Log-similarity for turbulent flows?