r>S hkuh;ma\. 



soom lo luo ihat tluMV is a j;ivat doal of vsigiiifioancv in iho '" oooilioiont 

 of spooitio irsisiauiv." Howovor, it is of intorost lo noio in ilic tahlc 

 that tho changvs of diuieiisioiis of maiiijaniti ami tluM'lo aiv so iaruo 

 v'oiu[>aivil witli tho tonsimi oooHioiom of obsorxod rosistamv tliat tho 

 toiision v'oothoiont of spooitic rosistaiioo is iiogativo, wlioroas tho tonsioti 

 iwlUoioiu of obsorvo«i rosistanoo is positivo. For tho othor inoials tho 

 ciM'rtvtiotv for ohaiijjo of figiuv is not larj^o oiiongh to ohango tho sign 

 of tho a>othoiont of obsorvoii n\sistaiioo. 



It is ill tho first phioo to ho romarkotl front tho tahlo that of tho sovon 

 substattcvs whioh aro ahtiornial wiih rospoot to tho siirn of tho prossttro 

 ivotlioiont. only two, bismuth and strontimn. aiv abnormal with 

 n\spoor to tho sign of tho tonsion cH>otticiont. This would soom to 

 indioaio somo ossontial dilVoronoo botweon tho oondtiotion moohanism 

 of thoso two stibstamvs and that of tho othoi-s. Lot ns disonss what 

 this ditVotxMtcv may bo in tho light of tho thtvry of motallio oondnotiott 

 whioh 1 ha\o ptvvionsly do\olopod. 



1 havo thought of ciwduotion as duo to a froo path mtx'hanism; 

 tho olassioal thtx^ry was a frtv path ihoory. Tho ditVortnicos oomparod 

 with .tho olassioal tlu\M"y aro those. In tho first place, tlie free paths 

 aro thotight of as loitg, booanso tho frtv olootivns an^ fow in ntimbor. 

 In normal motals. tho paths of tho oUvtn.>ns aro to bo thought of as 

 tlnxnigh tho snbstancv of tho atoms thomsolvos. The path may bo 

 terminattHi w hon tho eleotUMi makes the jnmp frtun one atom to the 

 next. Tho ohanct* of termination on making the jump will depend 

 both on the amplitude of atomic vibration and tho distantx^ apart of 

 tho atoms. Now if tho distamv apart of the atoms varies little com- 

 patwl w ith tho ohangi^s of amplitude, the variation of ft\x^ path may be 

 calcnlatoii in terms of the variation of amplitude only. The changes 

 of amplitude, nogkvting the ort'eots dne to changes of dimensions, may 

 be calcnlatoii for changes of prt^s^nre and tomperatnre, and so the 

 ehangi^ of path, and hoiuv tho changt^s of rosistamx^ may also be cal- 

 cnlaiixl. It is in ihrc»wing the entire bunion of tho variations on the 

 free p<ith. and in the method of computing the changes of the free 

 path, that my thixiry differs matJiematically frtim the classical thoor\-. 

 Now as a matter of faet, the changt^s of dimensions imder changt^s of 

 temperature an^ very small ctMuparoii with tlie changt^s of amplitude, 

 and the caloulattxi ehanges of rt^sistanct^ agree well witli the observed 

 ohangvs. Under ehanges of pressure the changt^s of dimensions are 

 several foUi larger, but still are so small compareti with the changes 

 of amplitude tltat an iiujx)rtant part of the pressure coelhcient 

 may bo (.vmpucoii. There is left an outstanding effect depending on 



