Magnetic Permeability of an Iron Bar. 279 



This comparison makes it clear that the reduction which 

 occurs in the thickness of the apparent air-space as the mag- 

 netizing force is raised cannot be explained as an effect of 

 magnetic stress in drawing the parts of the bar together. 



The magnetic stress (— - in dynes) amounts to less than 4 



kilogs. per square centim. when 33 is 10,000, and to less than 

 8 kilogs. per square centim. under the highest value of 33 

 reached in these experiments. Nevertheless the changes of 

 53 are associated with a reduction of the apparent air-space, 

 which is much greater than the reduction effected by an ex- 

 ternally applied load of 226 kilogs. per square centim. It is 

 startling to find the apparent air-space in the loaded bar 

 diminishing from 0*020 to 0"007 millim. as the magnetization 

 was strengthened, when the pieces of the bar were already 

 pressed together with a force so great that the additional force 

 due to magnetization was relatively insignificant. 



In the foregoing experiments the bar was simply cut in the 

 lathe, without any attempt to make the ends strictly plane. 

 In subsequent experiments comparison was made of the per- 

 meability of a bar, rough-cut in this way, with that of a bar 

 in which the cut faces were carefully shaped by scraping and 

 testing against a standard Whitworth-plane. For this pur- 

 pose a new bar was taken, and after being tested in the solid 

 and again in the rough-cut state (cut into two parts), the two 

 parts were sunk tightly in a pair of cast-iron blocks with 

 plane surfaces about 3 inches in diameter, so that the end to 

 be faced true was flush with the surface of the block. The 

 faces of the blocks were carefully shaped to true planes against 

 each other and against a Whitworth-plane, the surface of the 

 block itself and of the bar that was held in it being worked 

 up together. The bar's permeability was examined by taking 

 a curve of 53 and «£j in six states : — 



(1) Uncut bar; no load, 



(2) Uncut bar, loaded with 226 kilogs. per square centim. 



(3) Bar cut in two (at the middle). Ends as cut in lathe 

 (rough-cut). No load. 



(4) Ditto, loaded with 226 kilogs. per square centim. 



(5) Surfaces faced true at the cut. No load. 



(6) Ditto, loaded with 226 kilogs. per square centim. 



The results are shown in fig. 3, where four of the curves 

 are drawn in full, and the remaining two are indicated by 

 having the experimentally determined points marked sepa- 

 rately. The curves are numbered to correspond with these 

 successive operations. As in the former example, parve 2 



