It is the property that characterizes the symmetric distributions other than normal. Represents the density difference of the data as one moves away from the mean and thus the density of the tails of the distribution.

Where m are the central moments of the respective orders.

In particular, a distribution is said leptokurtic (g> 0), platykurtic (g= 0), respectively when: the data density to estrange from the mean, decreases more rapidly than in a normal distribution, the density of the data to estrange from the average, decreases more slowly than in a normal distribution, or the density of the data to estrange from the average is equal to a normal distribution.

Bibliography
GUSEO R., Statistica, CEDAM, 2006
LETI G., Statistica descrittiva, Il Mulino, 1999