTITUTION OF WASHINGTON. 



They also correspond closely with the dissymmetries produced by the 

 disruptive spark and with the displacements at the pole of the arc. 



(66) Certain resemblances, suggesting a possible indirect relation- 

 ship, are found between the displacing action of the tube-arc and the 

 pressure effect. 



(67) Measures of the Zeeman separation of 363 iron, 186 chromium, 

 and 718 vanadium lines have been made in addition to those previously 

 reported. 



(68) Tabulation and reduction to a standard field have been made 

 for 1,132 iron, 1,089 chromium, and 1,361 vanadium hnes. For most 

 of these lines there are two or more values in good accordance, though 

 some weak lines have been measured on only one plate. 



(69) Between X2979 and X6678, of 1,570 iron arc-Unes examined, 

 286 show displacements to the red and 80 to the violet when the 

 spectrum of light from near the negative pole is compared with that of 

 the center of the arc. At the positive pole the shifts are of the same 

 sign as at the negative, but only about one-fourth as large. 



(70) When hght is taken only from the center of the arc the current 

 may be changed from 5 amperes to 7 amperes without shifting the d 

 lines more than 0.001 a. Lines of group e are somewhat more sensitive. 



(71) The central section of the Pfund arc, carrying 6 amperes, is 

 IDractically free from pole effect. 



(72) Where the highest precision is sought, the slit should be placed 

 at right angles to the arc axis at its midpoint and the current main- 

 tained at a constant value. 



(73) Observations upon symmetrical lines with large pressure shifts 

 do not show a general increase of pressure in passing from the center 

 to the negative pole sufficient to account for the observed displacements. 



(74) The variation of pole effect with wave-length does not follow 

 the same law as pressure displacements. Therefore, an increase of 

 pressure, localized at the pole and in the core of the arc, does not appear 

 sufficient to explain the displacements. The wave-lengths of the sensi- 

 tive lines are not affected by a tenfold change in the density of the 

 iron vapor in the furnace, and are independent of a change in the 

 furnace temperature from 2,100° to 2,600° C. Except in a very few 

 special cases, the effect disappears in vacuo and in so far appears inde- 

 pendent of the fall of potential. Nor does the pole effect appear to be 

 intimately related to the differences in luminosity between the positive 

 and negative poles. 



(75) New values of the pressure shift have been obtained for some 

 typical lines in the iron arc spectrum by a method which practically 

 eUminates the pole effect. For groups a and b the new values accord 

 closely with the best previous work, but for group d the new value is 

 about one-half of the old, while for group e in the green the shift is 

 nearly zero. 



