Screen burn. How long does it take to affect a tube?

Everyone has seen old monitors/terminals with text clearly burned into the screen, or a grid of fuzzy boxes where the character cells line up, wear from many years of use and abuse. While graphic color tubes can and do develop screen burn (the Macintosh menu bar, and Windows taskbar being commonly found etched into discarded color monitors), the burn-in problem seems more prevelant on older, monochrome tubes. I use a fair number of old monochrome monitors, and a couple of terminals I use every day. Yesterday, I accidentally left a computer and it's monochrome monitor running unattended for about four hours (I forgot to shut it off/got distracted). When I returned, I carefully examined the areas where text had been on the screen, and could not detect any burn-in, although I can tell that the left side of the display has an oh-so-faint browning to it. Now, I am almost positive that this was there all along, after all - this thing lived a hard life before I found it. But, it got me thinking, how long does text need to be on the screen to burn into a monocrhome tube? Four hours doesn't seem like enough time - otherwise every PC monitor would have the WordPerfect ruler burned into it. I did some searching, but of course, could only turn up things relating to modern plasma screen TV's, which apparently suffer similar degredation from use. Were there any studies conducted, or actual ratings for picture tubes like this? The sort of tubes I'm referring to are standard monochrome CRT's, like the ones used in terminals and monitors in the early 80's. On a similar note, I was playing with a system test utility program for the Commodore PET. It tests the RAM and ROM, loops continuously and displays the test results and running time on the screen. The Commodore PET also uses this sort of monochrome picture tube - and unlike most other machines, you can't shut off just the screen and leave the computer running. So, this test program alternates every second or so between normal and inverse video modes. Now, since the screen is flashing, it shouldn't damage the phosphors - since it is my understanding that the phosphor burn phenomenon isn't so much the phosphors wearing out, but related to them being constantly excited, and therefore heating up and darkening. Is this correct thinking? While I don't plan on running this test for extended periods of time, I do wonder about it's potential effect on the screen, if any. It's interesting to see what's already burned into screens though, it gives an insight into what the computer used to do. I have terminals with logon promts burned into them, menu systems, and at least one with a dire confidentiality warning permently etched into the tube. I've got a Monitor IIe with "Apple //e" burned into the top - clearly someone left the computer on without a disk in it for a very long time to get that clear of a burn in. So, like I said, I just got curious about the whole thing, and figured I'd see if anyone else knew more on the subject. Obviously, I always try to be careful and shut things off when I'm not using them, but, you know, mistakes happen. -Ian