the limiting state of hardness for the temperature (1000) under consideration. But it also appears that the same limiting value of specific magnetism gradually reasserts itself, no matter how often the combined process of magnetization and subsequent indefinite annealing may be repeated. Moreover, a magnet which has approximately reached the limiting hardness for the given temperature, if remagnetized to saturation and annealed again at the same temperature, reaches a limiting magnetic condition, the value of which is nearly independent of the time of exposure.

The purely magnetic effect (permanent).-The inference enunciated at the end of the last paragraph still needs additional proof. Accordingly, our magnets Nos. 11 and 12 were now magnetized afresh and then exposed to steam in the uniform manner described. We thus arrived at the following results:

TABLE 69.-Specific magnetism, m, of saturated rods successively annealed at 100°.

We also remagnetized the steel parallelopipedons Nos. I... VI, with which the original experiments were made, and then annealed them in steam in the usual way, repeating the whole process a number of times. The following table contains the values of specific magnetism obtained: TABLE 70.-Specific magnetism, m, of saturated rods successively annealed at 1009.

If the present behavior of the magnets, where the steel has practically reached its limiting mechanical state for 1009, is contrasted with the above, where temper varies simultaneously with magnetism, a m

smaller and thoroughly uniform loss of specific magnetism is everywhere apparent. In the case of No. 11 the original loss amounted to 62.6-43.8 =30 per cent. nearly; whereas at present the average loss is

62.6

=5.3 per cent. Similarly, the average loss of specific

.19.62—17.93
19.62

=8.6 per

magnetism of the rods I... VI was originally

cent. nearly. Now we have only an average loss of cent. But owing to the fact that these magnets had been specially treated at the outstart-repeatedly heated from 150 to 600-an immediate comparison between the last results and those for No. 11 is not to be made.

We have thus arrived at a partial corroboration of certain results obtained by Moser and Riess," and subsequently also by Dufour.145 Following Moser and Riess, we have for the successive losses of a hard needle:

After the first magnetization.......
After the second magnetization.
After the third magnetization

Per cent.

44.0

6.1

4.4

But with this last result our observations are at variance. As has been stated, our data have invariably shown that when the maximum of permanent hardness corresponding to any temperature has once been attained, then the magnetic effects of repeated application of the same annealing process are identical, the losses of specific magnetism experienced by saturated rods constant.

The direct and indirect effect of temperature.—We conclude, therefore, that if it be our object to perspicuously represent the law of the phenomena in question, it is essential to discriminate between two species of magnetic loss. If the magnet is in such a condition-for instance glass hard-that the higher temperature (7) produces a mechanical effect, then this is invariably accompanied by a magnetic effect peculiar to itself, and as experiment has shown, of relatively very large intensity. The reasons for this behavior are obvious. The existence of magnetism is conditioned by a strain of a particular and characteristic kind. The same is true of hardness. It is very probable, therefore, that the partial disappearance of one of these strains from any cause whatsoever will materially interfere with the intensity of the other.146 Why the influence of the time of exposure to 100° is marked when the state of hardness is such that annealing produces both a mechanical and a mag

144 Moser u. Riess: Pogg. Ann., XVII, p. 403, 1829.

145 Dufour: Fortschr. der Physik, 1857, p. 435.

146 Whether mere magnetization produces a change in the temper of glass-hard steel is still to be investigated. In consequence of the very small variation of dimensions the anticipative effect must, of course, necessarily be small.

netic effect, is readily seen. For the latter effect must continue to vary until the limit of variation of the former has been fully reached; and the annealing effect of 1000 in case of glass hard steel is a diminution of hardness occurring at a very gradually decreasing small rate through infinite time.

When this has occurred-i. e., when the final state of hardness due to an exposure to T has been reached-we have to do with a purely magnetic phenomenon only. A rod magnetized to saturation and annealed at To experiences a direct effect-a loss of specific magnetism which is relatively small, nearly independent of the time during which T acts, and the cause of which may be loosely ascribed to a smaller coercive force at T and to the effect produced by the thermal expansion on the magnetic strain. We may add that while in the first case, where the rod itself undergoes a change of state, a limiting value of specific magnetism was not fully reached even after 22 hours of annealing; in the second, the action is certainly complete after the lapse of an hour, and occurs in such a way that the principal part of the magnetism is lost within the first ten minutes.

