In this work, a series of principal questions of the existence of documents in the information space is considered. Based on natural concepts of a document as an information object, the questions of document identification, concepts of the instance, copy, and original are studied, along with accompanying concepts of the registry, catalogue, and repository. Requirements for the identification and rules of the existence and use of documents are considered. An approach is suggested to make possible an adequate treatment of copies of documents for the case when copies are more available than originals. Features of an XML-based prototype are described. The work can serve as a basis for constructing document collections of an arbitrary nature, especially for documents naturally distributed in the information space.