“…Note that for s -> 0+, the argument of m in (2.5) converges to (F(r), F(t), 9{t), 0, 0) and the limit (2.5) for j-»0+ is therefore consistent with the normalization (2.2), which is assumed to hold only for 5 > 0, as m is not assumed to be defined for 5 = 0. Definition 2.1 differs only in technical details from the definitions given in [7,8], The domination conditions say first of all that the influence of the remote past is small; technically, they are adjusted so as to ensure that all the integrals occurring in the statements and proofs of the results are absolutely convergent. In particular, the assumption that m itself is dominated ensures that the integral in (1.1) is convergent for every process and every time; the dominance of the partial derivatives of m with respect to F, H, 6 , and G ensures the continuous differentiability of p. with respect to t and that the time derivative of the integral in (1.1) is given by the time derivative of the integrand with respect to t. The dominance of sdHm, sdGm permits one first of all to define the kinetic coefficients K, L by formulas (3.6), (3.7) in Sec.…”