This plot shows deltaV_z = V_z,w - V_z,e as a function of V_z. A correct T0 would show up as a band at deltaV_z = 0. A non-zero deltaV_z indicates some adjustment must be made to the T0. In this plot we see *two* bands, which means that somehow two different T0's are being used. This has been tracked down to different DB servers failing to sync with each other.

This second plot is a histogram of deltaV_z. It's from a later pass through the data where the entire distribution has been shifted closer to zero. The right panel shows two peaks. The peak at zW-zE ~= -0.6 cm is the correct value. Here, both V_z,e and V_z,w were reconstructed with the correct DB T0 values. The peak at ~ +0.15 cm is bad; V_z,e and V_z,w were reconstructed using two different T0's. It is broader because the incorrect tpcPadrowT0 did not apply properly for all sectors, so some are shifted and some aren't. The peak at ~ +0.08 cm (where both east and west jobs got a bad DB server) is almost impossible to see on this plot because it's so broad. The left panel shows only data which were analyzed with the correct DB values.