The chapter investigates the properties and diachrony of the much-debated German was-für construction based on data from historical corpora. It is argued that this construction originated from the was ‘what’ plus partitive genitive construction. The latter is claimed to be a construction stretching over two DPs, the first one consisting of a wh-element and a null noun, the second one being a genitive noun. Given the absence of phonetical evidence for the presence of a null noun in the first DP, it is shown that, during the Early New High German period, this binominal construction was reanalysed as a mononominal construction consisting of the wh-element was in combination with an indefinite NP. A number of properties of this construction (absence of partitive interpretation, possibility to split, etc.) can be explained straightforwardly by means of the diachronic development sketched.