The export command is a (ba)sh built-in. If that isn’t working, it probably means you’ve set something other than Bash for your shell. Try echo $SHELL to see what shell you are using.
Also, I’d try running the installer under the more common Bash shell just to be sure that isn’t the source of your problems. I think running exec bash --login ought to be sufficient to switch to the Bash shell temporarily for a one-time test of running the installer under that shell.
Good luck. 
I’ll give it a try on Wednesday when I’m back in the Institute but, it might sound silly to you, I don’t seem to have the command “export” on the F41 live-usb installation.
Thank you for your help and I’ll let you know the outcome.
Best regards.
Thanks again.
I’m learning something new everyday (as they used to tell us to do when at primary school).
I think that’s just about my level from the IT point of view.
News tomorrow.
I have an old iMac with Nvidia graphics that would need NVIDIA-Linux-x86_64-390.157.run. Building the Nvidia .run modules requires disabling nouveau, so needs an ssh terminal session from another system. I get:
make[2]: Entering directory '/tmp/selfgz1756/NVIDIA-Linux-x86_64-390.157/kernel'
warning: the compiler differs from the one used to build the kernel
The kernel was built by: gcc (GCC) 14.2.1 20250110 (Red Hat 14.2.1-7)
You are using: cc (GCC) 14.2.1 20250110 (Red Hat 14.2.1-7)
These are the same compiler (/usr/bin/cc is just a symbolic link to /usr/bin/gcc), so should not be a real problem. Here I see many serious errors in /var/log/nvidia-installer.log for functions called with mismatched arguments or the wrong number of arguments – all the sort of things that change with major kernel versions. Old systems can’t be expected to keep working, so unless you are looking for a learning exercise, the benefits from working on Nvidia’s old driver may not be worth the required investment of time.