Digital BW, The Print

I've just completed a curve in QTR for UT-FSN on Moab Entrada Bright White.  I printed out 
the step wedge and measured the densities on my densitometer, which, sadly, bit the dust 
immediately after finishing the readings.

In graphing my measured values alongside the values in the chart of ideal densities for 
dmax = 1.64 (I actually have a measured dmax of 1.65), I noticed that I have two slight 
bumps in an otherwise smooth curve.  The linearization function in QTR did a darn good 
job of matching the ideal density curve, with the exception of these two small bumps.

Does anyone have any ideas for further tweaking the curve produced in QTR to smooth out 
these bumps?  The bumps are at steps 4 & 5 and 8 & 9.  The 4 & 5 bump is darker than 
ideal by about 0.02, and the 8 & 9 bump is lighter than ideal by about 0.02.

Thanks,
Steve