specific gravity of not over 2.38. Some quartzites invert to cristobalite more slowly than others and brick with a lesser content of cristobalite have a lower spalling tendency and also do not show an appreciably greater permanent expansion when subjected to long-continued heating. Brick made from this type of quartzite may be properly burned when inversion has occured to such an extent that its specific gravity is slightly greater than 2.38. Examples with analyses are given. Metal-cased magnesite brick consist of steel containers of square or circular cross section, filled with dead burned magnesite. These are laid as headers in the furnaces. When heated the steel fuses and impregnates the magnesite forming a monolithic lining. Such a lining is more porous than one of magnesite bricks and has the advantage of better withstanding rapid temperature changes. Such bricks may be used in place of magnesia and silica brick in parts of the open hearth steel furnace and in electric steel melting furnaces.

Interesting facts concerning refractories in the iron and steel industry: C. E. NESBITT and M. L. BELL. In this paper the writers state the importance of refractories and emphasize the necessity for their greater efficiency in the iron and steel industry. This improvement can only be accomplished by the cooperation of the producer and the consumer. In the manufacture of iron and steel, refractories meet a wide range of temperature, while destructive agencies such as acid, basic or neutral slags, severe thermal changes, load, abrasion, impact and expansion are present in varying degrees of severity. Tests on refractory brick, easily and rapidly executed, which show a close relation to actual service conditions were developed for determining the resistance to these destructive agencies. The most important working qualities can be determined by two or three tests namely the spalling and hot crushing tests for silica brick, and the spalling, hot load and slagging tests for clay brick. Variations in the life of blast furnace linings, open hearth roofs, converter bottoms soaking pits and ladle linings are mentioned. Results are given showing the marked decrease in crushing strength and increase in spalling of silica brick defective from fire cracks, poor moulding, poor slicking, etc. The writers show the close relationship of the spalling test results with the life obtained in open hearth roofs. The effect of the degree of fineness or size of particles in silica brick is illustrated by re

The effect on certain

sults of the spalling test. qualities of clay brick produced by the method of manufacture is illustrated by spalling and load test results. The effect of the degree of fineness and the reduction of strength by heating of clay brick is also shown. From the comparative data it is evident that refractories require most thorough study. Simple practical tests which can be run in quantity and which give data showing variations in quality which reflect on the life of the structure should be adopted. A more uniform product can be secured if a careful study is made of the variations in manufacture which effect the important qualities.

Superior refractories: R. C. PURDY.

Refractory problems in the gas industry: W. H. FULWEILER and J. H. TAUSSIG. In the coal gas process the temperatures range from 400° C. to 1500° C. Rapid changes in temperature and expansion must be considered. Silica material is used in the retorts and the combustion chamber. Fire clay material is used in the recuperators and where the temperature is below 1000° C. In the water gas process the temperature may be 1700° C. in the generator, together with the slagging action due to the ash from the fuel. Abrasion occurring in removing clinker is important. In the carburettor the checker brick are heated to 1200° C. and sprayed with cold oil. Fire clay is used in the generator linings, but other materials are being tried. Cements used in construction frequently do not receive proper attention. Labora tory tests are useful in controlling the quality of materials.

be scattered into a comparatively broad beam. Geiger and Marsden showed not only that much small scattering occurred, but also that in passing through the atoms of a heavy element some of the a-particles were actually turned back in their path. Considering the great energy of motion of the a-particle, this is an arresting fact, showing that the a-particle must encounter very intense forces in penetrating the structure of the atom. In order to explain such results, the idea of the nucleus atom was developed in which the main mass of the atom is concentrated in a positively charged nucleus of very small dimensions compared with the space occupied by the electrons which surround it. The scattering of a-particles through large angles was shown to be the result of a single collision where the a-particle passed close to this charged nucleus. From a study of the distribution of the particles scattered at different angles, results of first importance emerged. It was found that the results could be explained only if the electric forces between the a-particle and charged nucleus followed the law of inverse squares for distances apart of the order of 10-11 cm. Darwin pointed out that the variation of scattering with velocity was explicable only on the same law. This is an important step, for it affords an experimental proof that, at any rate to a first approximation, the ordinary law of force holds for electrified bodies at such exceedingly minute distances. It was also found that a resultant charge on the nucleus measured in fundamental units was about equal to the atomic number of the element. In the case of gold this number is believed from the work of Moseley to be 79.

