1912-13.] Sensitive Magnetic State induced in Various Ways. 255 
state. Perhaps the most interesting fact in this connection is that in every 
case examined the greatest effect was produced by the last decrement of 
all, the final removal of the applied stress — the final removal of even a 
small stress being much more effective than the removal of a larger stress 
when some weight was still left on. 
Comparing now these results among themselves, particularly the results 
obtained for the first four specimens, which form a graded series, we notice, 
first of all, that in each the ( b ) curve, representing the effect of removal of 
stress, invariably lies above the ( a ) curve, which exhibits the effect of 
application of stress, and as a general rule the (c) curve lies below both. Also 
the (a) and (b) curves pretty nearly follow the same course, rising and fall- 
ing together, except in the case of Specimen II. (0*55 per cent. C), where the 
rise and fall in the (a) curve come later than in the ( b ) curve, and Specimen 
V. ( 0755 per cent. C), where the reverse is the case. 
Comparing now the general values of the sensitive state readied within 
the chosen range by each of the specimens, we find that higher values are 
reached by Specimen II. than by any of the others, and that these high 
values correspond to greater stress- values than do the high values of 
Specimen I. Specimen V. shows lower values than Specimen II., Specimen 
III. much lower values, and Specimen IV. lower values still ; but all these 
specimens show increasing values towards the end of the range, and there- 
fore probably show a maximum effect for stresses beyond the limits of the 
chosen range. 
All these facts, then, seem to point to the probability that, with an increas- 
ing percentage of carbon in the composition of the steel, a higher possible 
sensitive state may be induced, but that to induce it greater stresses are 
required as the percentage of carbon rises. That a harder steel should re- 
quire the application of a greater stress to produce a maximum effect is only 
what might have been expected, but further investigation is required, in 
which, by the use of thinner wires or greater applied stresses, increased 
strains may be produced, in order to determine definitely to what extent 
increasing percentage of carbon in the steel goes along with increasing values 
of the sensitive state induced. 
Coming now to a comparison of the effects produced by the two methods, 
we find some marked similarities between them, though the fact that the 
phenomena have only been examined over a very limited range in the 
second set of experiments prevents a complete comparison being made. 
It may be noted, however, that in both cases the (b) curve is generally 
above the other two, i.e. that removal of stress and cooling correspond in 
producing a greater effect than the other processes ; while application of 
