In the ongoing effort to figure out what's wrong with my calculated 2005 asymmetries, the like-sign (A_ls) and unlike-sign (A_us) asymmetries, as defined here are calculated and shown below.

By run: A_ls, A_us

It looks to us that something weird is going on with A_us in the first half of the fills. This seems consistent with the problem we're seeing with the single spin asymmetries, A_B, A_Y.

Thinking the issue may be related to the beam spin pattern, we separate the fills and runs by their spin pattern. Unfortunately, it seems like the majority of these runs/fills were of one spin pattern, P1 (B+-+- Y++--). The list of fills and their corresponding spin pattern can be found here.

Number of fills, runs per spin pattern.

The problem is visible in A_us, and seems to be more prominent for P1, but is it really a problem with that particular spin pattern, or just a function of the time (earlier in the run)?