108 



SCIENCE. 



[N. S. Vol,. I. No. 4. 



distribution of the magnetic and tlie elec- 

 tric force along the wire separately. 



b. The method of Bjerkness (Wied. Ann. 

 44, p. 74) in which two symmetricallj^ situ- 

 ated points of a long loop are connected to 

 the quadrants of a small electrometer and 

 the difference of potential measured. 



c. The thei'moelectric method [first sug- 

 gested by Klemencic (Wied. Ann. 42, p. 

 416)] employed by D. E. Jones (Rep. Brit. 

 Assoc, 1891, p. 561-562). The intensity 

 of the wave at any point of the wire is 

 measured by the thermoelectric effect pro- 

 duced in a thermopile placed in the imme- 

 diate vicinity of that point. 



d. The bolometric method first employed 

 by Rubens and Ritter (Wied. Ann. 40, p. 

 55, 1890). 



e. Perot's micrometric spark gap method 

 (C. R. t. CXIV., p. 165) by which the in- 

 tensity of the wave at any point is measured 

 by the maximum length of the spark gap 

 when attached to the wire at that point. 



The theorjr of each method is discussed 

 briefly but quite completely, and it is shown 

 very clearly that the results of the experi- 

 mental investigations cited above are in 

 good agreement with the theory and that 

 they all lead to the conclusion that the os- 

 cillations of the oscillator produce simple 

 waves, possessing a rapid rate of decay. 

 This is in accordance with Poincare's view 

 of multiple resonance. 



Bjerkness' experimental method (Wied. 

 Ann. 40, p. 94, 1891) of determining the 

 decrement of a resonator and Poiucare's the- 

 ory of it are then given and it is shown that 

 this decrement is a hundred times smaller 

 than that of the oscillator. 



A brief theoretical discussion of the 

 curves plotted by Perot fi'om the experi- 

 ments cited above closes this exceedingly 

 interesting and instructive part of the 

 book. 



It is pointed out now that the experi- 

 ments so far discussed do not decide the 



superioritj^ of the MaxM^ellian theory over 

 the older theories because it can be and has , 

 been predicted by older theories (Kirch- 

 hoff, Abhandl. p. 146) that the velocitj' of ~ 

 propagation of electromagnetic disturb- 

 ances along a long straight wire suspended 

 in air is the same as the velocity of light. 

 A review of some of the older exj)eriments 

 in this du-ection is then given. 



Direct Determination of the Velocity of Propa- 

 gation along Conducting Wires. — The earliest 

 experiments carried ou^t according to meth- 

 ods against which no serious objections 

 could be raised were those of Fizeau and 

 Gounelle (1850) over telegraph lines be- 

 tween Paris and Amiens, a distance of 314 

 kilometers. The method was similar to 

 that employed by Fizeau in the determina- 

 nation of the velocity of light. The mean 

 velocity was found to be 10' kilometers per 

 second for iron mre and 18X10* kilome- 

 ters per second for copper wire. They em- 

 ployed signals of, comparativelj' speaking, 

 long duration, and Poincar6 shoAvs by a ref- 

 erence to weU. known theoretical relations 

 that in this case there is a strong distortion 

 of the signals, so that a disturbance starting 

 in form of a short wave returns, after passing 

 over the whole line, in form of a more or 

 less steep wave fi-ont followed by a long tail. 

 This made the measurements verj' uncertain 

 and the velocitj^ of propagation necessarily 

 much smaller than it ought to have been. 

 The experiments of Siemens in 1875 avoided 

 this objection, in a measure, by employ- 

 ing the disruptive discharge of a Leyden 

 jar for the purpose of starting an electri- 

 cal disturbance on lines of varying length, 

 between about 7 and 25 kilometers. The 

 velocity found was in several cases nearly 

 250,000 kilometers for iron wire. Here 

 again the velocitj^ came out smaller than 

 that of light and for obvious reasons. 



The last and in all respects most success- 

 ful direct determination of the velocity of 

 propagation was that recently carried out by 



