I decided to measure the propeller RF oscillator accuracy tonight. I find that my random sample of one Propeller Board to be accurate to 1 ppm over the range of 1 Mhz to 100 Mhz. In my case, the error is always negative requiring a positive offset to correct back to the proper frequency.
From this, I should be able to calculate an error offset for the oscillator over the entire frequency range once the error correction at any given frequency is known for the particular propeller board.
Drift on this particular board at 10 Mhz appears to be around 2 Hz over 2 minutes time, which represents a typical WSPR transmission. My next set of experiments with drift control will be to temperature stabilize the propeller chip at about 50 degrees C and see how this affects drift.