Alan:
Who did you consult before making your measurements?
If you repeat a ride over the same mile course your rim reading should be within 0.1 rev of that in the first ride wether you have a Protege on board or not . Clearly this is not the case with your results, and you therefore have at least one problem unrelated to operation of the Protege.
I suspect that you may have labelled the rim backwards, so that numbers go down as the wheel rotates forward. For instance you thought your original miles were marked at 772.65, but they may actually have been at 772.35. If your second ride had been tighter by 0.4 this would have given you an apparant reading of 771.00 and an apparant difference of 1.65 as in the case of mile 10.
It's possible that you may have sometimes lost a count by letting the meter go to sleep and moved off without awakening it.
I am not clear as to why you do not estimate actual rim readings, but just go to the nearest division. Also, to quote down to one thousandth of a revolution is incorrect.
I am puzzled as to why you went with the two-magnet rather than the simpler one-magnet method. I suggest you try the latter.