On the supports of measure-valued critical branching Brownian motion (Q5899689)

In the paragraph preceding the statement of Theorem 3 in the introductory section of the original article, ibid. 16, No.1, 200-221 (1988; Zbl 0635.60094), a discussion of the almost-sure behaviour in the two- dimensional case is conspicuously absent. The result which fills this gap is the following one: If \(\emptyset \neq G \subset {\mathbb{R}}^ 2\) is bounded and open and \(X_ 0=\lambda\) (Lebesgue measure), then with probability 1, \(X_ t(G)\nrightarrow 0\) as \(t\to \infty\). Indeed if \(\lim_{t\to \infty}X_ t(G)=0\) on a set \(\Omega_ 0\) of positive probability, then \(\lim_{t\to \infty} t^{-1}\int^{t}_{0}X_ s(G)ds=0\) on \(\Omega_ 0\) as well. This however contradicts Theorem 2 of the author in Stochastics 18, 197- 243 (1986; Zbl 0608.60077), which implies that \(t^{- 1}\int^{t}_{0}X_ s(G)ds\) converges weakly, as \(t\to \infty\), to \(\xi\) \(\cdot \lambda (G)\) where \(\xi\) is a strictly positive random variable.