We describe the long-term evolution of a bipolar non-Hale active region that was observed from 1995 October to 1996 January. During these four solar rotations the sunspots and subsequent flux concentrations, during the decay phase of the region, were observed to move in such a way that by December their orientation conformed to the Hale-Nicholson polarity law. The sigmoidal shape of the observed soft X-ray coronal loops allows us to determine the sense of the twist in the magnetic configuration. This sense is confirmed by extrapolating the observed photospheric magnetic field, using a linear force-free approach, and comparing the shape of computed field lines with the observed coronal loops. This sense of twist agrees with that of the dominant helicity in the solar hemisphere where the region lies, as well as with the evolution observed in the longitudinal magnetogram during the first rotation. At first sight the relative motions of the spots may be misinterpreted as the rising of an Ω loop deformed by a kink instability, but we deduce from the sense of their relative displacements a handedness for the flux-tube axis (writhe) that is opposite to that of the twist in the coronal loops and, therefore, to what is expected for a kink-unstable flux tube. After excluding the kink instability, we interpret our observations in terms of a magnetic flux tube deformed by external motions while rising through the convective zone. We compare our results with those of other related studies, and we discuss, in particular, whether the kink instability is relevant to explain the peculiar evolution of some active regions.