.Associates: mean values.

. Elliptical method.

. Experimental.

. Quaternian series method.

Lower atom-numbers

I-3 -5 -7 9 -II- 13

2- ·4

3

-5

-7

ΙΟ -12t· 14t 11-[13-15]

all the elements are known except, at the most, four | elements as above indicated. Rutherford and Andrade's (Phil. Mag., xxvii., 854) experimental analysis of matter by excited radiation gives lines which assign lead to its proper and expected place in the extension of the series as indicated by X-ray analysis, and therefore no new elements are to be expected between gold and lead.*

Upon careful consideration of these supposed elements, falling as indicated next to ruthenium, osmium, and samarium, it will be seen, however, upon referring to "Studies in Valency," p. 17, that the lacunæ for these three elements in question are of the same character in the valency plot shown, indicating possibly that these may be natural breaks in an otherwise continuous chain, since, by closing the table up at these points, the irregularity becomes in itself regular. The number relations herein given support this contention.

The table under further consideration, with respect to the right-hand half of non-radio-active elements, lends itself to a complete arrangement of associate atoms (see CHEMICAL NEWS, cx., 25) in suchwise as to bring into

*The composite nature of the atom may be revealed by some characteristic radiation-such, for example, as the X-ray spectra of the elements as experimentally developed by Moseley. As a first approximation in theory, the two characteristic lines observed might arise from the presence of two types of atoms differing in atomic weight; but theoretical considerations do not necessarily point directly to such a view. These, however, are matters which the experimenters themselves can doubtless discuss to better purpose. For a general account of the interference and reflection of X-rays by crystals and the analysis of matter thereby, see "X-Rays," by Kaye (Longmans, Green, and Co., 1914), particularly Chapter XII., but the original papers should be studied. For a list of references leading up to the developments by Moseley, seeText-book of Inorganic Chemistry," by Messrs. Friend, Little, Turner, and Briscoe (C. Griffin and Co., Ltd., 1914), vol. i.. See also Tutton, CHEMICAL NEWS, 1913, cvii., 277, 289, 301 P. 50. but see particularly Moseley, Phil. Mag., December, 1913, p. 1024, and April, 1914, p. 703.

harmony the values by the elliptical method, those of the quaternian series when available (there being a limited number of these), and those by the summation method, also limited in number. Moreover, the values are not at variance with those obtained by experiment when the more accurately determined values are compared with the theoretical ones, as the accompanying table will show. In this table, however, the atomic weights enclosed by the circumscribed line are for one reason or another uncertain, notwithstanding the coincidences. The other atomic weights for the most part are probably correct; those obtained by taking mean values (m.v.) of the associate atoms as shown being the preferred ones, and these are supported by experiment. It must be remembered that the various methods employed represent tools which are capable of some adjustment, and finality is not possible in every case.

The setting aside of the inactive gases in these numberstudies should not be taken as an indication that they are not true elements, but rather that they are chemically inactive, and consequently count as nought when balancing or otherwise arranging the chemically active members. They appear to fit into the X-ray spectrum series, and take up a normal zero place in practically all periodic tables, and from various considerations the universal opinion is that they are true elements (see Soddy, Science Progress, 1914, viii., 654).

Since these relations are largely based upon the idea of branch systems of elements, the peculiarity of these branches should be kept in mind, otherwise some of the statements would seem contrary to accepted opinion; whereas the method of treatment is, I believe, quite as comprehensive in its systematic expression of fact as any other method, but this is a matter for others of wider experience to judge.

CHEMICAL NEWS, Jan. 8, 1915

}

These relations seem to harmonise with those published in the CHEMICAL NEWS, 1912, cvi., 37, when main and subordinate considerations (details) are not confused. (See also CHEMICAL NEWS, 1913, cviii., pp. 95, 188, 247, and 305). Greater interest than ever seems to be attached to the problem of exact atomic weight determination, and the following statement by Prof. Soddy (CHEMICAL NEWS, cviii., 169), which I am sure was not intended to minimise the importance of such work, should be of profound interest-if I may express what appears to be the opinion of others:

"The chemical analysis of matter is thus not an ultimate one. It has appeared ultimate hitherto on account of the impossibility of distinguishing between elements which are chemically identical and non-separable unless these are in the process of change the one into the other. But in that part of the Periodic Table in which the evolution of the elements is still proceeding, each place is seen to be occupied not by one element, but on the average, for the places occupied at all, by no less than four, the atomic weights of which vary over as much as eight units. It is impossible to believe that the same may not be true for the rest of the table, and that each known element may be a group of non-separable elements occupying the same place, the atomic weight not being a real constant, but a mean value, of much less fundamental interest than has been hitherto supposed. Although these advances show that matter is even more complex than chemical analysis alone has been able to reveal, they indicate at the same time that the problem of atomic constitution may be more simple than has been supposed from the lack of simple numerical relations between the atomic weights."

