632 PROCEEDINGS OF THE AMERICAN ACADEMY. 



trough, during the electromagnetic observations ; for in these it is 

 desirable to have all parts at the same temperature, except as differ- 

 ences may be set up by the magnetic action that we are studying. 



The asbestos pad, indicated by a dotted strip in Figure 8, at the 

 back of the plate bears against one face of the channel in the trough. 

 Between the plate and the other face of this channel is a space about 

 0.2 cm. wide, into which passes the single-turn "test-coil," of 1 cm. 

 radius, by means of which the strength of the magnetic field is measured. 

 The test-coil is in circuit with a ballistic galvanometer and the throw 

 produced when it is suddenly withdrawn from the field is easily and 

 accurately measured. The semicircular pieces above and below the 

 plate are of soft iron of nearly the same thickness as the plate, their 

 object being to insure uniformity of magnetic field over the whole width 

 of the iron under observation. Care is taken to have the plate a little 

 nearer to one face of the channel than to the other, so that it may not 

 be pulled forward on the test-coil side by magnetic action. At each 

 end of the trough near the top is a slot which holds a small brass bar, 

 having a head at one end and a nut at the other and carrying a stud 

 0.5 cm. long on its middle. This device prevents the walls of the 

 trough from bending toward or from each other during the magnetic 

 action. 



During thermomagnetic observations also, with a longitudinal grad- 

 ient of temperature, the plate was at first kept in the iron trough, 

 without, of course, circulation of the water therein. It appeared, 

 however, after a time, that the nearness of the walls of the trough 

 affected the longitudinal temperature-gradient, and after May 21, 

 1910, the trough was discarded during thermomagnetic measurements. 

 At this time disks of cork (see Figure 8) about 0.2 cm. thick were 

 cemented to the flat faces of the pole pieces of the magnet, which 

 faces are 4.0 cm. in diameter, and the exposed surface of each disk 

 was cross-cut with a file or saw, in order to reduce the naturally low 

 heat-carrying power of the cork. The asbestos pad at the back of the 

 iron plate was then placed against the cork disk on one pole of the 

 magnet, and the other pole was brought somewhat toward the plate, 

 but not near enough to pull the plate away fi'om the other pole. 

 Most of the thermomagnetic data which we use were obtained after 

 this arrangement had been adopted. 



The cork disks remained in use during all the subsequent measure- 

 ments, even when the iron trough wa^ put back for repetition of 

 electromagnetic observations. Previous to their introduction a facing 

 of asbestos had been used between the pole pieces and the sides of the 

 trough. 



