Ah, I missed their explanation. The 196% and 338% are still bizarrely high figures, given that the variance of a Poisson distribution equals the mean. 338% variance, and you don't ask what kind of distribution you're measuring?