Corrections to the paper "Arithmetic expansions associated with a rotation of the circle and with continued fractions"      Actually, the proof of the Erdös Theorem on the singularity of the Bernoulli convolution parameterized by the golden mean (pages 15-16), is wrong. We attempted to use the initial measure m as a measure invariant under the beta-transformation Ux = {x /a}. In fact, this interesting measure is actually only quasi-invariant under U, which was proved by us in [6] with a lot of important extras. In the same paper we have also given an exact expression for the measure n  which is U-invariant and equivalent to m as well as two new ergodic proofs of Erdös theorem for this case. Thus, not only that we have made amends but have also started a new line of research.