in reply to Math all gone wrong...

You are not doing enough trials. If you did 10000 trials you would see that the average number of errors approaches 100.