I havnt got any pics, but I could do a rough sketch.



From the frontal view you can see the blending into the wing, and from the side view you can see the blending into the cross sectional shape towards the front, the back part is big enough that it acts as a half end plate while the front is small enough to not add much weight but it still captures the vortex where it starts to blend towards the back, the tips where quite pointed to reduce weight and drag.

The idea was the vortex would sit on the edge of the winglet throughout appx the first 10 degrees AoA, thats where that blending towards the front comes in.

Because it could be bolted onto a wing that is symetrical aswell it would have equal properties at equal +- AoA values.

Unfortunatly I dont have access to the data from the tests, but from what I can remember (few years back) it was a semi symetrical wing and the peak gain in efficicency vs baseline was about 8% + AoA, and 6% - AoA, although with a symetrical wing that should be equal, probably about 7%, which is about the going rate on winglets.