1. In this paper we develop methods that are designed for writing out the distributions of functionals of the Brownian motion stopped at the moment when the local time reaches a given value on one of several levels. Such methods were considered for one level in [1,2], and for two levels in [3]. Let w(s) be the Srownian motion, E(w(s) -w(0)) 2 = s. By E x we shall denote the mathematical expectation with respect to the Brownian motion w(.) under the condition w(0) = x. The limit f0 t(s,y) = lim 1_ llt~,~+e) (w(u)) du,where ~IA(.) is the indicator function of a set A, exists with probability one (see [4]) and is called the Brownian local time at the point y up to time s. With probability one, the Brownian local time process t(s, y) is continuous with respect to (s, y) E [0, oo) x t/1.The inverse local time is the random stopping time e(t,z) = min{ s: t(s,z) = t}. Fix k levels zl,z2,... ,zk and associate to each level zt some time value tl. We are interested in the first moment, when the Brownian local time reaches at one of these levels z, the preassigned value tl. This stopping time can be defined in the following way: r = min{e(tl, zl): l= 1,2,...,k}, where t' and ~' denote the vectors (tl,t2,...,tk) and (z~, z2,...,zk). In this paper we consider additive functionals of the form .40)= c,tO, zl),where f is a non-negative function, ct >1 0, zl E R 1 , m < 00. In a certain sense the Brownian local time t(s, y) is an elementary additive functional of the Brownian motion. Namely, the following relation is valid: /0" s
f(w(u)) du = f(y)t(s,y)dy.Oo