The ‘Publishing the Philosophical Transactions’ project studies 350 years of editorial and commercial practices at the Royal Society’s publishing division. The financial data is manageable in large spreadsheets, but there is a huge amount of editorial data (relating to article submissions, refereeing, and editorial decisions) which is difficult for us to use at present. It is contained in hand-written ledgers, which are slowly being transcribed into a database. We estimate there will eventually be records for about 50,000 articles, with linked records for well over 100,000 people. We hope to use digital humanities techniques to analyse and visualise this data; but there are questions about how many errors or misattributions are in the transcribed data, how we would know, and what to do about it. These questions are much more problematic given the scale, which makes manual checking unviable.