Whatever else these relations may signify, they certainly broaden one's conception of Nature's working, in that the observing student must no longer look at things through one pair of spectacles but through several of different character or colour before the composite design is fully brought to view.

|

15

at high temperatures is non-magnetic, might itself become magnetic at lower temperatures. Non-magnetic manganese steel, however, is not transformed even at the temperature of liquid air (Hadfield, Four. Iron and Steel Inst., 1905, i., 179), and no treatment can make austenitic steels magnetic which does not increase the specific volume at the same time, and when this takes place it points conclusively to the transformation to a iron, for the thermal magnetic transition makes no appreciable alteration in the specific volume.

The development of enormous crystals in low carbon steels which have been strained and afterwards reheated to any temperature below Ac3 (Stead, Four. Iron and Steel Inst., 1898, i., 145; Chappell, Four. Iron and Steel Inst., 1914, i., 472), and in thin sheets of electrolytic iron cooled from above Ac3 (Stead and Carpenter, Four. Iron and Steel Inst., 1913, ii., 113), proves that iron possesses the same crystalline form above A2 as below it, and that therefore the A2 point is not evidence of allotropy. 3. The great and important distinction between the two forms of iron is their behaviour towards carbon, which is soluble in a iron only to a very small extent, but can dis solve with readiness in y iron, and on this preferential treatment rests the explanation of the unique position which carbon occupies as a cause of hardening by quenching in iron. There is no doubt that if another element can be found which posesses the same property of exclusive solubility in y iron, then that element also will possess to a similar extent the hardening power of carbon. 4. Pure iron, in common with all metals, is hardened after permanent deformation, and it is best described after such treatment as being in a condition of "interstrain." Metals in their normal unstrained condition are crystalline. That is, they are composed of an orderly arrangement of their constituent atoms arranged in a space lattice which conforms with whatever symmetry the crystals possess. Successful attempts have been recently made by the aid of X-rays to determine the actual position occupied by the atoms, and copper (Bragg, Phil. Mag., September, 1914), for instance, has been shown to have its atoms built up in a "face centred" cube-that is, a cube having an atom at each corner and one at the centre of each face. The effect of deformation on such systems must be to

By ANDREW MCCANCE.

I. THERE are three methods by which a pure metal may be made harder, namely, by mechanical deformation, by the addition of elements which form solid solutions, and by the application of heat treatment (in certain cases), but as yet there is no theory which can be said to demonstrate the underlying connection which must exist between these three means of achieving the same end. For it can be said with certainty that so long as the metal retains its chemical characteristics the same effects must be derived

from the same immediate causes.

It is with the hardening effect caused by heat treatment that this short paper will deal, but it can be shown that what is true for iron is also generally true for all alloys where corresponding conditions exist.

2. The great amount of attention which has of late been concentrated on the nature of the allotropic changes in pure iron has made it clear that from ordinary temperatures up to 1000° C. only two real allotropic forms of iron exist, the a and the y, and that the so-called 8 iron is in reality a iron which has become non-magnetic from purely thermal causes, the apparent liberation of heat at A2 being due to the rapid change in the specific heat which necessarily accompanies the ferromagnetic transition (McCance, Four. Iron and Steel Inst., 1914, i., 223; Benedicks, Ibid., 1914, i., 439).

Since a iron loses its magnetic properties above a defifinite temperature, it might be suggested that y iron, which Contribution to the General Discussion on "The Hardening of Metals," held before the Faraday Society, November 23, 1914.

and permanent deformation must cause so much alteration that the atoms along the planes of slip can no longer return to their positions of equilibrium. This corresponds to the condition of interstrain, and results in a hardening of the material. That the destruction of the crystalline arrangement goes so far as to render the material in those planes of greatest movement actually amorphous and devoid of crystalline form, as Beilby has proposed, has never been seriously questioned, but the writer thinks that the following facts are strong arguments against this theory of a hard, vitreous, amorphous phase. In the case of a dimorphous metal like iron, the vitreous phase to be truly amorphous must possess the same physical proper. ties, no matter from which crystalline form it has been produced by deformation, and since strained a iron loses none of its magnetic intensity, we must assume that amorphous iron is magnetic; consequently it is to be expected that strained y iron will also be magnetic since the same constituent is produced. The important results of Hadfield and Hopkinson (Four. Iron and Steel Inst., 1914, i., 125) show, however, that the magnetic intensity of austenitic manganese steel after severe straining is only augmented to o°3 per cent of the intensity of pure iron, and is still practically non-magnetic, although the hardness has increased from 200 to 490. It would seem that the iron in the interstrained condition has not lost the distinguishing properties which are the result of crystalline form.

