1. S. Franchetti:Nuovo Cimento D,6, 131 (1985).
2. A. Rahman, M. J. Mandell andJ. P. McTague:J. Chem. Phys.,64, 1564 (1975).
3. At the root of the difficulties was the imperfect knowledge of the real distribution of the radii of the shells, only tentatively replaced by the main peak of the radial distribution function.
4. On the other hand, the agreement between the calculated and the real spectrum in the high-frequency region, which—together with the general form of the spectrum—was the most significant result obtained in(1), is practically unaffected by the change of normalization, when coupled to a change in a second open parameter, namely the average correlation coefficientC between atomic displacements which alters from 0.23 to an equally plausible 0.20.
5. The classical value 2 for the exponents should be expected to hold only for ε→0,i.e. for very large wavelengths for which the solid behaves as a continuum.