>I seriously doubt that VFP could do that.
>The OS, on the other hand, might be a good candidate, especially if write cache is enabled (workstation or server) or OPLocking is active, an intermittent RAM problem and such.

Well, speaking loosely, I'm saying VFP did it in this sense:

1) VFP calls Windows and requests a write of a set of data (a cursor) to the temporary directory.

2) Some unexplained failure causes this set of data to be directed towards the network file.

In other words, it was not a network failure and it was not a server failure.

Tomorrow I will check out write caching and OPLocking.

BTW, the first record containing garbage in the trashed file corresponded to a record in the restored table which has not changed since December 1997.

Peter