This paper forms the continuation of a paper already communicated to the Royal Society of Edinburgh, and published in the Transactions (vol. xxxv. pp. 377–390). Strictly speaking, it is third of a series bearing on the same subject. In the first I showed that when a current is passed along a longitudinally magnetised nickel wire a twist is produced similar in character but opposite in direction to the twist which Wiedemann discovered to be produced in iron similarly treated. This was the chief result arrived at; but other results were also obtained, especially with regard to the influence of tension, which called for further investigation. In the later paper, the influence of tension was more thoroughly studied; and also the effect of change of temperature. I also discussed more fully the suggestion first made by Maxwell that the effect discovered by Wiedemann can be explained in terms of the changes of length which Joule observed in iron when magnetised. Obviously the same explanation must be extended to the case of nickel, which Barrett had found also to be subject to changes of length when magnetised. Mr Bidwell's recent elaborate measurements of these changes of length enabled me to make an approximate calculation of the twist in an iron tube longitudinally and circularly magnetised. The comparison of the result as calculated for a tube with the result observed for a wire of the same diameter, established, in my opinion, the sufficiency of Maxwell's explanation.