This is what happens when you do a thin-plate-spline interpolation with the theta radius set too small.
Happily, I figured out what was wrong almost immediately and was able to fix it in very short order, rather than banging my head against the wall for five hours. And now the data is all normal. And... kinda boring.
Okay, so now I feel the need to know what happens if you set it ( the TR) too *high*? Im sure its not as interesting, but I'm curious. ( Also, i think I had a shirt with that color pattern in the 80s...)