[Clascomment] OPT-IN:Beam-spin Asymmetries from Semi-inclusive Pion Electroproduction

[Sebastian Kuhn]

The paper is mostly well-written and shows interesting results. I will give a marked-up hardcopy to one of the co-authors for minor corrections. Here are some more substantial comments:
1) lines 77-80 are somewhat redundant with the next paragraph and should be moved there (and merged to give a single, cohesive narrative
2) line 123 states that F_LU is purely twist 3. Aren't there possible contributions from higher twists, as well?
3) Fig. 7 shows no trace of kaons. Maybe the z-scale (color) could be made logarithmic? Also, the discussion doesn't mention any background subtractions or corrections for pi+ and pi- - are you sure there are NONE?
4) Eq. 7: there is no mention of count rate normalization. Presumably, N+ and N- are normalized with (life-time gated) Faraday Cup counts for each helicity?
5) Some statements are bit stronger than warranted. E.g., Fig.10 caption says the pi- and pi0 kinematic coverage is "very" similar to pi+, but it has to be smaller. In line 403-404, it is claimed that systematic uncertainty (which, btw, should never be given in "%") is smaller than statistical in ALL measured kinematics. This disagrees with the pi+ results in Table I and Fig. 17
6) The discussion in line 393ff is rather complicated and confusing, and perhaps unnecessary in a publication. Fig. 12 doesn't really need to be there. Fig. 13 shows that there is SOME excess at large Chi2, beyond that expected from the curve.
7) The systematic errors due to Pbeam shown in table 1 are excessively small - especially for pi+ where the asymmetries themselves are large (and the error is proportional to them). E.g., according to Fig. 14, A_LU for pi+ and one z-bin has an amplitude of 0.0323, which should yield a systematic error of at least 0.0006 (for 2% uncertainty on Pbeam).
8) Fig. 15 and 16 should be made more appealing. The symbols are hard to distinguish (use color and thinner lines!), and the labels and numbers on the axes are too small. The word "uncertainty bar" is definitely not standard usage - error bar is the more common expression.