The following problem is solved...  But it's a wierd one and I'm 
wondering if it will repeat.

My wife's little PII 400/100.  RH9 install.  Was working fine this 
morning, then she heard a noise "like the neighbors hammering nails in 
the wall but it was in the computer" (I wasn't there).  Then her network 
connectivity stopped.  Standard M$ programmed response, she rebooted.  
On reboot, she didn't even get GRUB, just a black screen with blinking 
cursor where and when GRUB should have been.  I tried a few times, same 
black screen.  Tried Knoppix:  it didn't see my primary master!  
Massaged the IDE connectors and power connectors, retried Knoppix, ahh 
the partitions are back.  Did an fsck -c -f on all, clean.  But 
rebooting still gave me a black screen.  Rescue disk RH9, tried both 
reconfig and reinstall of both GRUB and LILO, same black screen.  Wiped 
the partitions except /home and installed Fedora.  Same black screen. 

Then I noticed something in the BIOS that I never noticed before: you 
can configure the order in which the "hard drives" are accessed to find 
an MBR.  I don't mean 1) removable 2) CDR 3) HD 4) net, I mean inside 
option 3, there are three more options, which was configured 1) Bootable 
add-in cards 2) Maxtor (SM) 3) Maxtor (PM).  I don't have any bootable 
cards (that I know of?).  Secondary Master is the wrong HD on this 
system, the MBR is on the Primary Master.  So I switched to exactly the 
reverse (PM,SM,bootable), and wow, GRUB came back! 

Okay...  So how can this be analyzed?  Did the RH install work on the 
old (bootable,SM,PM) configuration then suddenly stop working?  Or was 
it set to PM,SM,bootable and somehow reset?  Because there was some 
physical noise, maybe some physical problem caused the BIOS to reset?  
Maybe if a drive is physically yanked, the BIOS setting automatically 
does some corrections, then did the wrong correction back when the HD 
was restored?  Anyone else have this kind of wierdness?

--Alexander