392 Prof. Thomson, On some experiments on the [May 10, 



in those places where the field is not uniform and difficult to 

 calculate. The surfaces of the plates were worked very true, and 

 some small holes that were left in from the casting were filled up 

 with putty and then coated with gold leaf. The surfaces were so 

 true that though the electrodes were of considerable weight, yet 

 if they were placed in contact they adhered sufficiently to cause 

 the under one to be lifted when the upper one was raised. The 

 plates were maintained at the same distance apart by means of 

 three glass distance-pieces, two of which are shewn at AE and 

 DH, carefully made of the same length and their ends accurately 

 ground ; these were connected together by pieces of glass rod : 

 these distance-pieces were placed in the hollow part of the plates 

 so as to be out of the way of the discharge ; the plates were 

 placed in a box, LMNP, the side of which was a cylindrical piece 

 of glass and the ends of it brass discs, fastened to the glass with 

 marine glue ; into the upper one of these plates a piece of brass 

 tubing, R, was soldered in order to permit of the exhaustion of the 

 gas in the vessel ; between the top of the box and the upper plate 

 there was a spring, Q, which put the two into electrical connection. 

 The spark was produced by means of an induction coil. 



The following are the phenomena which occur as the air is 

 gradually exhausted from the box. At the pressure of the atmo- 

 sphere the spark passes between two points, being evidently de- 

 termined by some accident which makes the force a little greater 

 at one place than another; at this stage the discharge is very 

 unsteady and skips about from one point of the plates to another : 

 as the pressure diminishes the discharge gradually settles down 

 and remains at one place, and begins to present peculiar features 

 which are represented in the accompanying figures, in all of which 

 the negative electrode is supposed to be at the top. Fig. 2 re- 

 Fig. 2. Fig. 3. Fig. 4. 



presents the discharge when the pressure is that due to about 

 90 mm. of mercury ; it is shaped something like an Indian club 

 with the handle at the negative electrode. As the pressure 

 diminishes, the neck of the club lengthens, the lower part 

 broadens out, and a disc appears at the negative electrode ; the 

 appearance at a pressure of about 40 mm. of mercury is repre- 

 sented in fig. 3; the discharge being bluish near the negative 

 electrode, but reddish towards the positive. As the pressure 



