I found that in my previous updates I used about 50 or so more runs than did Kevin in his measurement. This should account for much of the uncertainty difference we see. I have, now, removed those runs from my list for a more direct comparison. The results, here, still include the geometric trigger which will grow my statistical uncertainties relative to Kevin's. Additionally, I now examine the effect of binning in pseudorapidity. I use the same bins as for the Sivers measurement: |η| < 0.5 and 0.5 < |η| < 1. Now, the asymmetries are calculated fill-by-fill and the weighted averages are performed over fill, trigger, beam, and eta. This will add an additional increase to the over-all uncertainty, since I will lose a few events from splicing the data and requiring all 12 azimuthal bins be filled for each detector half and each spin state.

p_T > 9.9 GeV/c

The higher p_T jets are more sensitive to quark effects. Thus, one expects the largest Collins signal to come from the high-p_T jets. I have chose 9.9 to correlate with my definition of jet-p_T bin edges. This is also a slight departure from Kevin's analysis where he places the cut-off at the more nicely round number of 10 GeV/c.

Figure 1: Total Range of Pseudorapidity

Figure 2: |η| < 0.5

Figure 3: 0.5 < |η| < 1

Table 1: π⁺

Table 2: π^-

The numbers for the full η range are very similar to those I had before. The linear fits are quite reasonable. The magnitudes are consistent for π⁺ and π^- with opposite charge-sign. The slopes are of 7.8σ and 7.1σ significance, respectively. For the η bins, the asymmetries are also consistent in magnitude with opposite sign between the two charge-states. For central pseudorapidity I find an asymmetry of 6.8σ and 6.9σ, respectively. There is a hint of increase in the z-dependence as one moves forward in η, but the statistical significance is quite small (< 1σ for π⁺ and 2.43σ for π^-). Similarly, when one integrates over all bins one finds hints--but nothing more--of an increase in the asymmetry as one moves forward in η (1.25σ for π⁺ and 1.18σ for π^-).

p_T < 9.9 GeV/c

While low-p_T jets are more sensitive to gluonic subprocesses we benefit from a tremendous amount of statistics. Additionally, they may be quite useful in comparison to the Run-11 results. Kevin has already posted low-p_T Collins asymmetries; and I, now, post my independent analysis.

Figure 4: |η| < 1

Figure 5: |η| < 0.5

Figure 6: 0.5 < |η| < 1

Table 3: π⁺

Table 4: π^-

Thus, for the full η range the charge-sign-dependence is still quite significant (3.5σ), however, the size of the effects and significance of the z-dependence is quite reduced relative to high-p_T jets. The integrated asymmetry returns a charge-sign-dependence of 2.76σ. The quality of the constant fit is also much better than in the high-p_T case. There are hints of an η-dependence, but nothing significant.

Summary

Figure 7

pT > 9.9 GeV/c	pT < 9.9 GeV/c

Anselm asked for a plot of the slopes as a function of η. In Fig. 7 I show the slope dependence on η. There does appear to be a fairly linear dependence on η which seems to scale with p_T. I'm bothered a bit by the high-η behavior, but perhaps it is not statistically significant. A double-check by Kevin will be quite useful, here.