Following the presentation from Mathias at last week's TPC meeting (see slide 10 here), I made some very simple plots from a pair of random runs from each of the 2019-2021 FXT datasets (32 runs from 16 datasets) of last point y vs. x of primary tracks when that point is close to the central membrane west side, z = [-1,+7] cm, using just 20k-30k events from MuDsts:

MuDst.Draw("PrimaryTracks.mLastPoint.mX2:PrimaryTracks.mLastPoint.mX1>>hh(100,-200,200,100,-200,200)",
  "abs(PrimaryTracks.mLastPoint.mX3-3)<4","zcol");

Here are the results, where I have also reduced the histogram maximum by half to suppress the domination of tracks that end at the outer perimeter of the TPC:

dataset	random run A	random run B
production_3p85GeV_fixedTarget_2019
production_4p59GeV_fixedTarget_2019
production_7.3GeV_fixedTarget_2019
production_31GeV_fixedTarget_2019
production_5p75GeV_fixedTarget_2020
production_7p3GeV_fixedTarget_2020
production_9p8GeV_fixedTarget_2020
production_13p5GeV_fixedTarget_2020
production_19p5GeV_fixedTarget_2020
production_26p5GeV_fixedTarget_2020
production_31p2GeV_fixedTarget_2020
production_3p85GeV_fixedTarget_2021
production_26p5GeV_fixedTarget_2021
production_44p5GeV_fixedTarget_2021
production_70GeV_fixedTarget_2021
production_100GeV_fixedTarget_2021

Clearly what Mathias has shown is indeed a repeatable pattern throughout all of these FXT datasets, hinting at something static in the TPC such as alignment.

Note that the 26p5 GeV dataset from 2020 suffered from outages of nearly all of the outer RDOs on sectors 13 and 14 (see plots below), causing a very large number of tracks to terminate in the inner sectors at 10 o'clock and 11 o'clock in phi, but the general behavior that is seen in other sectors can still be seen lurking underneath that artifact.

-Gene