>>>Seen something of the kind back in the nineties, on some networks, where the table header, with record count information, was not updated on some workstations, and those would create new record on the place of a recently added record. Something wrong with locking, I'd guess.
>>
>>Back in the dawn of time, I had this issue on occasion - turned out we had a bunch of bad network cards, and whenever one would go south, it would cause interesting things to happen, from table locks (in a buffered system) to garbage in records. Me and our hardware people went round and round about what could be causing these issues until we finally tracked it down to one machine with a bad card. We had about 6 or so cards go south before they finally got clearance to purchase new ones.
>
>You could get rich just living in the south, those years. Things would arrive from nowhere.

True, but they'd arrive broke...and who wants to fix 'em just to watch 'em go back where they came from? But that would help with the traffic around here.