The results of the winglet versus wingtip tests are in.  The tests were conducted with an airplane that has winglets installed and we made the IAS versus altitude tests with the same manifold pressure-rpm and exactly the same TIT.  And then we landed and replaced the winglets with standard wingtips and did the same tests.

  In a nutshell, we all knew the winglets have more drag than the wingtips.  If you were still laboring under the idea that they somehow added to the speed at any altitude, you are misguided.  A designer at Gulfstream once told me that if we were cruising at .55 Mach or greater, they could be an asset with detailed engineering and wind tunnel tests.  Otherwise they are a pretty, but unnecessary, drag producing device. The genesis of the winglets is around the venting (overboard spilling) of fuel because of a bad fuel selector valve while parked on a sloping ramp. The winglets objective was to get the vent higher than the wing fuel level.  They succeeded in doing that. The fuel valve has since been fixed and we hardly ever hear of a overboard venting of fuel --except in fast turns onto the runway with full fuel tanks. And they are incorporating check valves in the vents to prevent that.

  Here are the results of my test. All numbers indicate a slower indicated airspeed for the winglet airplane at constant and equal power settings, at the various altitudes and repeated using the standard wingtips.
Sea level =12 kn        
5000 feet =11 kn
10,000 feet =9 kn
15,000 feet =7 1/2 kn
20,000 feet =5 kn
22,000 feet =4 kn.

The surprise in these tests (to me) was that the highly touted benefit that the winglets cause a reduction in deck angle at altitude, and therefore reduced drag, was false.  I found that at the lower altitudes the deck angle was higher (averaging 1°) with the winglets.  At 22,000 feet they were roughly the same. Therefore the added area just helped to offset the reduced angle-of-attack.  After thinking about it I attributed it to the washout of the wing and that on the wingtip the angle-of-attack is a low-er therefore an increased angle-of-attack is needed to provide the lift.  Am I thinking correctly?

Any and all comments to add light on the subject are appreciated.
Charlie K.