become very much smaller, indicating that the charge upon the particles is decreasing. (3) a. When the position of minimum viscosity is reached the sol sets spontaneously to a rigid gel within the course of a few hours. b. An X-ray dose sufficiently in excess of that required to produce the maximum decrease will set the sol to a gel immediately. (4) In a particular sol studied the particles decreased in size up to the position of minimum viscosity. The percentage decrease in size was almost equal to that in viscosity. Beyond the minimum point the particles increased in size, reaching a maximum at the setting point of 1.6 times the normal value. In conclusion, I should like to thank Professor Crowther for his particular interest, kind help, and advice. He has intimated his intention, in a future paper, to review and comment upon the researches carried out in this laboratory during the past two years on the colloidal state of matter. I should also like to express my indebtedness to the Board of Scientific and Industrial Research for a grant which has made this work possible. Department of Physics, XLII. On the Thermal Measurement of X-ray Energy. By J. A. CROWTHER, M.A., Sc.D., F.Inst.P., Professor of Physics in the University of Reading, and W. N. BOND, M.A., D.Sc., F.Inst.P., Lecturer in Physics". Introduction. THE experiments described in this paper were com menced early in 1926 to determine the relation. between the ionization produced by a beam of X-rays and the energy of the beam. The matter is of importance both from a theoretical and a practical aspect. In the first place, early experiments, as for example those of Rutherford and McClung, indicated that the work spent per ion during Communicated by the Authors. Phil. Mag. S. 7. Vol. 6. No. 36. Sept. 1928. 2 D ionization by X-rays was considerably greater than the ionizing potential of the gas ionized, and it is desirable that this result should be tested further. In the second place, it is generally agreed that the most suitable method of measuring quantities of X-rays is by the ionization they produce in some suitable form of ionization chamber, and it is clearly desirable that experiments should be made to determine exactly what it is that an ionization chamber measures. At the time when we commenced our work there seemed to be no adequate data on this point, and it was hoped that such data might be supplied by the experiments we had in mind. Neglecting secondary effects which may be produced by the presence of the electrodes used for measuring the ionization current, the number of electrons produced per unit volume of air by the passage of a beam of X-rays can depend only on two factors, the energy conveyed by the beam, and the wave-length of the radiation, and can be written in the form I (E, λ). If the values of E and I can be determined for a sufficient number of values of λ, the nature of the function can be determined. The ionization, I, can be determined in some suitable ionization chamber, and the energy E by absorbing the radiation completely in a calorimeter and measuring the energy in the form of heat. This is what we set out to do. = The necessity for such measurements seems to have become apparent to other experimenters at about the same time, and during the progress of our experiments a number of such determinations have been published, amongst others by Kriegesmann *, Kircher and Schmitz †, Kulenkampff‡, and Rump §. It appeared at first sight that further work on our part might be superfluous. Unfortunately, however, there was such strong disagreement between the values published by the different authors that further investigation became imperative. Thus Kriegesmann finds that the work spent in producing a pair of ions varies from 87.8 volts for X-rays of mean wave-length 0.397 Å.U. to 132 volts for a wave-length of 0.166, and states that the "volts" required to produce an ion-pair increases as the wave-length diminishes. Kircher and Schmitz, on the other hand, give a value of 21 volts per ion-pair, and find that it is independent of the wave-length. Kulenkampff, in a very long and elaborate * Kriegesmann, Zeit. für Phys. xxxii. p. 542 (1925). § Rump, Zeit. für Phys. xliii. p. 254 (1927), and xliv. p. 396 (1927). paper, agrees that the energy spent is independent of the wave-length, but gives its value as 35±5 volts per pair of ions. Finally, Rump gives values for the same quantity ranging from 300 volts for long wave-length radiation down to about 40 volts for filtered radiation from a hard tube. It was clear, therefore, that in spite of much recent work, the constant required was still in very considerable doubt. It was also clear that its determination was full of pitfalls, and that great care would be necessary to avoid both errors of measurement, and errors in evaluation of the constant from the measurements taken. General Plan of the Experiment. The radiation was produced by a Shearer tube working on a two-kilowatt transformer, and having a molybdenum anticathode. Two windows in the tube, closed with aluminium foil 1/20 mm. thick, allowed two narrow pencils of rays, one vertical, the other horizontal, to leave the tube. The horizontal beam passed through a small ionization chamber A, having parallel plate electrodes of aluminium. This chamber was used for comparison purposes only, as a means of eliminating the effect of fluctuations in the output; the actual measurements used in calculation were all made on the vertical beam. The tube was controlled by adjusting the supply of air to it by means of a micrometer air-leak. As the tube is self-rectifying its condition cannot be determined with any accuracy by means of a spark-gap, since the spark-gap measures only the peak potential in the circuit, which, in the case of a self-rectifying tube, occurs in the half wave which does not pass through the