Writer identification is a difficult problem, it can be considered at different levels according to the applications, details can be observed or more globally the general aspect of a writing. We are going to present a new method to index and identify manuscript and other handwritten texts in ancient documents. Our method is based on Zipf law, originally used in mono-dimensional domains. Here same kind of relation is looked for in the image analysis domain. We use it as a model to characterize the distribution of patterns occurring in these special images that are handwritten texts. Based on this model some new features are extracted and we prove their efficiency for writer identification task.