Crowd count-SF march

I decided to use this methodology for counting people at the IAC march 10/26/02. I started at 12 o clock at civic center to count people already at the end point. I got a pretty accurate count of 433. The march started to pass by UN plaza at 12:25, adn it was clear that there were so many people that I couldn't possibly press my clicker fast enough to keep up. After reaching 1023 - totally underestimating the huge number of people passing by in the first 5 min, I started counting by tens. Because about 10 people were passing per second or more, I couldn't only vaguely keep up with my clicker, but I reached 20,000 - but this was haphazard guessing and I couldn't see the other side of the street, plus parts of the march got diverted down side streets as it got backed up down market.This includes no estimate of people who used BART etc either. I decided that an alternate method would be to take the time of the march 12:25-1:57pm and and multiply by people per second. that would be at least 8 per second for 70 minutes of that time, probably going up to 15/second during much of this time (stretching all across market street) and then 3-5 per second for the other 20 min. So that would be a minimum 70x60x8+(3x20x60)=37200.