THE AID OF THE ACHROMATIC FRINGES. 33 



of the same kind were also investigated with smaller fringes, but as they 

 were less uniform they, also, may here be omitted. The results of figure 34 

 for pure iron do not differ as much from the low carbon iron results of figure 33 

 as was anticipated. In the high fields, moreover, the former show definitely 

 that the contraction has not subsided, though its rate is decreasing. First 

 and third readings of the triplets were usually nearly the same, and but for 

 the vibration of fringes due to the extraneous causes, the results would have 

 been very satisfactory. 



With the aid of the results for Swedish iron, equation (3) in paragraph 17 

 may be resumed. With this end in view, the curve, figure 34, so far as the 

 higher fields (H> 200) are concerned, may be roughly reproduced by an equa- 

 tion of the form 



The data for the fields H = 2oo and 600 then give us A = io- 6 X7-3; 5 = 0.33. 

 The equation is not adequate, inasmuch as it either makes the final data too 

 small or the initial data too large. If we accept it, however, in order to deter- 

 mine the order of value of k, we may use equation (3) above and, eliminating 

 bl/l from (3) and (4), compute 



(5) k*=E(A-B/H*)/ 4 irH* 



This expression can have a meaning in the higher fields only; but if we use 

 it to find the order of values of k, the results are 



k = o.S 

 600 .6 



800 .4 



Hence one may infer from this tentative result that small residual values of 

 the susceptibility of iron, such as probably exist, are sufficient to account 

 for the contractions observed in strong fields. 



In addition to iron, a number of other metals, in particular bismuth, were 

 examined by the same method; but no elongation was observable in fields 

 up to 800 gauss. The bismuth rod was 39 cm. long. Large fringes were pro- 

 duced for which (AN/Ae) = io- 5 Xi.86. This makes AJ/Z = io~ 7 X2.43Ae, the 

 constants of the apparatus being as above. Assuming the Ae = o.5 could still 

 be recognized in a field of 800 gauss (the difficulty of observation being due 

 to the rapid heat production by the current), it follows that the expansion 

 must have been less than A///= io~ 7 . The same result applies to the other 

 metals. 



19. Magnetic elongations in a free=end coil. Profiting by the experience 

 of the preceding paragraph, I reinserted the original low-carbon iron rod and 

 completed a series of data in the larger magnetic fields, the coil being free to 

 expand at the end near the contact lever. These results are given in figure 35. 

 They are not superior to the earlier results, however, as was in a measure to 

 be expected, because all observations were made in triplets from which the 

 3 



