346 C. G. KNOTT AND S. KIMÜKA. 



After the be.st iiietliod of experinieutiiig had been by long trial 

 decided ii[)on, the character oï the experiment td part of the rewearch ^y•dü 

 in it.self very tediou.s ; and «ince month« of preliminary and other- 

 wiöc futile labour had already been .spent it :-eemed be;^t to po.^tpone a 

 .continuation of the experiment« till «ome future date. So far there 

 have been no opportunities for renewing the attack, other work fully 

 enL'TOs.sini>' our time. 



Vie are now pre[)ared to di«cu8.s the result« of the fi nal set of 

 experiment« with iron wire. 



The dimen«ion« of the tube bar were a« follow« : 



Total length of l)ar 102 cm. 



,, ,, tubular part 90 ,, 



External diameter (jf ,, ,, 4.4 ,, 



Internal ,, „ „ „ 2.2 „ 



The diameter of the iron wire u«ed was 1.2 mm. It projected about 

 a foot beyond the c<)ld (jpen eiid of tlie tul^e and wa« attached to a 

 spring dynamometer measuring p(/unds- weight. The dynamometer 

 wa« fixed to a «crew working in a lixed nut ; and Ijy thi« means the 

 tension could be increased or diminished as desired. 



In the final set of experiments each applied stress acted for at 

 least one wIkjIc day before the thermoelectric observation« w^ere 

 begun. The wire w;i« left ïnv this interval at the «nxlinary tempera- 

 ture of the air. 



The solid end of tlie cylinder wa« then heated to brio'ht redness 

 in a cliarcoal furnace ; arid after 2 hour«' heating the temperature 

 gradient l)ecame fdrly steady, as indicate«! l)y the thermometric cur- 

 rents on the iz'alvanometer. 



o 



There were five pairs (jf jiuictions, ten in all — five thermoelectric 

 and five thermometric. The [xjsitions of these juncti<jns along the 



