Gary, sorry, it just doesn't work that way. The fact that two GPS units displayed similar errors doesn't make them less in error. If you stand a 1/4 mile due West of a large iron ore deposit, you could have a hundred compasses, and they will all be in error. The factors that make any GPS unit report errors will make just about all such units in those circumstances report similar errors. For example, it's very common for a stationary GPS to report that the vehicle or vessel in which the GPS is located is either moving or turning, when if fact it's doing neither. Yes, there are well known exceptions to displacement hull speed calcs, but your OI 33 certainly does NOT fall in any of the excepted categories (e.g. multihull or ULDB). I understand you took photographs, but they can only prove the possible degree of error of your instruments given the circumstances. Try turning on your GPS while at the dock and watch it for 10 minutes. More than likely it will show your boat changing heading, picking up speed, etc.
BTW, 5 satellites is barely above the minimum number for accurately determining location (4). There are many authorative documents available online to explain why a GPS errors on speed. To pare it down the gist is:
Quote:
The most inaccurate is the method that tries to
estimate an average speed over some
“accumulated distance” between trackpoints.
Due to trackpoint inaccuracies, the line
connecting all track points is a zigzag, even
if the real path of a speed competitor is a
smooth or straight line. Since the length of
this zigzag is always longer than a
smooth/straight line, the “average speed”
determined with the “accumulated distance”
method always overestimates the real speed.
The less accurate are trackpoints (the less
accurate is a GPS unit) – the larger the
estimated “average” speed and the more
impressive is the “achievement”…

Think about it. The GPS reports position with an accuracy of 16 meters (you can't even get that accurate with 5 sats in view). That means the actual position is anywhere within a 16 meter RADIUS of reported postion. That's a circle with a diameter of about 100 feet. That makes it obvioust that what George said is absolutely on point: the sampling rate MUST affect the accuracy of the reported speed.