416 Mr. F. D. Brown on the Maintenance of 



the amount which the tap can open may be regulated. The 

 two ends of the lever are connected by strings at Si S 2 to the 

 loose pulleys P x P 2 of the clutch. These pulleys are made of 

 soft iron, and run on a spindle which revolves on the centres 

 A 1? A 2 . Revolving with the spindle and facing V 1 and P 2 are 

 two small electromagnets M 1? M 2 : one end of the coils of each of 

 these magnets is soldered to the insulated ring I 2 ; the two other 

 ends are soldered to I x and I 3 respectively. The binding-screws 

 Ui, U 2 , U 3 are connected electrically by means of springs with 

 these rings ; U 2 is further connected with the pole N of the 

 battery, while from JJ 1 and U 3 wires run to E 2 and E x respec- 

 tively. The disposition of the wires is shown in fig. 5 : it is 

 there evident that if the mercury in the manometer touches 

 the rod E 1? M 2 will become magnetic ; the loose pulley P 2 

 (fig. 3) will then tend to revolve with it, and the tap will take 

 up the required position a b ; while if the mercury touches E g , 

 M x will become magnetic, P 3 will revolve, and the tap will 

 assume the position a r b f , in which communication is made 

 with the air-pump. By this arrangement, therefore, the pres- 

 sure in the manometer, and in whatever apparatus is connected 

 with it, is caused to rise and fall within very small limits ; with 

 care these limits may be made to differ only about 0*25 millim. ; 

 and thus a practically constant pressure is attained. 



The current required to work the magnets is no more than 

 is furnished by a small Smee's cell; the magnets may be 

 made to revolve with a small turbine ; the air-pump may be 

 replaced by a Bunsen water-pump ; an air-pump, however, 

 is the only convenient apparatus for compressing air. 



It may be objected that the arrangement above described 

 requires motive power, which is not always at hand in a labo- 

 ratory. To meet this objection I endeavoured to construct a 

 double valve to be moved to and fro by two stationary mag- 

 nets ; but I found that magnets of ordinary size were not 

 powerful enough for the work, as the valve, to be of any use, 

 must fit perfectly air-tight, even when subjected to very high 

 pressures. I have not, however, given much attention to this 

 point, an engine which I use for many other purposes being 

 obviously the best source of the necessary power. 



In fig. 6 is given a section of the steam- or vapour-bath which 

 I employ for the comparison of thermometers and for the 

 measurement of the expansion of liquids in dilatometers ; it is 

 also suitable for the direct comparison of thermometers with a 

 standard air-thermometer. The bath is made of brass, and 

 consists merely of a boiler B surmounted by a double tube 

 D ,D similar to those first used by Rudberg and Regnault for 



