94: Prof. B. Hopkinson on the 



reason in these particular experiments why the wire should not 

 have been —^ inch diameter instead o£ ^ inch. The larger 

 size was chosen because it was intended ultimately to use it for 

 measuring the suction temperatures when the engine worked 

 in the ordinary way, taking in and firing a charge of gas. 

 But it was found that even this large wire always fused 

 before any observations could be taken. A still thicker 

 wire might, of course, have been used for the purpose, but the 

 correction would then be so great as to make the results 

 valueless. It seems clear that in an engine of this size and 

 compression ratio, it is a matter of much difficulty to get 

 accurate measurements of temperature by the platinum 

 thermometer, when the engine is firing. The temperature in 

 this engine rises in places to 2200° C. after the explosion, 

 and, probably, remains above 1710° C. (the melting-point of 

 platinum) for ^ second. It is therefore not surprising that 

 «ven fairly thick wires are melted. 



This paper may conclude with a short discussion of the 

 value of X, or the rate at which the wire loses heat per degree 

 difference of temperature between it and the gas. This 

 quantity is plotted in fig. 4 in terms of crank-angle. At the 

 top of the compression X is three times as great as at the 

 latter end of the exhaust. This is, no doubt, mainly due to 

 the fact that the thermal conductivity of the air increases 

 with the temperature. The increased pressure also probably 

 has some influence, since we have here to do with air in 

 turbulent motion, and X is not a function of thermal con- 

 ductivity only. During suction X is distinctly greater than 

 during exhaust, probably because of the more violent motion 

 of the air in the former operation. 



It has sometimes been said in connexion with attempts to 

 measure gas-temperatures by the platinum thermometer, that 

 the rate at which a fine wire loses heat to a gas per unit 

 length is independent of its diameter. This assertion is based 

 on the assumption that convection currents play an unim- 

 portant part. In that case the rate of loss of heat will involve 

 the logarithm of the radius of the wire and will not change 

 greatly with that radius, provided it be small compared to 

 the dimensions of the vessel in which the wire is situated. In 

 experiments such as those here described, however, convection 

 is exceedingly important , and the rate of loss of heat is, in 

 consequence, considerably greater for a larger wire. I recently 

 tried placing two platinum wires, one j^ inch and the other 

 ^inch diameter, in a long wooden box 6 inches square, 

 along which a current of air with an estimated velocity of 



