Measure theoretic aspects of the finite Hilbert transform

The finite Hilbert transform $T$, when acting in the classical Zygmund space $\logl$ (over $(-1,1)$), was intensively studied in \cite{curbera-okada-ricker-log}. In this note an integral representation of $T$ is established via the $L^1(-1,1)$-valued measure $\mlog\colon A\mapsto T(\chi_A)$ for each Borel set $A\subseteq(-1,1)$. This integral representation, together with various non-trivial properties of $\mlog$, allow the use of measure theoretic methods (not available in \cite{curbera-okada-ricker-log}) to establish new properties of $T$. For instance, as an operator between Banach function spaces $T$ is not order bounded, it is not completely continuous and neither is it weakly compact. An appropriate Parseval formula for $T$ plays a crucial role.
Comment: This is the final version, to be published in Mathematische Nachrichten