Spectrum of generic $\mathrm{CO_2}$ laser It is often quoted, that it's emission is 10.6um. But we also know that CO2 lasers have huge number of possible emission lines in 2 main bands: 10.6um and 9.6/9.4um.
But what is emission spectra of generic sealed tube CO2 laser (let's say ~1m long, ~40W CW) without any line selection devices?
Should I expect single line with highest gain, jumps between 2 neighbor lines, or simultaneous emission from all of them from ether single or both bands? 
I need this for optimization of CO2 laser collimation optics. 
 A: You should stimulate emission from the entire ro-vibrational spectrum.  There are a few modes in the P branch that will dominate, just because they have the largest oscillator strength.  They're close enough in frequency that your output should be a homogeneously broadened peak centered around 10.6 $\mu$m.  If you wanted to put an etalon or diffraction grating with 1 cm$^{-1}$ resolution at the output of the cavity, then you could pick out any of the transitions.  
A: It sounds like you have a particular laser, and you don't know what spectrum it's emitting? The best way to answer this question would be to measure your emission with a spectrometer, but if you're asking here, that implies you don't have one.
Let's pretend you got your laser from Coherent. You'll notice that different cavity designs can pick out different emission lines (9.2 $\mu$m, 10.2 $\mu$m, 10.6 $\mu$m, etc), so just knowing that it's a CO$_2$ laser isn't enough to tell you which line(s) are emitted. The spectrum emitted by any one of these models is roughly specified in its datasheet. For example, the GEM-100A 10.6 $\mu$m model is specified to emit from 10.55-10.63 $\mu$m, whereas the 9.6 $\mu$m model is specified to emit from 9.5 to 9.7 $\mu$m.
Where did you get your laser? Can you really not get a datasheet for it? Can you get a spectrometer? Can you build a spectrometer? I've built spectrometers for the visible with gratings and lenses, do you have gratings, lenses, mirrors, and sensors available for ~10 $\mu$m?
