The amplitudes produced by interp='aipslin' in accum appear to be mis-scaled by an order of magnitude.
Oops. The 'aipslin' mode (emulating what classic aips does when applying a CL table to data) interpolates the amplitudes linearly, and derives the interpolated phase from the linear interpolation of the complex solutions. The complex interpolation wasn't being divided by its own amplitude before being multiplied by the separately interpolated amplitude result. Thus, for interpolation over small phase ranges, the amplitude interpolation was approximately the square of what it should be. This will be patched into the stable Monday night.-- GeorgeMoellenbrock - 18 Dec 2006