LUMINOUS EFFECTS OF THE DISCHARGE. 637 



length of the mercurial column and the partial vacuum 

 above it, even if the latter is 20 cm long or more. The 

 spark in the rarefied air assumes then the form of a 

 broad ribbon-like flash having a green colour. The 

 colour is caused by the trace of vapour of mercury 

 which always exists in the Toricelliaii vacuum. 



For experiments of this kind, wires of platinum are generally 

 passed through the sides of glass tubes and fused into them so as to 

 be firm and air-tight, but the operation requires considerable skill. 

 For our present purpose it is sufficient to fix an iron wire with 

 sealing-wax into one end of a tube, which is open at both ends, 

 so as to close it air-tight. The tube should be l m long and at least 

 4 mm wide ; it should be taken wider if sufficient mercury is at the 

 student's disposal. Both ends are first heated until the sharp edges 

 are somewhat rounded off; then the tube is warmed through the 

 whole length and air sucked through it, in order to remove moisture 

 from the interior. An iron wire, 1 or 2 mm thick and 8 cm long, is 

 bent at one end into a ring ; the middle of the straight portion is 

 heated until sealing-wax melts on it, and after it has become some- 

 what cooler it is surrounded by a layer of sealing-wax so that it forms 

 a cylinder a little thicker than the bore of the tube. When the seal- 

 ing-wax is quite cold the end of the tube is heated, and the wire with 

 the coating of sealing-wax pushed into it, so as to close that end of 

 the tube completely. In heating the glass-tube it should be held some- 

 what slanting and the open end upwards, otherwise moisture from 

 the flame will enter the tube, which cannot be easily removed after- 

 wards. Nor should the end be heated more than just sufficient to 

 melt the wax ; if made too hot it is apt to crack, and besides, the 

 sealing-wax froths up into little bubbles when in contact with the 

 over-heated glass, becomes too liquid, and runs down inside the tube 

 along the lower portion of the iron wire, which should be left free. 



The tube thus prepared is filled like a barometer, as explained 

 previously (see page 234). The air must be swept out as com- 

 pletely as possible, by allowing an air- bubble, about 2 cm long, to run 

 to and fro along the tube ; small quantities of air will still remain 

 behind which will render the vacuum imperfect. 



The tube being filled and the open end immersed in mercury, it 

 is fixed vertically in the retort- stand. Into another re tort- stand an 

 iron wire, 10 or 20 cm long, is clamped vertically, so that one end of it 

 dips into the mercury of the cistern. To the other end, which is 