tube. The peak potential, when the tube is taking a current, will clearly be less, and may be considerably less than this. To overcome this difficulty a milliammeter with an open scale was inserted in the tube circuit, and an A.C. voltmeter was connected across the primary terminals of the transformer. It was found, experimentally, that a given pair of readings on these instruments corresponded to a definite state of the tube. If, for example, the voltmeter was set, by adjusting the primary resistance, to read 155 volts, and the milliammeter, by adjusting the leak, to read 3 milliamperes, the output from the tube was of definite quality and intensity, which was found to be reproducible at any time to an accuracy of at least 1 per cent. This method of regulating the tube is much more convenient than the use of a sparkgap, and seems to be at least as accurate. The anticathode end of the tube was efficiently watercooled. It was, however, of the utmost importance to prevent any thermal radiation from the tube from reaching the calorimeter in which the X-ray energy was measured. A "water-sheet" was therefore introduced between the tube and the calorimeter. This consisted of two parallel sheets of stout brass with a space about 1 mm. thick between them, through which a constant stream of water could be passed. A circular hole, 1.007 cm. in diameter, was pierced through the double sheet, the holes being covered with thin celluloid sheets to make the system watertight. This hole came immediately below the window of the tube, and served as the stop limiting the vertical beam passing into the calorimeter. Thus all the energy passing through the "water-sheet" passed directly into the calorimeter. Since it is convenient that the horizontal and the vertical beams should be of the same composition, sufficient thickness of cellophane was introduced into the horizontal beam to equalize the absorption of radiation in the vertical beam due to the water-sheet. The efficiency of the water screen is discussed in a later section. After penetrating the water screen the vertical beam passed either into the calorimeter, which is described later, or into the ionization chamber which was used as a standard for the experiment. This chamber was furnished with a shutter, actuated by a carefully graduated micrometer screw, which limited the beam entering the chamber. The collecting electrode of this chamber could be connected, when desired, to the same electroscope as that attached to the comparison chamber A. A standard mica condenser could also be connected to the electroscope when necessary. The details of the ionization measurements will be discussed more fully in a later section. The course of the experiments was as follows. The calorimeter was inserted in the path of the rays. The standard ionization chamber was cut out of the electroscope system, leaving the comparison chamber A attached to the electroscope and to the standard condenser. The tube was then excited, and kept working in its standard condition until the deflexion of the electroscope indicated a rise in potential of approximately one volt. The actual voltage was then measured by means of a potential divider in the usual way. The total charge which the X-rays had allowed to pass across the comparison chamber during the run could thus be measured. The condenser used was a subdivided microfarad, and the charge conveyed across the ionization chamber in a single run was usually of the order of a microcoulomb. As evidence of the certainty with which the tube condition could be reproduced, it may be mentioned that the time taken to produce a charge of one microcoulomb on the condenser for a given setting of the tube did not vary by more than a few seconds in about eight minutes, over a period of some months. In this way determinations were made of the heat produced in the calorimeter in relation to the charge conveyed across the comparison chamber. The calorimeter was now removed and the rays allowed to fall on the shutter of the standard chamber, which was now connected to the electroscope system. The standard chamber was charged to a potential of opposite sign to that of the comparison chamber, and the area of the shutter was adjusted until a balance was obtained in the electroscope with the rays passing through both chambers. In order to increase the sensitivity, the condenser was removed from the system during balancing experiments. It was possible to balance the two currents to an accuracy of one part in a thousand. Balances on different occasions, however, were only consistent to about one per cent. These variations probably indicate the limits of certainty to which the tube conditions can be controlled. Assuming the whole area of cross-section of the vertical beam in the plane of the shutter to be known, the ionization which would have been produced if the whole of the X-ray energy which is absorbed in the calorimeter had passed through the standard chamber, is equal to the ionization as measured in the comparison chamber multiplied by the ratio of the cross-section of the whole beam to the aperture of the shutter. The results of these determinations are given in Tables III. and IV., and a more detailed account of the observations in the following section. The Energy Measurements. The accuracy of the energy measurements depends on our being able to absorb the whole of the incident X-ray energy, and to transform it into heat at the proper place. Owing to the phenomena of scattered, fluorescent, and corpuscular radiation, it is impossible to find a surface which will do this. The only reliable method is to pass the radiation into a box, the aperture of which subtends a sufficiently small angle at the surface on which the radiation actually falls. From the point of view of ease of measurement a thermocouple made of thin foil presents |