Program iner() (SEP-61 page 399) found prediction and interpolation filters by minimizing the output power for a given input data. Program miss1() found partly unknown data by minimizing output power from a given filter. Program missf() found unknown data and an unknown filter by minimizing output power. Now we'll do that again except the filter will be constrained to be of the form of a plane-wave destructor (PWD) which is a simple two-dimensional filter. It is little inconvenience that a PWD is a two dimensional filter. A more serious inconvenience is that in practice the data space is a wall of data and the plane-wave model is only valid in regions of limited extent. We have already solved this problem without the missing data. Effectively the p for each window is independent of the others. The residuals were another matter, we had to add together the partially overlapping residuals, keeping track of how many windows overlapped each point in space. Now the question is, can I splice all this stuff together?