954 Dr. J. S. G. Thomas on Mutual Action of Convection 



resistance of the movable wire. This positive deflexion 

 attains a maximum value as shown at A, thereafter 

 decreasing until a maximum negative deflexion shown at B, 

 is attained, corresponding to an inclination, of 90°, in 

 accordance with results based on the curves in fig. 3. The 

 initial increase in resistance of the movable wire as indicated 

 by the initial positive deflexion, is in this case clearly 

 attributable to the increased effect of a downward convection 

 current, originating in the upward convection current from 

 the central wire. With increasing inclination of the plane 

 of the wires to the horizontal, the movable wire clearly 

 becomes increasingly under the influence of the ascending 

 stream, and the initial increase of resistance is succeeded by 

 a decrease oF resistance of the movable wire*. Tl e form of 

 the curve shows a fair degree of symmetry in the first and 

 second quadrants, but there is marked asymmetry of the 

 calibration curve about the ordinate corresponding to a 

 rotation of the plane of the wires through 180°. Thus the 

 maximum positive deflexion shewn at F, corresponding to 

 270°, is considerably larger than the maximum negative 

 deflexion shown at B, corresponding to 90°. The result is 

 clearly due to a retarding influence exerted by the upper 

 part of the wall of the chamber upon the convecti< n currents, 

 more especially that ascending from the movable wire. 

 With the movable wire disposed in the lower pair of 

 quadrants, the principal thermal effect is experienced by the 

 central wire, and this being situate at some considerable 

 distance from the wall of the chamber, it is to be anticipated 

 that the maximum deflexion under these conditions corre- 

 sponding to a rotation through 270°, would be numerically 

 greater than the corresponding deflexion for a rotation of 

 90° from the initial horizontal position of the plane of the 

 wires, the movable wire being then near the upper wall of 

 the chamber. The portions of the calibration curve shown 

 dotted at D and E in fig. 4 represent unstable values of the 

 galvanometer deflexion, these oscillating about 5 divisions on 

 either side of the mean shown in the figure. The instability 

 is probably due to turbulence arising owing to the wall of 

 the vessel not conforming to the contour of the stream lines 

 ■in its neighbourhood. The asymmetry of the thermal effect 

 experienced by the wires when the movable wire is respec- 

 tively in the upper or lower pair of quadrants, is clearly 

 seen from a comparison of the curves C and D with E and F 

 in fig. 5. In the former pair,, the movable wire was above 



* Similarly, the depression at the maximum of the calibration curve 

 given in a previous paper (Proc. Phys. Soc. he. cit. p. S05) may Le 

 explained. 



