The converse of Fatou's theorem for Zygmund measures (Q1568897)

The classical Fatou theorem states that a bounded analytic or harmonic function in a half-plane has a non-tangential boundary limit at almost every boundary point. But this statement says nothing about the existence of the limit at a given boundary point. Theorems concerned with the existence of the boundary limit at a given point are called pointwise Fatou theorems. Actually, Fatou gave sufficient conditions for the boundary limit at a point to exist; his conditions imply the existence of the limits almost everywhere. These sufficient conditions are stated in terms of the boundary measure, which appears in the integral representation (like the Poisson integral) of the function. So, the Fatou theorem is of Abelian nature. The converse result, as any Tauberian theorem, requires some additional assumptions to hold, and the simplest of such a Tauberian condition is the positivity of a function. The necessity of Fatou's sufficient conditions for positive harmonic functions was proved by \textit{L. H. Loomis} in 1943 [Trans. Am. Math. Soc. 53, 239-250 (1943; Zbl 0061.23305)]; for the detailed references concerning pointwise Fatou theorems see, e.g., the paper under review or the reviewer's paper [Potential Anal. 3, 379-389 (1994; Zbl 0807.35049)]. The present authors consider another Tauberian condition for pointwise Fatou theorems. Namely, a complex measure \(\mu\) on \(R\) is called a Zygmund measure, if there exists a constant \(C>0\) such that \(|\mu (I)-\mu (I')|\leq C|I|\) for any two adjacent intervals \(I,I'\) of the same length. For Zygmund measures the authors prove that if the Poisson integral of \(\mu\) for a half-plane has a finite non-tangential limit \(L\) at a point \(x_0\in R\), then \(D\mu (x_0)=L\), where \[ D\mu (x)=\lim _{t-s\rightarrow 0,s<x<t} \frac{\mu ((s,t))}{t-s}. \] An analoguous result on normal limits is stated in terms of the symmetric derivative \[ D_{sym}\mu (x)=\lim _{h\rightarrow 0} \frac{\mu ((x-h,x+h))}{2h}. \] The case of infinite limits is considered as well; the criteria in this case are different.