Heya, that turned quite some interesting information. 
I think (and hope) you may find these Xorg logs very interesting, as they certainly surprised me, which is why I attach all the four described log files. (Rest assured, 3 are tiny).

Here is what happened:

first of I had a number of reboots spontaneously landing my X desktop on vt7... bummer :) I did however confirm this way, that indeed switching to vt1 will _not_ garble my primary display output on my nvidia card when X is on vt7 proper.

On the next reboot, I had a garbled text console on vt1, so I knew before looking that the X session was probably not on vt7 that time: indeed it was on tty2! "-isig" was listed, so it wasn't armed and dangerous. However, looking at the X logs was a big surprise:

/var/log$ ls -ltr --full-time /var/log/Xorg.*
-rw-r--r-- 1 root root  6423 2010-09-10 00:40:01.096801267 +0200 /var/log/Xorg.0.log
-rw-r--r-- 1 root root  6423 2010-09-10 00:40:01.272800811 +0200 /var/log/Xorg.1.log
-rw-r--r-- 1 root root 15698 2010-09-10 00:45:17.064801287 +0200 /var/log/Xorg.2.log.old
-rw-r--r-- 1 root root 13521 2010-09-10 00:47:06.201536090 +0200 /var/log/Xorg.2.log

Those .0, .1 and .2 (old) logs are in surprisingly rapid succesion, not? Indeed the first two contain:

(EE) Sep 10 00:40:01 NVIDIA(0): Failed to initialize the NVIDIA kernel module. Please see the
(EE) Sep 10 00:40:01 NVIDIA(0):     system's kernel log for additional error messages and
(EE) Sep 10 00:40:01 NVIDIA(0):     consult the NVIDIA README for details.

just before their respective shutdowns. Only the third attempt (Xorg.2.log.old) worked. Now, this was the one that I actually logged onto, while it was running "-isig" on vt2. I played around a while waiting for a crash opportunity. Unfortunately that took a while. Therefore I switched to vt1, submitted 'stty -F /dev/tty2 isig', chvt 2 and was able to immediately cause X to crash by pressing '2' (sorry for being impatient). As you will be able to see, Xorg.2.log.old then gets terminated with the following sequence:

(II) Power Button: Close
(II) UnloadModule: "evdev"
(II) Power Button: Close
(II) UnloadModule: "evdev"
(II) Logitech Optical USB Mouse: Close
(II) UnloadModule: "evdev"
(II) AT Translated Set 2 keyboard: Close
(II) UnloadModule: "evdev"
(II) Macintosh mouse button emulation: Close
(II) UnloadModule: "evdev"

I suppose the many repeats of 'NVIDIA(0): Setting mode "CRT:1280x1024_75+0+0,DFP:2048x1152+1280+0"' before that are because of my repeated switching of vt1 to vt2 and back before the X session crashed. Also note that I recall seeing a different end sequence (involving the SIGINT keyword) a fair while back. This may be due to several reasons: (a) it might have been different before the allegedly released fix (something has been changed by ubuntu devs a while back) (b) it might have been the trigger key ('2' in this case); I used to only recall crashing with the Enter key, which might result in a different X termination mode :)

One more interesting observation: in none of my X sessions (be it on vt2 or vt8) have I been able to make normal use of the scroll lock light (pressing scroll lock does _not_ cause it to toggle), making the fact that it can sometimes spontaneously flash directly before crashing X a bit more interesting.

Finally, you should know that the last logfile is from the relaunched X session, which is happily running on vt8 right now where I type this report. 

Cheers