Well, I lucked out. We had a one day "heat wave" on Friday as predicted - temps got into the upper 40s, maybe even a bit higher. A lot more pleasant to ride, although the course was more crowded with fair-weather joggers, walkers and cyclists, from whom I endured no small dose of enmity.
Calibrated at home, then went to the course after lunch. Met the race people and located the finish line, which was on an extension inside a two-plus loop. Then did the ride as follows:
Went to a point on the loop just beyond the extension, common to both (Point A). Rode the extension to the finish took a reading, then zeroed out my Protegé, which I used to lay out mile splits. Rode the course backward, laying out splits and taking readings at Point A and took a reading on the other side of the loop near where the desired start was - I just picked a point by an easily recognized lamp post and called this "Point S."
Then I rode to A, took another reading, for 1 complete loop.
Rode to S, took a reading, then rode to A & took a reading. 2 complete loops and 2 rides of all segments. I continued on to S a third time and took a reading.
After getting home and recalibrating, I figured out the total distance of extension, 2 complete loops A to A, plus one A to S. Came up 66.46m short of 21,097.5m, which I added the next day w/ a steel tape (and by which time the temperatures had dropped below freezing). By today it was even colder (12F at 7 AM) when the runners raced over the course for the first time. I heard that they actually had to break ice film that had formed on the cups at the water stations