Some variants of the Feynman alpha method in critical and accelerator driven sub critical systems

Some problems related with the Feynman variance to mean ( v / m ) method commonly used in reactor noise analysis are studied. These are: (i) an exact analysis of the effect of an extending (paralysable) dead time in the circuit; (ii) an approximate analysis of the same problem for non extending (non paralysable) dead time; (iii) effect of the non Poisson character of the neutron source — a problem of current interest in the context of accelerator driven systems. Our methods and results are different from those obtained in some recent publications.