Knowing the mass of the impinging a-particle and of the atom with which it collides, we can determine from direct mechanical principles the distribution of velocities after the collision, assuming that there is no loss of energy due to radiation or other causes. It is important to notice that in such a calculation we need make no assumption as to the nature of the atoms or of the forces involved in the approach and separation of the atoms. For example, if an a-particle collides with

another helium atom, we should expect the a-particle to give its energy to the helium atom, which could thus travel on with the speed of the a-particle. If an a-particle collides directly with a heavy atom-e. g., of gold of atomic weight 197-the a-particle should retrace its path with only slightly diminished velocity, while the gold atom moves onward in the original direction of the a-particle, but with about one fiftieth of its velocity. Next, consider the important case where the a-particle of mass 4 makes a direct collision with a hydrogen atom of mass 1. From the laws of impact, the hydrogen atom is shot forward with a velocity 1.6 times that of the direction, but with only 0.6 of its initial speed. Marsden showed that swift hydrogen atoms set in motion by impact with a-particles can be detected like a-particles by the scintillations produced in a zinc sulphide crystal. Recently I have been able to measure the speed of such H atoms and found it to be in good accord with the calculated value, so that we may conclude that the ordinary laws of impact may be applied with confidence in such cases. The relative velocities of the a-particles and recoil atom after collision can thus be simply illus trated by impact of two perfectly elastic balls of masses proportional to the masses of the atoms.

While the velocities of the recoil atoms can be easily calculated, the distance which they travel before being brought to rest depends on both the mass and the charge carried by the recoil atom. Experiment shows that the range of H atoms, like the range of a-particles, varies nearly as the cube of their initial velocity. If the H atom carries a single charge, Darwin showed that its range should be about four times the range of the a-particle. This has been confirmed by experiment. Generally, it can be shown that the range of a charged atom carrying a single charge is mu3R, where m is the atomic weight, and u the ratio of the veloc ity of the recoil atom to that of the a-particle, and R the range of the a-particle before collision. In comparison of theory with experi ment, the results agree better if the index is taken as 2.9 instead of 3. If, however, the

recoil atom carries a double charge after a collision, it is to be expected that its range would only be about one quarter of the corresponding range if it carried a single charge. It follows that we can not expect to detect the presence of any recoil atom carrying two charges beyond the range of the a-particle, but we can calculate that any recoil atom, of mass not greater than oxygen and carrying a single charge, should be detected beyond the range of the a-particle. For example, for a single charge the recoil atoms of hydrogen and helium should travel 4 R, lithium 2.8 R, carbon 1.6 R, nitrogen 1.3 R, and oxygen 1.1 R, where R is the range of the incident a-particles. We thus see that it should be possible to detect the presence of such singly charged atoms, if they exist, after completely stopping the a-particles by a suitable thickness of absorbing material. This is a great advantage, for the number of such swift recoil atoms is minute in comparison with the number of aparticles, and we could not hope to detect them in the presence of the much more numerous a-particles.

In order to calculate the number of recoil atoms scattered through any given angle from the direction of flight of the a-particles, it is necessary, in addition, to make assumptions as to the constitution of the atoms and as to the nature and magnitude of the forces involved in the collision. Consider, for example, the case of a collision of an a-particle with an atom of gold of nuclear charge 79. Assuming that the nucleus of the a-particle and that of the gold atom behave like point charges, repelling according to the inverse square law, it can readily be calculated that, for direct collision, the a-particles from radium C, which is turned through an angle of 180°, approaches within a distance D=3.6 X 10-12 cm. of the center of the gold nucleus. This is the closest possible distance of approach of the a-particle, and the distance increases for oblique collisions. For example, when the a-particle is scattered through an angle of 150°, 90°, 30°, 10o, 5o, the closest distances of approach are 1.01, 1.2, 2.4, 6.2, 12 D respectively.

In the experiments of Geiger and Marsden,

the number of a-particles scattered through 5° was observed to be about 200,000 times greater than the number through 150°. The variation with angle was in close accord with the theory, showing that the law of inverse squares holds for distances between 3.6X10-12 cm. and 4.3 X 10-11 cm. in the case of the gold atom. The experiments of Crowther in 1910 on the variation of scattering of B-rays with velocity indicate that a similar law holds also in that case, and for even greater distances from the nucleus.

