354 



0. G. KXOTT AND .s. KIMÜKA. 



In the last Table we see, almost at a glance, the pr();n'i'e.s.s of thiii"'.s 

 as the tension increased. The graphs of Figure 11. are obtained by plot- 

 ting the electromotive Ibrces c<)iTesponding to one temperature in 

 terms of tensions. These sIkjuRI correspond in general features to tlie 

 cm'ves obtained by Colin and Ewing. In a very general they do s«) ; 

 but they are nuich more irregular. This perliaps is not surprisinu" if 

 Ave bear in mind the fact that each graph is made up out of as many 

 ditt'erent days' experiments as there are points. If we leave out of 

 consideration the experiment for tensicm (>, the remaining points on 

 each graph arrange themselves in a fairly regular manner. There does 

 not, however, seem to be any sufficient reason for omitting this experi- 

 ment. Tor the peculiar deviations of aJI the p<jints belonging to it 

 cannot l)e easily explained as due to any errors in reduction either to 

 temperature or to electrom<jti\ e force. The same pecidiarity appears if 

 we use the unreduced thermometric readings in drawing the curves. 

 Un the other hand, the galvanometer constant was almost exactly the 

 same day aftci' day (as may be seen from Table I), excepting for tlie 

 two last series (jf ex])eriments at the liighest tension and the ünal zero. 



In draAving our conclusions we nuist liowever Ijear in mind the 

 smallness of tlie cjuantities tal)ula(ed iu Table \\ . The probable errors 

 of (jl)ser\ ation are of tlie order of tlie smaller cjuantities given in that 

 Table; so that it would be out oi' the (juesti<ni to attach anv impor- 

 tance to N'alues less than 5. 



j\e\'ertheless, we ai'e able t<) recoi^'nize in the graphs fiu-ured a 

 certain ordered succession of changes ; and there can be no doubt as to 

 the signilicance of the values for the aftei'-strained Avire. Here Ave 

 have a result apparently ncAV to the subject ; we are not aAvai-e that the 

 possibility of such an effect has eyeu been hinted at Ijy })revious 

 workers. We have already expressed the nature of this result bv 

 saying that the Thomson Effect in an iron Avire imdergoes a permanent 