Hanriot (Comptes Rendus, clv., 713, 1502) exposed small cubes of metal to a fluid pressure of 10,000 kgs. per sq. cm., and although no measurable deformation could be detected, the hardness had increased in every case as

follows:

The microstructure was the same after this treatment as before, while to suggest that the hard vitreous phase had been formed as the result of pressure would, of course, be out of the question, since iron which expands on becoming liquid would have its melting-point raised ar 1 not lowered by pressure.

Deformation does not necessarily produce an amorphous constituent, but only results in some permanent alteration of the atoms from their positions of equilibrium in their crystalline space lattice; and, conversely, it will be true that any metal whose atoms are prevented from reaching their positions of equilibrium will be in a condition of interstrain, and will be hardened.

5. To harden steel it is necessary to heat it above the Ac point-that is, to heat it to the temperature at which the carbon is all in solution, and then to cool it at a rate which is sufficiently rapid, but the greatest hardness is obtained when the temperature reaches the highest change point-that is the Ac3 for hypoeutectoid and Ac cementite for hypereutectoid steels. This is shown in Fig. 1, giving the hardness of three steels with 0.35 per cent, o'86 per cent, and 1.18 per cent carbon, and the temperature of quenching in degrees above the Ac1.

With 0.35 per cent carbon the structure, after quenching from just above the Ac, will consist of ferrite and martensite areas (separated by transitional troostite, &c.), and the increase in hardness from this to just above the Ac3, when the structure is wholly martensitic, corresponds with the gradual replacement of the ferrite by the hardest con stituent. With 1.18 per cent carbon the greatest hardness is obtained by quenching from just above the Ac cemen

tite, and the softening which takes place on increasing the temperature corresponds with the appearance of austenite in the microsection. A very important result is that when the carbon reaches about o.7 per cent a maximum hardness is obtained in the quenched state, and this value is not exceeded in any pure carbon steel. If the hardness were the direct consequence of carbon in solution, the hardness would be__proportional to the amount of carbon dissolved. That it is not so shows that the action of carbon is indirect, and that the hardening element is the iron itself.

6. It is a fact well known that steels which at he emperature of quenching are non-magnetic are magnetic in their hardened state, and this change must have taken place during the time occupied by the quenching. Since it has been previously shown that y iron does not itself become magnetic, it is necessary to conclude that during the time occupied in cooling from the higher temperature to the lower the original y must have changed to the magnetic a condition. Measurements of the electric resistance show that the carbon has not changed its condition, but has remained in solution, and so we recognise the fact that under the conditions of quenching the transformation of the iron from the y to the a state can take place independently of the change in the state of the carbon from a state of solution to the undissolved state. These two transformations, which on slow heating and cooling take place simultaneously, can behave as independent reactions, although in a sense which is limited. For though martensite may be considered as an enforced solution of carbon in a iron, it must not be forgotten that there is still some iron present with the carbon, which is absolutely necessary to maintain its solubility, and any change in the state of the carbon causes the transformation of this y iron to a. It might on this account be suggested that the iron is chemically combined with the carbon, but an examination of the curve (Fig. 2) connecting the loss of

CHEMICAL NEWS,

Jan. 8, 1915

magnetic saturation intensity with the carbon content shows that they iron increases much more rapidly than the amount of carbon, which points to purely physical influences.

7. It has been shown elsewhere (McCance, loc. cit.) that it is possible to calculate the velocity of quenching in a round bar on certain assumptions which are closely realised in the early stages, at least of quenching, and if the curve thus obtained is compared with the curve connecting the electric resistance, or the loss of magnetic intensity of a steel with '86 per cent carbon, it is seen that they are closely similar, so that the amount of carbon in solution and the amount of y iron retained in the cold state are proportional to the rate of cooling, at least until the carbon is completely in solution.

With a given rate of cooling the progress of the transition from the equilibrium position for y-crystals to that

17

The first of these can take place independently of the second, but when the second takes place it necessarily in| volves the first. Three possible conditions can therefore result:

(1) Both transformations are completed: this is the normal pearlitic condition of steel, which is soft. (2) Both transformations are suppressed; this corresponds to the condition of pure austenite, which is also comparatively soft.

(3) When (a) takes place but (b) is suppressed-that is, the complete transformation is partially suppressedthen the structure is martensitic and the a-iron formed under this condition is interstrained and very hard.

Case (3), then, is the important one from a hardening standpoint, and to be possible it is necessary that a difference should exist in the respective velocities of transformation of (a) and (b), so that with an appropriate rate of cooling the slower transformation can be suppressed while the faster is not greatly affected.