We have seen that Marsden was able by the scintillation method to detect hydrogen atoms set in swift motion by a-particles up to distances about four times the range of the incident a-particle. In Marsden's experiments a thin-walled glass tube filled with radium emanation served as an intense source of rays. Since the lack of homogeneity of the a-radiation and the absorption in the glass are great drawbacks in making an accurate study of the laws controlling the production of swift atoms by impact, I have found it best to use for the purpose a homogeneous source of radium C by exposing a disc in a strong source of emanation. Fifteen minutes after removal from the emanation the a-rays from the disc are practically homogeneous, with a range in air of 7 cm. By special arrangements very intense sources of a-radiation can be produced in this way, and in the various experiments discs have been used the y-ray activity of which has varied between 5 to 80 milligrams of radium. Allowance can easily be made for the decay of the radiation with time.

In the experiments with hydrogen the source was placed in a metal box about 3 cm. away from an opening in the end covered by a thin sheet of metal of sufficient thickness to absorb the a-rays completely. A zinc sulphide screen was mounted outside about 1 mm. away from the opening, so as to allow for the insertion of absorbing screens of aluminium or mica. The apparatus was filled with dry hydrogen at atmospheric pressure. The H atoms striking the zinc sulphide screen were counted by means of a microscope in the

usual way. The strong luminosity due to the B-rays from radium C was largely reduced by placing the apparatus in a powerful magnetic field which bent them away from the

screen.

If we suppose, for the distances involved in a collision, that the a-particle and hydrogen nucleus may be regarded as point charges, it is easy to see that oblique impacts should occur much oftener than head-on collisions, and consequently that the stream of H atoms set in motion by collisions should contain atoms the velocities of which vary from zero to the maximum produced in a direct collision. The slow-velocity atoms should greatly preponderate, and the number of scintillations observed should fall off rapidly when absorbing screens are placed in the path of the rays close to the zinc sulphide screen.

A surprising effect was, however, observed. Using a-rays of range 7 cm., the number of H atoms remained unchanged when the absorption in their path was increased from 9 cm. to 19 cm. of air equivalent. After 19 cm. the number fell off steadily, and no scintillations could be observed beyond 28 cm. air absorption. In fact, the stream of H atoms resembled closely a homogeneous beam of a-rays of range 28 cm., for it is well known that, owing to scattering, the number of a-particles from a homogeneous source begin to fall off some distance from the end of their range. The results showed that the H atoms are projected forward mainly in the direction of the a-particles and over a narrow range of velocity, and that few, if any, lower velocity atoms are present in the stream.

If we reduce the velocity of the a-particle by placing a metal screen over the source, it is found that the distribution of H atoms with velocity changes, and that the rays are no longer nearly homogeneous. When the range of the a-rays is reduced to 3.5 cm., the absorption of the H atoms is in close accord with the value to be expected from the theory of point charges. It is clear, therefore, that the distribution of velocity among the H atoms varies with the speed of the incident a-particles, and this indicates that a marked change takes place in the distribution

and magnitude of the forces involved in the collision when the nuclei approach closer than a certain distance.

In addition to these peculiarities, the number of H atoms is greatly in excess of the number to be expected on the simple theory. For example, for the swiftest a-rays the number which is able to travel a distance equivalent to 10 cm. of air is more than thirty times greater than the calculated value. The variation in number of H atoms with velocity of the incident a-particle is also entirely different from that to be expected on the theory of point charges. The number diminishes rapidly with velocity, and is very small for a-particles of range 2.5 cm.

It must be borne in mind that the production of a high-speed H atom by an a-particle is an exceedingly rare occurrence. Under the conditions of the experiment the number of H atoms is seldom more than 1/30,000 of the number of a-particles Probably each a-particle passes through the structure of 10,000 hydrogen molecules in traversing one centimeter of hydrogen at atmospheric pressure, and only one a-particle in 100,000 of these produces a high-speed H atom; so that in 109 collisions with the molecules of hydrogen the a-particle, on the average, approaches only once close enough to the center of the nucleus to give rise to a swift hydrogen atom.

We should anticipate that for such collisions the a-particle is unable to distinguish between the hydrogen atom and the hydrogen molecule, and that H atoms should be liberated from matter containing free or combined hydrogen. This is fully borne out by experiment.

From the number of H atoms observed it can be easily calculated that the a-particle must be fired within a perpendicular distance of 2.4 X 10-13 cm. of the center of the H nucleus in order to set it in swift motion. This is a distance less than the diameter of the electron, viz. 3.6 X 10-13 cm. The general results obtained with a-rays of range cm. are similar to those to be expected if the a-particle behaves like a charged disc, of radius of about the diameter of an electron,

7