The article presents the effectiveness of the Mahalanobis distances 
in case of the text origin identification. 
In order to calculate this metrics were used text features 
of the original texts and the texts generated on the basis of the original texts. 
As a generation method there were used the synonymy and Markov chain method.