This intimate connection between the rate of cooling and hardening is further illustrated by the behaviour of alloy steels. The addition of manganese, for instance, by lowering the temperature of the transformations greatly decreases the velocities with which they proceed, so that with a certain percentage (which varies with the carbon content) the transformation of the carbon may be almost completely suppressed by air cooling, and the normal condition after such treatment is martensitic, accompanied by hardness and brittleness-in fact, all the characteristics of quenched pure carbon steels. Water quenching, on the other hand, results in the complete suppression of both transformations, and the steel becomes austenitic and soft, while the same result is obtained by air cooling when the percentage of manganese is further increased.

Commercial manganese steels after air cooling are nonmagnetic though fairly hard and brittle, but such steels usually contain about 1.2 per cent of carbon, and this condition is probably brought about by the deposition from a state of solution of the cementite in excess of the euctectoid composition.

Similar results are obtained by the addition of nickel, and the following facts are also instructive. A nickel steel containing 3.6 per cent of Ni and o'19 per cent carbon had a normal hardness of 183, which, after quenching in water from 1000° C., had risen to 444, but a steel containing 0.19 per cent carbon without the nickel increased after the same treatment from 143 to 218; that is, while the plain carbon steel increased in hardness 65 points by quenching, the nickel steel increased by 259. Nickel is known not to form a carbide in this range of composition (Arnold and Read, Engineering, 1914, pp. 463 and 468), consequently the state of the carbon is similar in both, and the increased hardening power must be due to the influence of the nickel on the iron. It has enabled the iron to retain a greater degree of interstrain by lowering the temperature at which the transformation of the y to a iron takes place.

8. Tempering allows the metastable conditions which

and by the Author.

of a-crystals will be arrested, and will ultimately be stopped by the increase in the internal viscosity at the lower temperature. The crystalline transformation, in other words, will be partially suppressed, and a condition produced which is essentially similar to the condition of interstrain produced by deformation, and the hardness which results is accounted for by this condition.

During the cooling of hypoeuctectoid steel from a temperature at which it consists of a homogeneous solution of carbon in y-iron, its complete transformation into a-iron and cementite comprises two individual transformations:

(a) A change in the state of the iron from to a. (b) A change in the state of the carbon.

hypoeutectoid steels the carbon is deposited from solution and the specific resistance falls, while at the same time the interstrained condition of the iron is gradually lost with the formation of a ferrite recognisable under the microscope (McCance, Four. Mechan. Eng., 1910, p. 1664). As a consequence, the magnetic permeability under low fields increases, the remanent magnetism decreases, and the peculiar shape of the curves connecting the magnetic properties with the temperature of tempering is found to be exactly similar for interstrained iron as for quenched steels-a similarity which extends also to the rate at which the hardness is lost (Maurer, Rev. de Mit., 1908, p. 711).

Turning now to the consideration of the tempering of high carbon steels which contain a proportion of pure austenite, the curves of specific volume electric resistance

H.

=

=

and hardness of one containing 1.66 per cent carbon due to Maurer are given in Fig. 3.

The smooth curve for the electric resistance shows that the carbon is deposited continuously, and this causes a decrease in the specific volume, but at 150° C. to 250° C. there is a sudden increase which is without effect on the electric resistance, and must, therefore, be due to the transformation of y to a iron, and since this takes place independently at a low temperature under conditions of high internal viscosity, the a-iron formed will be in a condition of interstrain, and an increase in hardness should result. This is strikingly confirmed.

9. The importance of internal friction in retarding transformations and preserving the metastable state has been pointed out, and in solids this internal friction is very similar to sliding friction between solid surfaces. A fixed amount of super-cooling below the transition range, for instance, is necessary before transformation can take place in solids, and the difference in the temperature of transformation on heating and cooling is a measure of the internal friction. Twenty-five per cent nickel steel has a transformation-point at 550° C. on heating, while on cooling it is transformed in the neighbourhood of - 100° C., which shows a range of 650°; but manganese steel containing about 14 per cent Mn has a larger difference in this respect than any other, since on heating the Ac point occurs about 550° C., while the Ar, does not even take place at the temperature of liquid air, a range of more than 740°.

Many of the abnormal properties which manganese steel possesses are due to this great friction. Deformation

renders non-magnetic nickel steel magnetic through the transformation of the y-iron to a with a decrease in density, but manganese steel, on the other hand, is practically unaffected magnetically.

Tempering at a temperature just below its change-point results in the gradual conversion of the y-iron to a, and the deposition of the carbon from solution, but the a-iron is interstrained and the hardness is increased. It is conceivable that very prolonged tempering would cause this hardness to be lost again, with the production of a soft manganese steel which is magnetic.

10. When pure austenite, formed by quenching 2 per cent manganese steels, or the austenite of high carbon steels, is immersed in liquid air, the greater super-cooling overcomes the internal friction, and they to a transformation takes place, although the carbon change is unaffected. The electric resistance is not changed, but the specific volume increases as well as the hardness, through the production of interstrained a-iron. Maurer obtained the following results :