1. Ku?, M., Scharf, R., Haake, F.: Z. Phys. B ? Condensed Matter66, 129?134 (1987)
2. Mehta, M.L.: Random matrices and the statistical theory of energy levels. New York: Academic Press 1967
3. Mehta, M.L., Des Cloizeaux, J.: Indian J. Math.3, 329 (1971)
4. Dyson, F.J.: Commun. Math. Phys.47, 171?183 (1976)
5. Actually, (2.6) requires l1?l 2?...?l n . When some of thel 1 coin-cide such thatk different values are assumed, 1?k?n, with multiplicitiesn 1,n 2,...,n k and ?n i =n, the factor 1/n! must be replaced by 1/n 1!n 2!...n k! For the Taylor coefficients,E 1, given in (2.11 or 14), the distinction in question is irrelevant, due to the appearance of another factorn 1!...n k !/n! accompanying the summation over the configurations {l i }