Having just found this thread, I'd like to thank Malmo, Keith Stone, Illinois Master and Scotth for correcting the misunderstandings that many people still appear to have, concerning course measurement. Unfortunately, it seems to be an uphill battle.
I'm a runner, land surveyor, race director and course measurer and I, too, get tired of other runners questioning the accuracy of courses, because their garmin gives a different distance. Garmins are wonderful training tools (I use one myself on a daily basis), but they are not suitable for course measurement (for one thing, they can't be calibrated).
On another note, to answer one of Keith's questions, here's an extract from the manual:
Measure the course. Ride the bicycle over the course, following the shortest possible route as it will be available to the runners on race day. At least two measurements over the course are required for certification. Use the first measurement to establish tentative start and finish marks. Use the second (and any subsequent) measurement to check the distance between those same marks. In particular, during the second measurement, simply record counts at the points already marked during the first measurement. Do not make new marks on the road during the second measurement. If you measure on different days, calibrate both before and after measuring on each day.
So, yes, you can measure on different days (and I have done so for marathons because I prefer to do the whole measurement in one day, rather than segment it, then come back on another day for the second measurement).
Also, as a surveyor, I would never use survey equipment for course certification, unless it was truly a straight line course on a consistent grade. To date, I have never measured a course that fits those criteria (I've been measuring courses for 25 years).
Dave