104 Prof. A. F. Sundell's Researches 



observed the following changes with change of pressure. 

 With a tolerably wide slit the spectrum could be measured at 

 a pressure of 12 millim. Feeble continuous light began at 

 557. The bands 22 and 23 appeared as a continuous band of 

 mean wave-length 5192. Further, 24, 25, 26, 27, and 28 

 appeared united into one band, which was also the case with 

 29, 30, 32, 33, 34, 36, 37. Nos. 11 and 31 appeared at a 

 pressure of 5'5 millim. ; and all the bands became visible at 

 2*3 millim. When the pressure sank below 0*2 millim. all the 

 bands became decidedly weaker, the feebler bands disappearing 

 first with increasing exhaustion. Thus, at a pressure of 

 0-02 millim. the bands 1 to 9, 21 to 26, 31, 34, 35, and 38 had 

 disappeared, and the bands 11 to 17 had united to form a con- 

 tinuous glow. At 001 millim. there were left only 10, 27, 

 28, 33; and at 0'0023 millim. only 27, 28, and 33: Whilst at 

 0'0013 millim. the band 28 also vanished. If I pushed the 

 exhaustion still further I could see no air-lines, although the 

 tube was still faintly luminous, the mercury-line 546 being 

 quite distinct. No luminosity was produced by single dis- 

 charges ; at a pressure of 0007 millim. the tube only became 

 luminous very seldom, and if the pressure was less than 

 0"0003 millim. there was generally no light to be perceived. 

 With the same velocity of the induction machine the sparks 

 followed each other much more rapidly than at higher pres- 

 sures. We may therefore conclude that with high values the 

 tube no longer acts as a conductor, because the rarefied air is 

 no longer present in sufficient quantity. 



With a very high exhaustion the walls of the tube fluoresced 

 strongly, especially near the positive coating ; and at each 

 discharge this coating emitted a sharp sound, like that of a 

 spark. Some of Crookes's kathode-rays were seen at a pres- 

 sure of 0*002 millim. 



In one case, however, each discharge produced light in the 

 tube, although fully exhausted. The tube had been heated in 

 an air-bath to a temperature of about 250° C. for several hours 

 daily during ten days, in order to disengage the air from the 

 sides of the tube as much as possible. The pressure was re- 

 duced on each occasion to a few millionths of a millimetre by 

 continuous pumping. In the intervals while the tube was not 

 heated, the pressure rose from one day to another by about 

 0*00015 millim. After this preparation I examined the effect 

 of discharges, and found that a tolerably strong luminosity 

 was produced, which was finely stratified throughout the 

 tube. In the spectroscope onl}- the five previously mentioned 

 mercury-lines 579, 546, 492, 483, 436 were seen. Since pure 

 air generally shows only the mercury-line 546 in high vacua, 



