@2fd7551e, for old books the standard OCR models of Tesseract (which is used by ocrmypdf) are often not sufficient. Look here for better models: https://github.com/UB-Mannheim/tesseract/wiki#tesseract-models-for-historic-prints. 

 @476c1b14 thanks. Sadly, in my case, the book is from the 1950s. So has a fairly modern font. Just an incredibly bad scan!