Nada on the dithering....
ffmada, it just don't make any sense to send a dithered mix to a mastering house. Here is why.
Dithering puts "noise" at about -96db and that noise is generally around 3db loud. But it is shaped noise, so usually, you don't hear it. But in effect, you now only have 93db of "clean" audio.
The problem with dithering a mix down to 16 bit is that the mastering house is probably going to have to dither it too. Not a good thing to dither twice as you lose yet another 3db of s/n ratio.
The best way to submit your mixes to a mastering house is at the highest resolution possible. This may mean burning a DATA CD with the higher then 16 bit files. In the case of 24 bit recordings, as long as you keep it at 24 bits, you don't need to apply dithering at all after any DSP, because the quantinization errors will be far before where further dithering will take place, and even if you kept the audio at 24 bits, you would not be able to hear the errors because they are too quiet for the human ear to detect.
Even if you HAVE to mix at 16 bit, it would be better not to apply dithering IF you are going to send that mix to a mastering facility. Yes, you will truncate the file to 16 bit, but at least the whole 16 bit is there, with no added dithering noise yet. Let the mastering house deal with that.
Now, the above paragraph ALSO depends on how the mastering house is going to deal with the mastering. If they are going to run the 16 bit files through a D/A converter, then through analog gear, then through a A/D converter, then you MAY need to apply dithering at mix. But, this whole conversion process is probably going to start adding noise to the audio signal, so it will dither itself in this case.
If they are going to master with digital tools, dithering should be the last thing done. But remember that Digital Signal Processing is best done with higher bit rates and higher sampling rates, so once again, you should supply higher bit rate files to the mastering house.
Ed