Measurement of Electromagnetic Radiation. 55 



last is very small compared to the interval between two 

 sparks, it is quite possible, and even probable, that the actual 

 current during the earlier part of each spark greatly exceeded 

 this, but this is a superior limit to the value of the square root 

 of the mean square of the strength of the current in the par- 

 ticular case. 



It should be noted that two errors, each of which is 

 numerically insignificant, have, for convenience, been made. 

 One is the supposition that the forces on the elements near the 

 ends of the resonators are the same as though they were con- 

 tinued together with the stationary wave. As a matter of 

 fact, it is only the last few millimetres for which there is a 

 falling off. The second error is made by supposing that the 

 middle half of the wire is all at one distance from the axis 

 on one side, and the two end quarters at an equal distance 

 from the axis on the other side. As a matter of fact, the wire 

 has to cross over, but as the two forces almost exactly balance 

 on the small part which crosses the axis^ it was not worth 

 while to complicate the simple statement of conditions illus- 

 trated by fig. 4 for the sake of an imaginary degree of 

 accuracy which the nature of the observations would not 

 justify. 



Second Method, 



As the first method, carried out with all the delicacy which 

 under the conditions we could make use of, gave no promise, 

 we determined to find by some direct process whether the heat 

 which Mr. Gregory had inferred from the expansion of the 

 wire was real. For this purpose we made use of one of the 

 most delicate heat-measuring instruments that exist, which is 

 specially suitable for detecting very small quantities of heat 

 diffused over a great length of wire, namely the convection 

 air-thermometer of the late Dr. Joule*. With this instru- 

 ment he was able to detect the direct heating-effect of the 

 moon's rays unconcentrated by lens or mirror. 



His instrument consisted of a tube about 2 feet long and 

 4 inches in diameter, divided longitudinally by a blackened 

 pasteboard diaphragm, leaving spaces at the top and bottom. 

 In the top space a piece of magnetized sewing-needle, fur- 

 nished with a glass index, was suspended by a single silk fibre. 

 Any excess of temperature on one side of the partition caused 

 an air-current, which at the top passed from the hot to the 

 cold side and deflected the needle. 



* Proc. Manchester Phil. Soc. vol. iii. p. 73. 



