from Long Vacuum Tubes, 731 



secon iary spectrum in the central portion of the tube 

 requires the presence of traces of: water-vapour. The proper 

 diameter and length of the capillary can be found best 

 by trial. It a short piece of tube about 1 mm. in diameter 

 is melted down at the centre until the walls almost join, and 

 then quickly drawn out to a length of 80 cm., it will be of 

 the right order of magnitude. Having about the thickness 

 and flexibility of. a horse-hair, it can be inserted between the 

 tube and generator by means of sealing-wax. A better 

 plan, perhaps, is to draw the tube out to a length of, say, 

 "1*5 metres, and starting with the full length, gradually 

 reduce it until the proper dimension is found. 



The current is supplied from a small 25,000 volt trans- 

 former, operated by a 60-cycle 110-volt current, as much as 

 30 amperes being fed to the primary on some occasions. 



In joining the generator to the tube by the capillary 

 " all-glass " seals are preferable, as sealing-wax is apt to 

 liberate traces of vapour which introduce the cyanogen 

 bands. The generator (see fig. 1) is filled with a dilute 

 solution of phosphorus pentoxide, and is kept permanently 

 connected with a 12-volt storage battery, which is auto- 

 matically cut oat when the liquid falls below the electrode. 



When starting the apparatus for the first time, the current 

 should be turned into the generator before the liquid rises 

 too near the top. If it continues to rise, a higher voltage 

 must be applied. As soon as one is sure that the hydrogen 

 can be supplied more rapidly than it flows out through the 

 capillary, the battery can be disconnected until the liquid 

 has risen quite up to the small tube at the top ; in this way 

 the air in the generator is most rapidly removed. It is well 

 to begin with a capillary which will give a Orookes dark- 

 space of about 5 mm., with the stop-cock to the pump wide 

 open ; by partially closing the cock a narrower dark-space 

 can be held continuously. 



It is best to aim. the tube very carefully at the spectroscope 

 before joining it to the pump. It should be at a distance 

 of at least 3 metres from the slit, for reasons given in the 

 earlier paper, and is best aimed by mounting a small tungsten 

 automobile lamp in front of the collimating lens, opening 

 the slit wide and bringing the end-on portion of the tube 

 into the line of the emergent beam, the final exact aim being 

 secured by looking through the tube at the illuminated slit. 



In operating the tube, a small direct-vision prism is in 

 constant use for observing the central portion and deter- 

 mining the spectroscopic character of the light. 



In actual work it is preferable to have the tube opened 



3 C2 



