680 Prof. Fleming and Mr. Richardson on the Effect of an 



on the balls this deflexion hardly ever remains steady, the 

 needle wanders to and fro over the scale, and the observer 

 can at best but take a mean reading. The result is that the 

 points plotted for the resonance curve do not lie well on a 

 smooth curve, and it is particularly difficult to plot the 

 important part of the curve near the maximum value. This 

 difficulty has been experienced by other observers. Thus, 

 Mr. R. A. Houstoun has recently made a series of measure- 

 ments of spark resistance and decrement with spark-balls of 

 various metals (see Proc. Roy. Soc. Edin. 1908, vol. xxviii. 

 p. 369, or ' The Electrician/ vol. 62, p. 636) ; and he remarks 

 that even with great care taken, measurements of logarithmic 

 decrement of oscillation made with the same spark-gap and 

 circuit do not agree well together, the resistance of the spark- 

 gap appears to vary irregularly. Again, Messrs. J. E. Taylor 

 and "W. Duddell in some radiotelegraphic measurements made 

 in 1905 (see Journ. Inst. Elect. Eng. vol. 35, p. 321, 1905) 

 also complained of the difficulty of obtaining good measure- 

 ments with a spark-transmitter. These difficulties have their 

 origin in the actions taking place in the spark-gap. If, 

 however, a steady blast of air at a suitable pressure is thrown 

 between the spark-balls, this irregularity is greatly reduced 

 provided the spark-gap is not long. The deflexion of the 

 galvanometer becomes greater and much more steady, and 

 accurate observation of the values of the current a corre- 

 sponding to known values of n becomes much facilitated. 



With the above described arrangements a resonance curve 

 can be easily taken as follows: — The jet of air from the Lennox 

 blower conveyed by a rubber pipe ending on a glass nozzle 

 is allowed to play between the spark-balls, and these are 

 connected together hj a condenser of known capacity in 

 series with an inductance which has either been measured or 

 predetermined. This inductance is preferably formed of 

 round copper wire and may be rectangular in form, and its 

 high frequency resistance and inductance can then be calcu- 

 lated by known formulae. 



The cymometer is then placed alongside this rectangle so 

 as to have induced oscillations created in it, and by means of 

 the hot-wire ammeter inserted in its circuit the mean-square 

 value a of the cymometer current is taken for various 

 settings of the cymometer circuit, which give it various 

 assigned natural frequencies n. If A and IST are the maximum 

 values obtained we then calculate the value of the sum of the 

 decrements of the spark and cymometer circuits from the 

 Bjerknes-Drude formula given above *. Since the resonance 



* See ' The Principles of Electric Wave Telegraphy/ by J. A. Fleming. 

 Longmans & Co., pp. 221-223. 



