It is only since about the mid-1950s that technology innovation has become the subject of extensive empirical research and data collection. The possibly dominant part of this work has taken the form of case studies, each based typically on a sample of firms or innovations in a single industry over a short or medium period. Research work of this type has several serious limitations: the sample firms are usually just those which were willing to cooperate; these firms tend to change significantly during the time under study in terms of important characteristics, such as the qualities of workers and management, size, market conditions, products, and methods of production, many of which would nevertheless be assumed to be constant in each firm or the same for different firms; the methods used to obtain the data from firms also vary among studies and the information obtained is often subject to considerable errors. Consequently, the empirical findings that have been reported are neither particularly reliable nor comparable. Some of these serious drawbacks are also shared by the aggregate R&D data for industries and countries. Moreover these data are patchy or missing for the years before the Second World War.