This is for my own education mostly. After fiddling around with Izotope and the brickwall filters I got to thinking.
 
Why not brickwall the low frequencies and the hi ones- say over 18000 that no one hears or cares about on standard playback equipment. Won't we then be able to push the signal higher without distortion for a louder file?
 
Also, why does a hi pass at say 40hz cause a file touching zero to clip? 

backwoods
This is for my own education mostly. After fiddling around with Izotope and the brickwall filters I got to thinking.
 
Why not brickwall the low frequencies and the hi ones- say over 18000 that no one hears or cares about on standard playback equipment. Won't we then be able to push the signal higher without distortion for a louder file?
 
Also, why does a hi pass at say 40hz cause a file touching zero to clip? 

Anything under your target low frequency (we can use 40 hz for this example) should be removed....anything up high that is doing nothing, can also be removed. I do it all the time. But some things may show a little activity in those ranges too, so you have to be careful. Some of this will be genre specific too.
 
For example, if we remove everything below 40 Hz in a r&b song or rap song, we just affected any bass drops they may have had in the song.
 
If we remove 18 k and above (which is usually safe to do) there are a few plugins out today that accentuate like 22 k. I see them as senseless but some guys swear by them and I have a few clients like that. They literally feel that high end air is making a difference for the better where to me, it's adding hiss that can mess with the audio.
 
So depending on what style of music you are working on will determine where you high pass and low pass. I think you'd be better off high passing and low passing over brickwalling the stuff. This way you have a little more control over what you allow to pass through naturally where as brickwalling it could introduce artifacts.
 
A file touching 0 dB is clipping in the digital realm unless you are literally getting readings of -0 dB. But at regular 0 dB without the minus, you'll show clip points no matter what you add to the audio unless you are cutting something drastically. But just about anything you bring in will clip a file like that. It may even clip just passing through the plug without touching a parameter on the plug.
 
Your best bet is to keep anything un-mastered at -3dB peak...anything you master, -0.3 and no hotter than -0.1 dB peak. But after something has been mastered, there is a good chance anything you add will make the song clip...even if you're high passing.
 
Though -0 dB doesn't show up as a clip, you shouldn't really go that high if you can help it as you're just way too close to the clipping zone and chances are you just may go over.
 
-Danny

It is regular practice to cut low frequencies below the lowest note needed.  For flute for example you don't need anything below 190Hz so that can safely be removed. 
 
As for the high frequencies, cutting them will tick off all the audiophiles who convinced themselves they can hear the higher frequencies, or that leaving the high frequencies in will somehow make the lower partials sound better, (which defies all physics).  But is generally is safe to leave them in unless they are filled with noise you want to remove.  CD's and mp3 limit themselves to a max of ~20kHz anyway. 
 
 
 
 

"Also, why does a hi pass at say 40hz cause a file touching zero to clip?"
 
Ringing?
 
What filter, and filter order or slope are you using?
 
best regards,
mike

Also, why does a hi pass at say 40hz cause a file touching zero to clip?  Much more important question, why is the signal touching zero in the first place?
 
Interesting article in SOS September issue on gain staging. Although it fails to mention a really good approach to system calibration one of the main things that comes out of it is there is no reason to be anywhere near 0dB FS at any point in your production. Except for mastering and that can be a time when a signal comes close to 0dB FS but until that point it should not and does not need to.
 
People just need to keep well away from it and keep everything well under, turn your monitor level up.
 
On another issue one thing I find helps with keeping your mix clean is to think about the highest frequencies that various parts are actually going up to and putting a LPF with the cutoff set at some point up just past the highest necessary frequency that is can really help to keep a mix clean. It may sound like it is doing nothing on the track but it all adds up and helps keep unnecessary high freq clutter to a minimum. It works down the other end too as Daniel points out. If something only goes down to 200Hz then a HP set somewhere below that will help to minimise low freq clutter. The trick is to ensure these cutoff frequencies don't really impact on the sound. In the case of the LPF sometimes the cutoff freq can even be set on purpose to roll off some HF and make a track sound smoother and better.

