Binned observables multiplied by abs(x)?

[Christian Speckner]

To me, it seems like binning an observable O(x) over x results in a histogram of abs(x)*O(x). As an example, I have attached the result of binning \theta (polar angle) and \cos\theta as functions of \theta for a constant 2->2 matrix element with massless phasespace (attached together with the sindarin input and the hacked matrix element). Contrary to the expectation of seeing distributions like \sin\theta resp. 1, the result looks like \abs\theta*\sin\theta resp. \abs\theta. I've attached the faulty histograms together with the WHIZARD 1.95 result (which looks correct to me).