80 Messrs. K. Honda and T. Terada on the Change of 



§ 4. Concluding Eemakks. 



The results obtained in the present investigation may be 

 summarized as follows : — 



(i.) In Swedish iron and tungsten-steel (PL IV. figs. 45 

 and 46) , the change of elastic constants is generally positive, 

 but its amount is extremely small, less than 0*5 per cent. The 

 change by the indirect method is several times greater than 

 that by the direct method. It is interesting to notice that 

 the change of elasticity and that of rigidly almost coincide 

 with each other, not only in their general aspects, but also 

 in their quantitative relations. 



(ii.) In nickel (fig. 42), the change of elastic constants is 

 remarkably large, amounting to about 15 per cent, in the 

 change of elasticity and 7 per cent, in the case of rigidity. 

 The elastic constants first decrease and then increase, as the 

 field becomes greater. The elastic constants by the indirect 

 method are numerically greater than those by the direct 

 method. 



(iii.) In 28*74 per cent. M (fig. 48), magnetization increases 

 the elastic constants by a small amount. The changes given 

 by the two different methods nearly coincide with each other 

 for a moderate tension. In 5072 per cent. Ni and 70*32 

 per cent. M (figs. 49 and 50), the increase of the elastic 

 constants is remarkably large, approaching in amount the 

 change of rigidity in nickel. Except with a very weak 

 tension, the change by the indirect method is greater than 

 that by the direct ; but this difference becomes less as the 

 tension is increased. With a tension of 3 or 4 kg. per square 

 millimetre, it nearly vanishes. 



From the results above given, it is evident that there are 

 some cases in which relation (3), given in the earlier part of 

 this paper, does not hold even approximately, so that 

 equation (1) cannot be freely used in any quantitative 

 discussion. 



There are many analogous cases in the problem of magneto- 

 striction. As will be seen from our subsequent paper to be 

 published presently, the change of magnetization by stretching 

 a magnetized wire does not agree with the change deduced 

 from the results obtained by magnetizing the wire in the 

 unstrained as well as the stretched state. Similar phenomena 

 are also observable in the change of magnetism caused by the 

 twist. Next, we may cite the case of the Wiedemann effect. 

 It is well known that in iron and nickel, the twist produced 



