This paper describes the ASSIST instrument, a Fourier transform spectrometer operating in the infrared and designed for ground-based measurements of downwelling radiances.  The paper is clearly written, and the instrument as well as the data processing are well described. I have a few questions/comment, though, which are listed below.

Line 83: 

You describe the beamsplitter as "nominally flat". Don't you have problems with multiple reflections at parallel surfaces?

Fig. 3 Figure caption:

"For simplicity, rays are traced at 550 and 1650 cm-1 for only one arm of the interferometer and from two field points located at infinity: zero and 21 mrad relative to the optical axis (blue and red, respectively)"

With this description I'd expect four beams: 550 cm-1 with zero and 21 mrad and 1650 cm-1 with zero and 21 mrad.  How do I know which rays belong to 550 cm-1 and which rays belong to 1650 cm-1?

Line 206:

"a NIST-traceable in-house calibration procedure"

Can you spend a few words on this calibration procedure? What is the reference?

Line 215: "Section 33.3" must be "Section3.3"

Line 221:

"a modular air conditioning unit that can maintain a steady-state internal temperature for external

temperatures up to at least 40◦C."

which steady-state internal temperature is used?

Line 304, Section 3.2:

Do you have any idea if the non-linearity may change over the lifetime of the instrument due to detector aging? 

Could the modulation efficiency change over time? Are laboratory characterization measurements for the non-linearity coefficients

foreseen from time to time in order to monitor and to account for such possible changes?

Line 370:

"Figure 13 shows the typical responsivity (magnitude of the complex gain) 

in the 500 to 3000 cm-1 range for the forward scan direction."

Just out of curiosity: I understand that the phase of the gain is sweep direction dependent, 

but is the magnitude the same for both sweep directions?

Figure 13:

How do you treat the absorption lines in the response curves? 

I assume that these absoprtion lines lead to erroneous radiances in the calibrated data?

The same question holds for the atmospheric lines in the instrument offset (Figure 14).

Figure 16 (a):

What are the red dots in the curve?

Line 428:

"is proportional the cosine"

should read: "is proportional to the cosine"

Fig. 21:

The red line turns into yellow around 1750 cm-1 and becomes white between 1825 and 1900 cm-1 on my screen. This is a bit irritating.