Same thing done the other way....... There is a catch in it though... Before executing it, you need to have Windows partition mounted at /mnt/winc, right?? Now, while loading the bootloader you don't know of any filesystem tree. Then where will "winc" be mounted??? Only after booting into linux will you be able to access /mnt - unless of course it is onto a different partition. But again, it would be better if you use the device name + partition no. combination instead of *nix filesystem specifics.

I did it mainly because I used to frequently update kernels... it doesn't make too much sense to write the bootsector somewhere else, dd it to a file, copy it to the windoze partition etc. every time.

If you do it only once, during installation, then of course it amounts to the same thing, and my method is probably more more roundabout.