JBumstead — Heating Ejects produced by Rontgen Rays. 3 



them upon one side ; when one of the vanes is illuminated, it 

 is heated, and the molecular reaction causes it to be repelled 

 from the neighboring wall. But it is quite possible to reverse 

 this procedure and have the wall opaque and the vanes trans- 

 parent, and, although the attainable sensitiveness is probably 

 less than in the other case, it has obvious advantages when one 

 is dealing with Rontgen rays instead of ordinary light. The 

 walls can be made of the metals under investigation, and of suit- 

 able thickness to absorb a considerable fraction of the rays inci- 

 dent upon them ; while the vanes may be made very transpar- 

 ent to the rays and be thus far less in the way, and be less 

 heated by the rays (independently of the heating of the sub- 

 stances under examination) than would be possible with the 

 thermopile or bolometer. In the construction of the radiometer 

 and its gradual adaptation to the present purpose, I was particu- 

 larly fortunate in having the advice and assistance of Professor 

 E. F. Nichols, who was also in Cambridge, and who most kindly 

 put at my disposal the results of his long experience with the 

 radiometer. I wish here to express my thanks to him for a 

 much shorter period of apprenticeship to the instrument than 

 would have been possible without his help. 



Apparatus. 



The radiometer and its adjuncts passed through many pre- 

 liminary and tentative forms before a satisfactory arrangement 

 was obtained. The final form, which proved fairly sensitive 

 and manageable, is here described. 



The vanes were made of aluminium foil which weighed 

 about l mg per sq. centimeter. Each vane measured 8x10 milli- 

 meters with its greatest dimension vertical, and the two pieces 

 of foil were stretched between two very thin horizontal rods 

 of glass at top and bottom, which in turn were kept at the 

 proper distance apart by their attachments to the central rod 

 of the suspended system. The inner edges of the vanes were 

 4 mm apart. The various joints were made by very small drops 

 of alcoholic shellac, which were baked on with a hot glass rod. 

 The system was put together on a flat brass table and, after 

 a few trials, it was not difficult to get one in which the alu- 

 minium vanes were very fairly smooth, plane, and parallel to the 

 central rod. The mirror, for use with telescope and scale, was 

 attached to the central rod 3 cm above the middle of the vanes ; 

 it was usually about 5 mm square and was made of specially 

 selected microscope cover glass. All the mirrors used gave 

 very satisfactory definition. In the final system the weight 

 of the mirror was 13 milligrams, that of the rest of the sus- 

 pended system 5 milligrams ; the moment of inertia was 