I was using 0 as an arbitrary figure. Putting any hi pass on a file raises the top level it seems. Just easier to say hi pass 0 makes it clip rather than hi pass-3 makes it go higher than -3. I'm trying to understand these things. Just looking at an EQ curve it would suggest the volume will go down with a hi pass- not so.
 
Do you guys instead of brickwalling sometimes use several hi pass filters on the same file?
 
here we go: http://www.soundonsound.com/sos/dec05/articles/qa1205_3.htm
                   http://www.gearslutz.com/board/4376759-post13.html

I think it is a good point that you raised about how levels can go upward when using a HPF. My point is that if your gain staging is correct and the average and peak levels are well below 0 dB FS than it does not matter. As long as you make your limiter the very last stage in the mastering process then an increase in level is just never going to happen.
 
With HPF (and LPF of course) the slope of the filter is important rather than using more than one in series. Sometimes it is better to move the cutoff freq much higher but use a 6 db/oct slope and you will get the sound you are after but on other ocassions it is better to use a 48 dB/oct slope and have the cutoff as low as it can go so you are letting as much low energy through as you can but then slamming off everything under a certain frequency.

If your levels are at 0, as Jeff said they should not be before you add eq, there is a possibility of the program thinking the output is increasing even when you cut instead of boost, especially in the bass range due to phase shift and general calculation errors.  You may not be clipping, but the meters think you are.  Give yourself plenty of headroom to make adjustments and increase the levels at the final stage. 
 
In a DAW it is perfectly fine to decrease the level of the audio and increase it later.  The noise floor is virtually non-existent and there will be no loss of resolution. 

backwoods
Putting any hi pass on a file raises the top level it seems.

This is an example illustrating "ringing" in a EQ filter. The gui doesn't identify the filter design or the filter order.
 

 
The "Q" is reported as 8.14 which is very steep. (3dB/octave?) If that "Q" was below 1.4 it would be far less steep and probably would not have any ringing.
 
 
I don't have a great understanding of EQ math so I will not pretend to be able to answer your question. If you want to understand why hi-pass, low pass, or shelving EQ filters "ring" or resonate when they are set to high "Q" factors with steep slopes you'll have to learn about EQ filter math and the various types of EQ filter designs.
 
I generally try to avoid "ringing" by using hi-pass, low pass, or shelf filters with low "Q" factors and gentle slopes, but sometimes I exploit the ringing by choosing appropriate settings.
 
Good luck learning about the math. There are some folks here who seem to know it very well. I usually gloss over every time I repeat the process of endeavoring to learn it.
 
When you really need a steep slope with out ringing, the stacking of instances can be an effective solution.
 
 best regards,
mike
 
 

Don't worry about the math. Just remember that steep slopes should generally be avoided -unless you actually want the effect Mike showed above. And sometimes, you actually do. It can be a cool effect. Try it on kick drums. But for mastering, probably not. Which is why we use linear-phase filters for that application. 
 
Cutting very low frequencies is common practice in mastering, although 40Hz is a tad high for most genres. 30 Hz or so is a more reasonable target. The thing is, even though you can't hear those frequencies (and your listener's playback system can't reproduce them), your dynamics processors can hear them just fine. Excessive energy down there can affect bus compressors and limiters in unexpected ways.
 
As for limiting the extreme high end, the only reason I can think of for doing that might be to reduce the likelihood of aliasing when played back as a wave file on a really cheap player. I wouldn't bother, though - if they're that cheap, the heck with 'em. OTOH nobody will miss those frequencies, and they're gonna get tossed anyway if you encode to MP3. But generally they're not going to cause any harm, and as Daniel noted above, some audiophile may hook up a spectrum analyzer and then send you an angry letter.