>All my programs use tables on mapped drive x:
>when the main server goes down all machines are mapped a new x: to that server. If I hard code addresses, I would have to change all references to them. I am using win98 SE. The server is simply a workstation. Data and forms are backed up to other work stations. There are nly 10 work stations using this application. If all fails, I can map to a CD and run everything from there. I need only to poll all machines for one file datestamp and remap to it.

In that case that's fine - as long as you can access all the workstations. For remote maintenance, I've found that W9x drive mappings are unreliable - if the machine to map to is not up when the workstation boots, there's a messagebox where the user can click to unmap the drive, so I had to have ways around it. In the days when FPD apps were used under W9x, it was batch files which mapped the drives and then ran the apps. Nowadays, I'd probably have some .ini file locally which would contain the full UNC path to the app.

Well, I figure this is all rhetorical stuff - your problem is solved (knock on wood that it doesn't return in worse form), and that's what matters.