104 



as tliey were before determined b_v tlie au(lior') these measurements 

 gave two values lor tlie reduclion factor, wliicli agreed down to 

 0.1 7„. Besides, some measurements with carbon disulphide liave 

 been made, wliich sufficiently harmonized with those of water. 

 Carbon disulphide is less satisfactory for a measurement of the 

 n)agnetic field tiian water. The angles of rotation are then, indeed, 

 larger, but the greater temperature coefficient renders a greater 

 accuracy in the temperature determinations necessary, in consequence 

 of which an accurate result is after all more difficult to reach. 



Further a numlier of measurements have l)een made with the 

 titanium chloride at different temperatures l)y cooling the water at 

 its entrance into the jacket by means of ice, or by raising the tem- 

 perature. The temperature coefficient of the constant of rotation 

 appeared, however, to be so small, that it could not be determined 

 with certainty from the observations. The thought of applying a 

 temperature correction has, therefore, been aimiidoned. In the mea- 

 surements with the quartz amalgam lamp the temperature was on 

 an average 17.9^, in those with the arclamp 13.4°. 



The results of the measurements with the quartz lamp are recorded 

 in the last column of the above table; those with the arc lamp 

 referring to observations on six different days, follow below. 



The more recent theories of the magneto-optical phenomena which 

 are founded on the theory of electrons, show that there is a connec- 

 tion between the magnetic rotation of the plane of polarisation and 

 tlie Zeeman effect, which the lines of the substance's free vibrations 

 present. By starting from simplifying suppositions, and assuming one 

 free period, the magnetic rotation as far as the sign and the order 

 of magnitude is concerned, can in many cases be explained by the 

 assumption of a magnetic resolution of spectrum lines as it is given 

 by the elementary theory '). For a more complete explanation it is, 

 however, necessary to take more free vibrations into consideration. 



It has appeared from investigations l)y Drude and others that 

 the ordinary dispersion of transparent substances can generally be 

 represented l>y an expression with a small number of free vibrations, 

 among which ultrared ones, corresponding to vibrations of positively 

 charged particles, and one or min-e ultraviolet free vibrations of 

 negative particles. The ultraviolet frequencies cause the greater 

 part of the dispersion. 



') Versl. Kon. Ak. van Wet. 1896/97 p. 131. tJomm. Leiden Suppl. 1. p. 76 

 Aicli. Néei-l. (2) 2 p. 366 (1899); (2) 6 p. 825 (1901). 

 ■•) These Proe, Vol. 5, p. 413, Gomm- Leiden ;»J"- 82. 



