120 Messrs. J. S. Ames and W. J. Humphreys on the 



fully measured of as many of the lines as possible. In certain 

 cases eye-observations were also made. The results for each 

 element may be thus briefly stated : — 



1. The lines of any one series of a particular element are 

 shifted alike, i. e., according to the same law, which may be 

 written 



A\=\/3(p 1 -p ) ; 



where X. is the wave-length, AX- is the shift produced by an 

 increase of pressure^ —v , /3 is a constant for any one series 

 of a definite element. 



2. The constant /3 is different for the different series of the 

 same element, the change being such that, very nearly, /3 for 

 the principal series is one half for the first subordinate and 

 one quarter that of the second subordinate. 



3. The constant /3 is different for the same series of different 

 elements. Special attention is called to this fact in another 

 note in this Circular. One apparent regularity which demands 

 attention is the fact that, approximately, the value of /3 for 

 similar elements (e. g. zinc, cadmium, mercury) varies as the 

 cube root of the atomic weight. 



No satisfactory theory has been advanced to account for 

 these shifts of the spectrum-lines when the arc is under pres- 

 sure. There is every evidence that it is not due to a tempe- 

 rature effect of any ordinary kind. It would certainly be 

 expected that the outer envelope of an arc would be at a much 

 lower temperature than the core, and that different electric 

 currents might cause different temperatures in the arc ; yet 

 no shift due to these variations has been observed. It is 

 our intention to observe the arc under these conditions with 

 a Michelson refractometer as soon as possible, and so to learn 

 what actually occurs. Again, the temperature of the arc 

 should be much greater than that of a Bunsen-flame which is 

 being fed with sodium ; but the difference in the wave-length 

 of T>i under the two conditions is not perceptible with a 21-foot 

 concave grating, 15,000 lines to the inch, i. e., a shift, if any, 

 must be less than 0'002 of an Angstrom unit. The words 

 " temperature of the arc " are used with considerable hesi- 

 tation, because so little is known as to the mean condition of 

 the molecules of the vapour which are producing the light. 



One can easily understand, however, that if the pressure on 

 a gas is increased, the number of collisions per second must 

 increase ; and it is not impossible that this increased internal 

 energy of a molecule, as thus produced, is the immediate 

 cause of a change in the actual size of the molecule. The 

 extent of this change would depend upon the looseness of con- 



