Let X be a strongly symmetric recurrent Markov process with state space S a
nd let L-t(x) denote the local time of X at x is an element of S. For a fix
ed element 0 in the state space S, let
tau (t) := inf{s: L-s(0) > t}.
The 0-potential density, u({o})(x, y), of the process X killed at T-0 = inf
{s: X-s = 0), is symmetric and positive definite. Let eta = {eta (x); x is
an element of S} be a mean-zero Gaussian process with covariance
E-eta(eta (x)eta (y)) = u({0})(x, y).
The main result of this paper is the following generalization of the classi
cal second Ray-Knight theorem: for any b is an element of R and t > 0,
{L-r(t)(x) + 1/2(eta (x) + b)(2); x is an element of S} = {1/2(eta (x) + ro
ot 2t + b(2))(2); x is an element of S} in law.
A version of this theorem is also given when X is transient.