1) Sti and Stv appear to use the same energy loss functions, gdrelx and gfluct, obtained from G3.  There is a difference in the way Sti and Stv call gfluct (which is renamed gsigma2 in both packages)...

In Stv:
fdEdXErr2 = gsigma2(fZ/fA,fDens,fCharge2,fM ,beta2,1.);

In Sti:
  double err2=gsigma2(_Z/_A,1.,z2,m ,beta2,1.);

Sti is obtaining a significantly larger error in gasses (density < 1 ) and a significantly smaller error in solid materials (density > 1), because the energy loss squared (?) is directrly proportional to density.

Sti may correct for this later... but I need to trace the spaghetti code which is Sti.  Or Sti may neglect energy loss fluctuations entirely.