Hello IO540,
To be properly calibrated, the airspeed indicator would have to be removed from the aircraft. The means of calibrating the airspeed indicator would probably involve direct connection to a meriam fluid (an oil with a density more precise than, but otherwise the same as water) manometer. This means is able to provide calibration accuracy well within one knot in the speed range you mention.
Due to the design of the mechanism inside the instrument, it is possible for there to be either one or two points along scale which are right on, and all others a little out. Depending how far out (knot or two), a correction card might have been provided. More error than that should be adjusted out.
If the process was accomplished with the instrument in the panel, there would have been several extra opportunities to introduce errors (though I would have expected them to be the other way).
It is also necessary to vibrate the instrument continuously during this process. The normal forces within the mechanism are so low that it is easy to have a many knot error due to internal friction (even in a perfect condition instrument). The most effective way to vibrate the instrument is to gently hold the blade of a screwdriver in your fingers, rolling it back and forth, while letting the plastic handle bounce off the top of the instrument case.
I spent 5 hours of test flying a deHavilland DHC-2 Beaver, following the relocation of the engine, to verify the static error had not changed (which would affect the accuracy of the airspeed). I tried the GPS method, and out of frustration gave up, and did 10 zero wind timed miles instead - that worked better. There was no change anyway.
I'm sure the GPS method works with precise flying and patience (as it was recommended to me by Transport Canada Flight Test department), but with the client watching the mounting bill for flying hours, the simpler method answered the question faster!
Cheers, Pilot DAR