August 30, 1919] 



SCIENCE 



215 



0"^ 10"" iO"^ 30^' 40] 



(?\ :^ 



^ ■§ /{^ k S^ 



^ f ^ i ^ ^ 



m, Anm. 



0"^ 10"^ ^0'^ 30^ 40 



10'^ ZO'^~i>0^ 40'' 

 S%hMUimi,40cm/. 



Fig. 1. 



In Fig. 2, made under the same circmn- 

 stances on a different day, the phenomenon is 

 totally changed. The needle turns in the 

 middle of the interval between the turning 

 times of the pulling weight. There is no 

 phase difference, while the drift has been ac- 

 centuated. The triplets are larger, though the 

 motion within the branches is still uniform. 

 Similar observations were obtained with a 

 downward drift. So I might adduce examples 

 with all kinds of phase-differences, in some 

 of which what was called " inertia " in relation 

 to rig. 1, comes just iefore the reversal of the 

 pull! For the same reason alternations in 

 periods of one minute each rarely succeed. 



It is obvious therefore, that in addition to 

 the gravitational attraction there is in all 

 these cases evidence of the development of an 

 attractive (or in Fig. 2, of a repulsive) force 

 more or less rapidly after the weight is turned. 



3. Radiation, — The extraneous forces origi- 

 nating in M are clearly referable to radiation. 

 We may argue plausibly that, if M be warmer 

 than m, there is excess of convection on the M 

 side and a corresponding part of the pressure 

 is converted into kinetic energy. Attraction 

 apparently results. In the opposite case (colder 

 M), there is repulsion such as is evidenced, 

 for instance, after the semi-periods in Fig. 2. 

 The relative magnitude of the radiation forces 

 is astonishing. One has merely to warm the 

 ball M with the hands, in order to increase the 

 " gravitational attraction " iive or ten times. 

 Again on cooling the ball in tap water only a 

 few degrees below that of the room, repulsion 

 may be obtained. Thus when the external 



temperatures are increasing even if very 

 slowly, outside objects like M are hotter and 

 the excursions of § 1 are large ; and vice versa. 



The warmer ball remains effectively though 

 deoreasingly so, for hours, even when it has 

 become cold to the touch. Normal experi- 

 ments are not again feasible imtil the day 

 after. 



4. Radiation in Vacuo. — At this point it was 

 therefore necessary to build another apparatus, 

 capable of being exhausted. This was done, 

 and experiments similar to the last performed, 

 by exhausting the interior in successive steps 

 of 0-10 cm., 10-20 cm., etc. Thus again the 

 interior was cooled relative to the exterior and 

 there was an influx of radiation, the character 

 of which was made evident by hanging the 

 needle somewhat obliquely to the vertical walls 

 of the case.^ The ball M was discarded. It 

 was found that the attractive forces obtained 

 in each of these successive steps of exhaustion 

 (allowing the needle to get back to equilibrium 

 before the next step) gradually diminished 

 with the decrease of pressure, until between 

 60 and 70 cm., there was no appreciable effect. 

 For higher exhaustions (70-74 cm.) the at- 

 tractive forces were reversed and became 

 strong repulsive forces.^ In other words at 

 this point the radiometer forces supervene 



3 1 give tMa explanation witli some reservations. 

 All that is in question is a reversible inequality of 

 radiation on the two sides. 



4 Deflections of +15 cm. and - — ^15 cm. were 

 observed, respectively, at the first and last drop of 

 pressure, whereas the gravitational deflection is 

 but 3 or 4 cm. 



