CONVECTION OF HEAT BY AIR CURRENTS. '45 



result of one action, but of at least three — radiation, convection, and conduction. 

 Leaving the last out of consideration meanwhile, the law of radiation is known to 

 involve higher powers of the temperature excess than the first. The amount of heat 

 carried off by convection may be assumed to be proportional to temperature excess. 

 Consequently, if the amount of heat radiated during unit time at a given temperature 

 excess remain constant and comparatively small, while the amount lost by convection 

 be considerably increased, the total rate of cooling due to both causes will become more 

 nearly proportional to temperature excess. A rough estimate, deduced from results 

 given later on, may be made of the relative amounts of heat lost by radiation and 

 convection. When the copper ball was exposed to an air current of a speed of 1000 

 metres per minute, the rate of cooling, at a temperature excess of 50°, was 2|-° per 

 minute due to radiation, and 7° per minute due to convection. 



This seems to me an explanation of the result quoted above. 



I would suggest that Fokbes' method of determining thermal conductivity might be 

 improved by the application of the above result. The bar experimented upon might 

 serve for both parts — statical and cooling — of the experiment, and in both it might be 

 exposed to a current of air passing across its breadth. The amount of heat lost by 

 cooling during the statical experiment might then be determined with greater exactness 

 than in the usual way. 



7. As regards the rate of cooling at a given excess of temperature with different 

 speeds of the cooling current of air, the results obtained are given in tabular form in the 

 Appendix to this paper, Considering the many minute points in regard to which the 

 experimental conditions might vary, the results are fairly concordant. 



For a temperature excess of 50° the rates of cooling at different speeds of the air 

 current were plotted as ordinates in a curve, the abscissas in which were the different 

 speeds. For 80° and 100° the same was done. The curves are shown in fig. 6. 



A glance at these curves will show that the rate of cooling at a given temperature 

 excess increases with the speed ; at first almost proportionally to the speed, but after- 

 wards more slowly. With regard to this peculiarity I would offer the following remarks 

 by way of possible explanation. 



Were the copper ball exposed to an air current whose speed gradually increased, 

 there will obviously be reached a speed at which the motion of the air will cease to be 

 steady motion ; vortices will be formed, or the motion will become turbulent. It is 

 reasonable to suppose that, until this speed is reached, the rate of cooling at a given 

 excess will be proportional to speed. Beyond that speed, there being less and less 

 steady motion of air past the surface of the ball, the cooling effect of the current will 

 be less than proportional to its increase in speed. In this way the change in curvature 

 of the curve might be explained. I attempted to determine the speed at which the 

 motion of the air ceases to be steady, by allowing sal-ammoniac fumes to be drawn into the 

 tube, gradually increasing the speed of the air current, and observing the direction taken by 



