Plotting the mfilt versus b values from the UCD scaler analysis for all energies gives the upper figure which shows a clear deviation for the extrapolation 2 AGeV data. As a guidance I ran a few thousand UrQMD events at 4 AGeV (b=0-15fm) and plotted the charged particle multiplicity (minus the initial 178 protons) versus the impact parameter. A comparison seem to indicate that the 2 AGeV data have a (small?) problem.

After normalizing the mfilt distribution to the mfilt value at half the plateau and normalizing to the height of the plateau (due to different trigger bias one can't just normalize to the counts) I put in the impact parameter cuts according to UCD as lines of the same color. Here are the kumac which produces this plot, the kumac containing the comis utilities (you will have to edit them a bit) and the rz file containing the histograms. Here are the plots for the
2 fm cut
3 fm cut
4 fm cut
5 fm cut
6 fm cut
7 fm cut
8 fm cut
9 fm cut
They show that the 1.85 AGeV data are consistently different from the other beam energies. If it is true that in this representation the fm cuts should be close to each other I can estimate the cuts for the 2 AGeV data by taking the mean of the other energies. This new value is shown in the following plots as black line
2 fm cut
3 fm cut
4 fm cut
5 fm cut
6 fm cut
7 fm cut
8 fm cut
9 fm cut
The mfilt values extracted this way are
UCD BNL
2 fm (1.8 %) 190 181
3 fm (4.2 %) 175 163
4 fm (7.4 %) 160 146
5 fm (12 %) 140 125
6 fm (17 %) 125 107
7 fm (23 %) 105 83
8 fm (30 %) 85 60
9 fm (38 %) 60 38
One possible explanation may be a difference in the mfilt distributions for
2 AGeV
4 AGeV
6 AGeV
8 AGeV
The UCD 2 AGeV distribution could be one made with runs from the more central trigger setting.
If my fm cuts are used the 2 AGeV data agree extremely well with the common crossing of the other beam energies.