Beam squint and Stokes V with off-axis feeds

Radio telescopes with off-axis feeds, such as the (E) VLA, suffer from "beam squint" in which the two orthogonal circular polarizations sampled have different pointing centers on the sky. Its effects are weak near the beam center but become increasingly important towards the edge of the antenna power pattern where gains in the two polarizations at a given sky position are significantly different. This effect has limited VLA measurements of circular polarization (Stokes V) and introduced dynamic range limiting, wide-field artifacts in images made in Stokes I. We present an adaptation of the visibility-based deconvolution CLEAN method that can correct this defect "on the fly" while imaging, correcting as well the associated self-calibration. We present two examples of this technique using the procedure "Squint" within the Obit package which allows wide-field imaging in Stokes V and reduced artifacts in Stokes I. We discuss the residual errors in these examples as well as a scheme for future correction of some of these errors. This technique can be generalized to implement temporally-and spatially-variable corrections, such as pointing and cross-polarization leakage errors.