September 19, 1919] 



SCIENCE 



281 



one can thus predict what occurred in the 

 outer world several days ago — too bad it is not 

 the other way ! But these successive isotherms 

 as they creep through the wall are not given 

 to prophecy. Like our politicians, they have 

 developed a tactile sense and prefer to feel 

 their way. 



If we turn again to Fig. 1, it will be noticed 

 that the lower readings for which the ball M 

 is hehind the case and thus screened from the 

 wall are quite variable, though not usually 

 as much so as the upper readings, in which 

 nothing intervenes between ball and wall. It 

 is thus obvious that the ball parts with what 

 it has absorbed with extreme reluctance. I 

 have therefore been tempted to ask whether 

 with the thermal radiation something else may 

 not have passed which finds in the lead ball so 

 tenacious a receptacle. 



2. Efflux of Air. — Let me now pass to a 

 different class of experiments given in Figs. 

 3, 4, 5, 6, in which the ordinates are again 

 scale readings, y, but the periods between 

 them only about 3 minutes. Apparatus 77 

 with an airtight case is here used and it is 

 exhausted in steps of 10 cm. of mercury each, 

 very slowly, but otherwise much after the 

 manner of treating the fog-chamber. Fast ex- 

 haustion would throw the readings out of scale 

 and there might be interference from air cur- 

 rents. For this reason pressure increments 

 from infiux are here not very trustworthy. 



The effect of this efflux is to slightly cool 

 the inside of the case. Thus the ball M, or 

 the environment, is temporarily warmer than 

 the needle m, or the inside. In Fig. 3 (note 

 the small scale) the ball M is discarded and the 

 exhaustions, 0-10, 10-20, . . . 70-Y2, 72-74 cm., 

 are marked as to the large numbers on the 

 curve. The needle in these experiments was 

 somewhat oblique to the ease, the right end 

 being nearer the front window and the left 

 end near the rear. A relatively warmer 

 window if attracting throws the reading into 

 larger numbers. Exhaustion is made when 

 the needle is in an equilibrium position, as 

 indicated by the little circles on the curve. 

 Fig. 3 shows therefore, that the effect of ex- 



haustion is actually a temporary radiant 

 pressure increment on the cold side of the 

 needle, the amount of which gradually dimin- 

 ishes until the exhaustion from 60 to 70 cm. 

 (pressures 16 to 6 cm.) has been passed. After 

 this the effect changes sign and the temporary 

 increments now increase with great rapidity 

 on the warm side of the needle. A few trials 

 were also made for influx (74-70, 70-60 cm., 

 etc.), in which a total reversal is in evidence, 

 not very smooth, because of the experimental 

 difficulties mentioned. 



The ball M was then replaced and put on 

 the far side, so that its attractions (gravity, 

 warmth) would counteract the attraction of 

 the plate glass window. The result of this 

 change of environment is given in Fig. 4. 

 The window (less in mass but nearer) first 

 acts, as before; thereafter the ball M (further 

 but with much more mass) produces an op- 

 posed effect. The branches are doubly in- 

 flected and the zero points (circles) lower than 

 before, owing to gravitational attraction. 

 Finally the changes of sign of the radiant 

 forces occur as in Fig. 3; but at the high 

 vacua (70-74 cm.), the window effect is ab- 

 sent. Here also the needle begins to show 

 vibration. 



The attempt was now made to hang the 

 needle symmetrically to the case, or to the 

 environment, in the absence of the ball M, by 

 twisting the torsion head slightly. Fig. 5 

 shows that the apparatus is over-corrected. 

 The plenum radiant forces are negative, the 

 vacuum force positive, the change of sign 

 occurs in lower exhaustions (higher pressures) 

 than before, and the radiometer forces are 

 relatively less important. 



In the case of Fig. 6 the ball M is restored 

 in a position to counteract the radiant force 

 of Fig 5. Hence Fig. 6 is a reversal of Fig. 

 4, just as Fig. 5 is of Fig. 3, with the differ- 

 ences just stated. 



3. Infiux. — In conclusion I will adduce an 

 example of a third group of experiments, 

 made in the manner stated in relation to Figs. 

 1 and 2. In the new experiments the ease was 

 exhausted to different pressures, p (cms. of 

 mercury), but left with a slight leak of about 



