Removing linear trends? Wa?

Solved off-site

Hey, was wondering if anyone here could give just a nudge in trying to understand what in the world is supposed to happen here.

The basic problem is a long time series of values, which is to be analyzed in various ways. The first part of the problem is simple enough:

Remove the recurring cyclic pattern by transforming to the frequency domain and zeroing appropriate values.

Fair enough, Matlab using fast Fourier transform and a bit of coding takes care of this. But then this problem statement comes up:

"Remove the linear trend from the data before filtering, and then replace it after filtering".

And I must be spectacularly dense or something, but I really don't get what you're supposed to do here. Remove the linear trend? I've done a linear regression, showing that the data conforms fairly well to a simple linear function - is it something to do with that? But remove? Replace?

So yea, if anyone could give me some kind of hint at how this is to be done I'd be extremely grateful! The text book I have is quite amazingly unhelpful. I don't mind being dumped links to some good source on the subject; in the end what I'm after is to actually learn and understand the material, but right now I could really use some help to stop banging my head against a wall.