[Clascomment] OPT-IN:Single and double spin asymmetries for deeply virtual Compton scattering measured with CLAS and a longitudinally polarized proton target

[Volker Burkert]

Dear Silvia Pisano and all,

This paper is the result of a very big effort, and it is great that we are converging 
on this full paper. Here I have some, hopefully constructive, comments on some parts 
of the paper that could be improved.  
Volker


General: 

1)	You have many figures that include color. Unless you want to pay a lot of money to 
have them included with color in the printed version, you should include Figure XX (color online) 
for each Figure that includes color. 

2)	Throughout the paper different use of units is made. For example, Q2, t, W is often used 
with c=1, while in other places Q2, t have units (GeV/c)2 . Make it consistent throughout. 

3)	If the current authors list should become the final “lead authorship” group, this would 
establish a 4-tier ranking. If I include the ranking within the Orsay authors, it is even a 5-tier 
ranking.  Unfortunately, the previous PRL has already started this, which broke with the 
tradition from ALL previous DVCS paper to place the primary analyzer(s) first, and the 2nd 
tier authors are the core discussion group listed in alphabetic order. If anyone believes one 
is getting more credit or recognition for being in place “#3” compared to place “#5” they 
are mistaken. 

Detailed comments: 

Eqn.(1): Instead of defining only Q2, it would be better to define all kinematics quantities 
(Q2, ν, W, xB , ξ, t, p, p’) in one block of equations before describing the hard scattering 
requirements. It becomes difficult to read if most of the definitions are squeezed into 
the text. It also avoids using quantities such as “t” before they are defined (eqn.(2)), also 
Fig.1 is referred to before quantities used in the figure are defined ξ, t. Also, in the notation 
in Eqn.(1) Q2 will be <0, but in the figures later it is quoted as Q2>0. I suggest to use the 
usual definition q2 = (e-e’)2 and Q2 = - q2. 	

Line 132-135: 
There is no reason to put down the Shifeng CLAS measurement as being of “limited statistics” 
and “non-dedicated” while the Hermes data with even less statistics, and came 4 years (!) 
later, are just referenced without qualification (as it should be). I consider our 2006 data 
as pioneering results. 

Line 262:  the phrase “less than a meter’ should be replaced by the exact quantity as is done 
for the other distances, e.g. is line 273 for the distance to the CLAS center.

Line 300:  “DVCS events” should be replaced with “epγ events” or “single photon events”.

Line 345: “..low-number-of-photoelectrons peak” should be replace with “single-photoelectron 
peak”. [Note that the sharp peak in Fig. 5 is exactly at the one-photoelectron position.] 

 Line 252-253: The momentum is actually measured by the drift chambers from their curved 
trajectories in the Torus magnetic field.

Line 265-266: I don’t think these are neutrons. [Note the 2ns structure is visible in Fig. 8 as 
sharp lines that are independent of energy. This indicates that these are out-of-time photons. 
If they were neutrons, one would not see the 2nsec beam structure.] I suggest replace 
“to separate the photons from neutrons” with “to select the in-time photons”.    

Line 410-413: There we argue that π0 were not included in the simulation to explain the 
discrepancy with the MC simulation for EC. However the next section starts with the headline 
“Exclusive π0 simulations”. A referee may ask why were they not included in the simulations 
with the DVCS/BH events. 

Line 436: Why did the cut allow events with more than one photon if we include only single 
photons in the asymmetries? That seems to increase π0 contamination that then will have 
to be subtracted later.

Line 479:  “W” should be defined in Eqn(1). 

Line 487: “smearing” sound like it includes human (GPP) intervention, I suggest to use 
“broadening” instead.

Line 498: “..these data appear to be strongly dominated…” replace with “..these data are 
strongly dominated..” 

Line 500: replace “appear to be” with “are”. 

Line 505: replace “heavier” with “larger”. 

Line 583-585: “xB shows the strongest deviations from the constant behavior … but a
 constant fit is doable” Replace with something like this:  “There may be a small xB 
dependence, although the dilution factor is not inconsistent with a constant behavior”.  
[To better quantify this, a straight line should be fit to the data. The slope may not deviate 
by more than 1.5sigma from a zero slope.] 

Figure 12: The MM2 distribution clearly peaks closer to the (pion mass)2 (~0.018) 
than the photon mass (0). This should be mentioned and discussed as it already 
shows that π0’s dominate the “epγ” sample in EC. This is still true after all the exclusivity 
cuts. [In line 436 it is discussed that event were selected with “at least one photon”. 
The question I have is, do the exclusivity cuts include a single photon only constraint 
or not, and if not, why not? ]

Figure: 13: The thin black lines showing the binning should be made thicker, in the top graph
 they are hardly visible. 

Line 713 and Fig.17. The carbon (red) data show a shift towards more positive values of ALU, 
which may indicate a bias in the carbon data. Two data sets to be comparable at the 3sigma level 
does not indicate good agreement.  The more relevant comparison is to compare the leading 
twist contributions, i.e. sinφ moments. I suggest to fit the two data sets with an offset fit parameter, 
i.e. AUL=  a + bsinφ, and compare the b values for both data sets. My guess is that they would 
agree better than the 3sigma quoted for the comparison of the unfitted data sets.  

Line 863-866 and Table IV: The table compares errors for the asymmetries BSA, TSA, DSA.  
The quotation of relative errors is not a useful way of showing asymmetry uncertainties 
as the asymmetry is already a relative quantity. The absolute asymmetry errors should be 
given, e.g. AUL ± ΔAUL . The consequence of showing relative percentage is demonstrated 
in lines 863-866, where one has to explain why the quoted (relative) errors are so different 
for the different asymmetries while the absolute errors are basically the same. 

Line 949, 996, 1049: These definitions should be given in equation mode with an identifying 
number, not squeezed in between the text. Also, these definitions are repeated in the captions 
of Fig.18, Fig. 20, Fig. 23. If they are properly defined with numbers, they should be referenced 
as such. 

Figures 18, 20, 23: The graphs include ultra thin lines showing the fits to the data. The lines 
are nearly invisible. 

Line 1147-1152: The qualitative conclusion about “scaling” is insufficient. If we want to conclude 
something about scaling, we should make it quantitative. The data should be analyzed with a 
straight line a + bQ2 at fixed xB. The uncertainty in b would be the relevant quantity to conclude 
about “scaling” behavior. Since we have only two points in Q2 at fixed values of xB a fit is not 
required, just simple algebra.  I suggest to do this exercise.

Conclusion section: One should include something about improved analysis once the new 
cross section and beam asymmetry  data are available. Also, the 12GeV DVCS program should 
be referred to. 

Happy holidays! 
Volker