The Lee, Oehme and Yang (LOY) theory of time evolution in two state subspace
of states of the complete system is discussed. Some inconsistencies in
assumptions and approximations used in the standard derivation of the LOY
effective Hamiltonian, H(LOY), governig this time evolution are found.
Eliminating these inconsistecies and using the LOY method, approximate formulae
for the effective Hamiltonian, H(eff), governing the time evolution in this
subspace (improving those obtained by LOY) are derived. It is found, in
contradistinction to the standard LOY result, that in the case of neutral kaons
( - ), cannot take the zero value
if the total system the preserves CPT--symmetry. Within the use of the method
mentioned above formulae for H(eff) acting in the three state (three
dimensional) subspace of states are also found.