Again, if you were to provide a better illustration of the context, say a minimal working example of what exactly is being pursued and why, chances are the starting premise will become moot, “one required operation is to pack 32-bit integer values into an array of 64-bit integer containers.” and better ways to approach and design the larger aspects with modern Fortran will become evident. And the rest of the discourse here will be witter.
There is a lot of textual description in the original post and the one that immediately followed but details are missing. It comes across as nonstandard steps were taken in the 70s and 80s toward instructions the language didn’t support: the language still doesn’t support the operation of interest directly but how to now replace the nonstandard code with modern Fortran syntax!? That makes little sense in the larger scheme of things now that the readers in the discourse have had their fill of being humored.