The mean value of the square of a generalized score function is shown
to be interpretable as information associated with a continuous random
variable. This information is in particular cases equal to the Fisher
information of the corresponding distribution.