The popular (quick) way was to put an RTS in the second cassette buffer with a SYS in the code pointing to it. (on the PET) and save the program from that to the end using the MLM. Then if someone SAVEs it regular and tries it it just crashes. The most ingenious PET protection was to put a bit of code in the FIRST cassette buffer, so you had to save it from the second cassette port using the MLM to make a runnable copy. Second would be to plug a bit of code into the chargot (I think) routine in low memory which will trick the computer in running the program once loaded (this could easily be bypassed on the PET, but on the 64, it was pretty fool-proof. On the 64 if you can get to the vectors before the user can try SAVing, you change the jump address for SAVE (locations 818,819) to your own location. The character varys on the computer. I think Transactor or Compute did a table of em once. So was this to twart those modifying the code, or tracing it?... Oh... add a REM shift-l after it, once the 'fix' it it still doesn't work. Only problem with the delete character is if you list to a printer they will all show up. That same PET program had an ingenious listing protection too, the first two lines were only listable and they were effectively masked by rems with a bunch of delete characters. What the programmer did was break the line links after the second line so the computer thought it was a two line program. But when run the two lines restored the link via POKEs, once it was running past that point, it restored the line break. If you are curious to see this marvel in action it is on-line; check out my Flash Attack page and download either of the PET versions, this was written by Tim Stryker years before he founded Galacticomm.