This brought all the old overlay stuff back to mind. Early computers were memory-limited, and what you used for code, you couldn't use for data. Even if you use many megabytes for code, you make little impact on the available RAM for data. Hence, you don't need overlaying.
I used to use computers with under 100 kb - and quite successfully. By the time I was using a PC with 1000 kb or 16000 kb I had abandoned overlaying. Now, even a cheap computer is likely to have 2,000,000 kb!
I just finished (is one ever finished?) putting three old programs together with a ClearWin+ GUI. The three programs all retain their complete original data structures, and the GUI operates with a fourth data structure that is independent of the others. When it comes time to run one of the constituent programs, I simply map the GUI data structure to that of the relevant program, copy the data over, and call the old program's main routine as a subroutine. Instead of printing the results, I hand them back to the GUI for plotting. The whole shebang operates in Fortran 77 mode (none of the newfangled stuff) with dozens of COMMON blocks. Works like a dream.
In order to be in trouble over space now, you would have to be looking at solving problems many times larger than anything you ever did with MS Fortran.
Jimmie the one - I'm doing exactly what you are asking about. The only downside, as far as I can see, is that you lock yourself into FTN95/Clearwin.
Eddie