An increasing number of documents in companies and other organizations are now only available electronically, and exist in several versions updated at different times. In order to provide efficient support for temporal textcontainment queries (query for all versions of documents that contained one or more particular words at a particular time) temporal text-indexes are needed. In this paper we present DyST, a dynamic and scalable temporal text index. The goal of DyST is to provide the same efficiency in terms of search cost and space usage as the previous approaches developed for small and medium document databases, while at the same time providing only logarithmically increasing search cost for very large databases. We present the architecture of DyST and describe how inserts and searches are performed. Based on a prototype we will also present an evaluation of performance based on real-life temporal documents.