Proceedings of a Symposium held at National Physical Laboratory, March 1953, pp. 21-30, published by HMSO. Key points from the article include:
in building the “calculator” (i.e. LEO) Lyons’ intention was to get it into operation as quickly as possible because they felt that until it had actually been in use over a period of time for clerical purposes, “the optimum form of such equipment could not be decided”.
Pinkerton states that modifications and “additional features” were required to the EDSAC design that the LEO is based on “to make the installation effective on clerical work” and that this included a “larger store, means for introducing data into and extracting results from the calculator much faster than was possible with the EDSAC, and a foolproof method of checking data recorded to the machine”. Pinkerton suggests that although LEO has been in use for over 18 months, it is “not yet as reliable as would be necessary for carrying out a regular and intensive programme of clerical work”. http://www.chilton-computing.org.uk/acl/pdfs/Automatic_Digital_Computation_Symposium_Mar53_text .pdf
page32, see