1889. 1 on Optical Properties of Oxygen and Ozone. 471 



and tlie absorj^tion of the ultra-violet rays by the latter gas begins at 

 a higher point, namely about X 2450, than that of uncombined 

 oxygen. In fact, we see the anomalies of the selective absorjition by 

 compounds as comjDared with that of their elements when we take the 

 case of water, which has a remarkable transparency for those ultra- 

 violet rays for which oxygen is opaque. 



These observations show that all stellar spectra observed in our 

 atmosphere, irrespective of the specific ultra-violet radiation of each 

 star, must be limited to wave-lengths not less than X 2700, unless we 

 can devise means to eliminate the atmospheric absorption by observa- 

 ,tions at exceedingly high altitudes. 



We have extended our observations to much longer columns of 

 ^oxygen. A steel tube 18 metres long (see Fig. 2) w^as fitted with the 

 same quartz ends as had been used with the shorter tube, and with 

 two quartz lenses symmetrically j)laced inside the tube, one near 



Fig. 2. 



I ^miA 



Jni LENS LENS j c . 



'' S ^Mn.m 



n rziz^^ 



Section of Steel Tube. 



each end, so that when an arc lamp was placed about 14 centimetres 

 from one end of the tube, the image of it was formed on the slit of 

 the spectroscope at the same distance from the other end. 



When the tube was tilled with air only at ordinary pressure, no 

 absorptions could be detected, but when the air was replaced by 

 oxygen at the pressure of the atmosphere the absorption of A was just 

 visible, though neither B nor any other absorption-band could be 

 traced. As the pressure of the oxygen was increased, A became much 

 darker and more distinct, and B came out sharply defined. The 

 absorption-band about A 5785 was next seen, and the dark bands 

 about X 6300 and X 4770, were just visible when the pressure reached 

 20 atmospheres. 



At a pressure of 30 atmospheres A was very black, B also strong 

 and sharply defined, and the forementioned bands were all quite 

 strong and had the same general characters as when seen through the 

 shorter tube ; the band about X 6350 also could be seen, but there 

 was only a bare trace of that in the indigo about X 4470. At 60 

 atmospheres these last two absor23tions could be well seen, all the 

 other bands were very sti ong, B still quite sharp, but A somewhat 

 obscured by a general absorption at the red end. At 90 atmospheres 

 this general absorption at the red end seemed to extend to about one- 

 third of the distance between A and B ; but A could still be seen, 

 w^hen the slit was wide, as a still darker band on a dark red back- 

 ground ; B was still sharp, and the other absorptions all strengthened 



