Original hint something strange was going on in the new vs. cvs file was very different behavior with and without full STAR geometry -   Ilya's post.  This was traced to differences in ELED block in the new file - which instead of being pure lead was instead a lead alloy of lead, tin and calcium.  Further tests showed the ELED block with the alloy was very sensitive to the LOW_EM option while the cvs version was not - Ilya's post.  We did some "experimental coding" and found that this was true even for a mixture made of 100% lead.  Plot below shows total reconstructed energy/thrown energy with a 100% lead "lead alloy."  The difference between the left and right plots are a line "Medium Ecal_Lead" under the Material defintion in the block.  It does not change the LOW_EM behavior.

A change was suggested by Jason that should have made the lead alloy the tracking medium.  Tests showed this created even stranger behavior, with the reconstructed/thrown energy now like the plot below.

We concluded that some very strange things we do not understand seem to happen when lead is in a mixture.  For now we will retain pure lead.

Further, it appears to be only a problem for high Z mixtures because the stainless steel mixture which replaced Iron did not cause these kinds of problems.