Table of contents

A description is given of a photoelectric recording spectrometer for Raman spectroscopy incorporating a high-intensity excitation system and a novel amplification device. Operation of the exciting mercury arc lamps from 50 c/s. mains imparts to the light signal, and consequently to the current from the photo-multiplier tube, a 100 c/s. periodicity. After amplification by a narrow band-pass amplifier, rectification is achieved in a push-pull phase-sensitive (homodyne) rectifier, by "mixing" the signal with a fixed voltage of identical periodicity. The rectified signal is fed through a filter to a cathode follower valve which provides the power to operate a recording milliammeter. This method of rectification, in conjunction with appropriate filter, effectively reduces the bandwidth of the amplifier, with consequent great improvement in the signal/noise ratio. With this arrangement cooling of the photo-multiplier is unnecessary for normal work. Using a spectrometer of small aperture and resolution and a Raman tube of 35 ml. capacity, traces of the Raman spectrum of carbon tetrachloride show no appreciable background noise.

The photographic plate method has been used in an investigation of the neutrons from the reaction ₃⁷Li + ₁²D → ₄⁸Be + ₀¹n.

The results are in general agreement with those of Richards, but show the structure of the neutron groups more clearly. In particular, the shape of the broad group attributed by Richards to a level in ⁸Be at 3.0 ± 0.4 MeV. is discussed in detail.

The total energy release is 15.0 ± 0.15 MeV., and evidence is put forward for the existence of energy levels in ⁸Be at 2.8 MeV., 4.05 MeV., 4.9 MeV., and 7.5 MeV.

It is shown that the collective electron and London-Heitler models are not to be regarded as different approximations to the same exact wave function for solids in which, according to the former model, there is a partially filled zone of energy levels. It can thus be shown why nickel oxide in the pure state is a non-conductor, although it contains an incomplete zone. The properties of the metals nickel, palladium and platinum are discussed in the light of these results; platinum differs from nickel in that the orbital contribution to the moment of the elementary magnets is not quenched. A discussion is given of x-ray absorption edges, and it is shown why exciton lines are absent for metals.

The principles outlined in previous parts, published in the Proceedings of the Physical Society in 1946 and 1948 respectively, which dealt with the direct current circuit and with electrostatics, are here applied to magnetism, with the object of showing how the basic concepts are defined in terms of the operations actually performed in measuring them. This part is confined to a discussion of the vector B, which is shown to be measurable independently of a knowledge of any other magnitude. It is solenoidal, and can therefore be determined at all points, even those within solid bodies.

This paper concludes the inquiry which was undertaken by N. R. Campbell and the author with the object of elucidating the connection between the concepts and principles of electromagnetism and the experimental operations actually performed in the laboratory that constitute their factual basis. In Part III the vector B was established as measurable everywhere, even within solid bodies. The vector H and the scalar μ=B/H are now shown to be measurable in special circumstances by means of the magnetometer and permeameter, but in general their values depend on a hypothesis, which is stated. The significant facts concerning the magnetic properties of real materials are briefly reviewed, and the two-fold aspect of the science is emphasized: the self-consistent mathematical theories based on postulates, on the one hand, and the complex of experimental laws on the other. It is important that the relation between the two shall be clearly established, and not merely assumed as self-evident.

An experimental method of preparing specimens of N1O in the form of coherent strips is described. Simultaneous measurements of electrical conductivity σ and Hall effect R are then made at various temperatures up to about 700° c. and the thermo-electric power dE/dT measured immediately afterwards, all on one sample. The experiments are then repeated on different samples, and they show that log σ, log R, and dE/dT are linear functions of 1/T, where T is the mean absolute temperature of the specimen. The graphs for log σ and log|R| against 1/T each consist of two rectilinear portions, one covering the high, the other the low temperature results. Thermo-electric measurements yielded results only for the high temperature part.

The theory of impurity semiconductors is developed and found adequate to explain the variation, at any rate at higher temperatures. It is found that an activation energy of about 2 ev. at high temperatures, and from 0.3 to 0.64 ev. at low temperatures, is required for excitation of electrons to the conduction band. The concentration of impurity centres operative in the high temperature range is about 8.6 × 10²⁰, the electrons having a mean free path of about 10^-6 cm. The concentration of free "carriers" varies from 10¹¹ to 10¹³ as the temperature rises. NiO behaves like a defect semiconductor, "hole" conduction predominating. The significance of these results is discussed.