The great effort necessary to make historical sources machine readable is a bottle neck for the application of computational methods in history. This makes their reuse for secondary analysis very important; at the same time electronic type setting as well as newly emerging OCR promise to make many more texts available for analysis than so far. As the variety of machine readable data is greater in the historical disciplines than in sociology, the data archives of sociology are only partially useful as model. At least four activities are needed to change the situation: an understanding of the data formats actually used, as well as a drive for standardizing them, organizing models for the preservation of these data for the long run and explicit training of younger researchers for the possibilities of secondary analysis. An agenda for an international workshop to address these problems is derived.