The counting of numbers in the streets of London is a pretty arbotrary procedure.

Arbitrary? Do you actually know how they do it in Britain so you can say with reason it is arbitrary? Here's how it is done in Germany with big demonstrations:

(1) They take a picture from a helicopter. (2) They blow it up to wall size. (3) They put a regular grid on it. (4) They count the number of people in three small grid squares (5) They average the three counts to get an estimation of people within one square. (6) They multiply by number of squares covering the demonstration (with some error correction for the fringes) to get an estimate of the number of people in the demonstration. (7) They repeat the procedure twice with different grid squares to get an idea how variable their estimate is.

Of course, they still could chose not to publish the best estimation for political reasons. But a likely explanation is also that the bias is in the perception of those having a strong motive for wishing some types of demonstrations smaller and some bigger than they actually are.

As I said I don't know what is true in this special case. I do know, however, that the routine counting procedure in Germany is not arbitrary.

Wolfgang