The reasons fully appear why Moser and Riess found that when soft and annealed rods were used the losses were not only small, but occurred with the characteristic rapidity of those here enunciated. In this case an annealing effect due to 100° is manifestly impossible, and the peculiarities of the purely magnetic phenomenon are alone observed. It would moreover appear that the latter for a given temperature, T, is independent of the material used, of an intrinsically magnetic nature. At least Moser and Riess found for this loss

When the needle was soft

When tempered blue

When tempered cherry red

Per cent.

13.6

13.4

13.7

We will waive this matter here, as it is our object to investigate it specially, paying particular attention, moreover, to the effect incident upon a variation of the dimensions L/D. Such an effect is already, though somewhat obscurely, apparent.

Pre-existing magnetization.-If the inference derived in the foregoing paragraph be correct, then must it be immaterial whether a glass-hard unmagnetic steel rod is first annealed, say in steam, at 100°, until the final mechanical state for this temperature has been practically reached and then magnetized to saturation, or whether the rod, originally saturated, is annealed and then remagnetized, as in the previous case. The ultimate result must, in other words, be independent of pre-existing conditions so long as these are effects of a lower order than correspond to the given temperature.

In order to give this question, which partakes of the nature of a crucial test, due experimental consideration, two rods, Nos. 13 and 14, of equal length, were broken from a glass-hard sample of the same

thickness (0.084 cm.) and material as Nos. 11 and 12. The constants of the new magnets are:

Of these, No. 14 was magnetized to saturation; No. 13, however, left unmagnetized. Both were then exposed to the action of steam, and the progress of the annealing investigated by repeated measurements of the specific resistances of the rods. Unfortunately, a piece of No. 13 was accidentally broken off in clamping. The new rod (No. 13) was 8.7 cm. long and weighed 0.363 g. The two wires in their present condition would not, however, permit us to discuss the question from a sufficiently broad standpoint, and we, therefore, selected three other wires of the diameter 0.2 cm., so chosen as to present nearly the same specific resistances, viz:

The rods were tempered by sudden cooling after heating to redness in the flame of a blast-lamp. Out of the first but a single magnet was taken, No. 15, and but one, No. 16, from the second; while to the third and most homogeneous of the three the two shorter magnets, Nos. 17, and 18, owe their origin. It was intended to have the lengths of Nos. 15 and 16 and of Nos. 17 and 18 identical, but it is difficult in the case of wires of this thickness to break them off at a prescribed mark. Small variations of length are, therefore, unavoidable. The constants of the four magnets (0.21 cm. thick) are:

These were now treated in a manner analogous to that applied to Nos. 13 and 14; 11 hours of annealing in steam at 1000 transferred them into the final state of temper for this temperature, not completely, it is true, but sufficiently so for the purposes. All were now remagnetized. Nos. 15, 16, 17, 18, acted on by steam for some time, and their magnetic behavior examined, were then remagnetized again, and once more annealed. Nos. 13 and 14, however, were first exposed to a lower temperature, that of boiling methyl alcohol at 66°, during a certain in. terval; and not until the magnetic limit for 66° had been fully reached

were they exposed in steam, in order that the limiting value of permanent specific magnetism corresponding to the new temperature (100) might in its turn appear. The data expressing the magnetic effect of these operations are detailed in the following tables, 71 and 72. As before, W denotes the resistance per meter of length of rod at to, s the corresponding specific resistances at zero.

Magnetic effect of annealing-Final result.-In this series of results. our views are fully corroborated. When the limiting state of hardness conditioned by the temperature of annealing T has once been reached, then it is wholly immaterial, in so far as the subsequent magnetic behavior is concerned, whether the rod was originally a magnet or not. The consecutive values of m for Nos. 15 and 16, as well as Nos. 17 and 18, after 11 hours of annealing and remagnetization have been applied, manifest a perfectly similar progress throughout. The same is true of Nos. 13 and 14, both while in vapor of methyl alcohol and while in