and Retractions of Rods of Iron and Steel. 187 



its rate of progress and found it to be 3*6 divisions in ten mi- 

 nutes. Therefore these experiments were made on the rod 

 while it had a slow motion of elongation. The mean of the 

 elongations is 1*22 division; the mean of the retractions is 1*04 

 division, which subtracted from 1*20 gives *16 of a division for 

 the effect of the heat on one demagnetization. The reduction 

 of Table I. gave *12 of a division for this effect. The difference 

 in the two results I thus account for : — While the bar was 

 slowly expanding from the heat radiated by the helix, the circuit 

 was made and the elongation was immediately observed ; but 

 about five seconds elapsed before the reading could be obtained 

 and the circuit broken ; and during these five seconds the rod 

 was expanding, but so slowly that its amount could not be read, 

 but was often visible. That this minute expansion could not be 

 determined was to be expected ; for if the rod elongated from 

 heat 3*6 divisions in ten minutes, it elongated only *03 of a di- 

 vision in five seconds, and 03 of a division was a quantity too 

 small to be measured on the scale; but it nevertheless existed 

 there, and during the continuance of 18 make-circuits would 

 amount to , 03xl8 = *54 of a division — quite an appreciable 

 quantity when we come to calculate the mean with this fraction 

 contained in the sum of the retractions given in the last column 

 of the Table. Therefore, to obtain the effect of the heat deve- 

 loped at the moment of demagnetization, we should subtract *03 

 from *16, the heating effect of demagnetization determined with- 

 out this correction. This gives *16 — '03 = *13 of a division, 

 while from Table I. we deduced '12 for the value of the same 

 effect. The difference of only '01 division in the two results is 

 not, however, to be taken without some reserve ; for in the cal- 

 culations I assumed that the rod had the same rate of expansion 

 under a closed circuit as under an intermittent one ; and this I 

 did because I have no means of determining the difference, if any 

 exists. 



Experiments similar to those just given were made on all the 

 iron rods ; and similar results were obtained. 



Relations existing between the number of break-circuits, the heating 

 of the rod, and its elongation. 



At this stage of the investigation it became of interest to de- 

 termine the above relation. For that purpose I drilled a hole 

 6 inches deep in the direction of the axis of rod No. 3, of Nor- 

 way iron, and inserted into this hole a thermo-electric couple 

 formed of two wires (one of copper, the other of iron). This com- 

 pound wire was wrapped first with two layers of waxed silk, then 

 with twelve layers of floss-silk ; and over these layers I coiled two 

 more layers of waxed floss-silk, leaving, however, the point of 



