cosmetic

Within an OLE context, when Multiple over-lines are used, 
The end-point of the left edge increases leftwards.
The end-point of the right edges increases and decreases repeatedly.

The expected behavior of the right edge
   a) Increase uniformly (less desirable)
or b) remain constant 
The expected behavior of the right edge is to 
   a) Increase uniformly like the left edge (less desirable)
      This seems to be jagged at 400% zoom due to pixel rounding at 100 %zoom.
or b) Remain constant. 

The moment you click on the object and enter into formula editing
  the jaggedness becomes less noticeable, however still there.

In Standalone Open-office-Math or in formula-editting , this non-uniformity 
  is still present but very slightly, I think this gets 
  magnified in OLE context.
By close observation it seems like the right-endpoint is calculated by adding
  a fixed length to the after-pixel-rounding location of the left-end-point. 
  I speculate that that the right thing to do is to calculate the right-endpoint
  by adding a fixed length to the location before-pixel-rounding 
  left-end-point and then pixel-rounding that.
  But then in the OLE-context mode the right edge seems able to take any 
  non-rounded values (in comparison to the first overline). 
  Which means maybe there is no necessity to pixel-round the location 
  of the left-edge
  I might be wrong.

also effects widevec, widehat, widetilde, underline, overstroke

I think it is better to not have the line to increase
  on either side. This permits two overlines to be considered
  like a double overline with technically identically lines
  of equal length. which means keep the left-end-point
  the same as the previous left-end-point.

underbrace and overbrace don't seem to have this issue.

on XP, openoffice release version 2.1

just aesthetics, low priority.