Pakistan, Nov. 1 -- A search result from an old newspaper is an invitation to inspect a page. It is not yet a reliable quotation from that page. The distinction matters whenever a student, journalist or researcher copies the machine-readable text supplied alongside a scan. The searchable version may contain an error that the image makes obvious. If the quotation will support an argument, the reader should make that comparison before carrying it into a new article.

The Library of Congress explains the issue in its Chronicling America research guidance. Optical character recognition, commonly shortened to OCR, converts images of letters and numbers into searchable text. The process can misread unusual type, small print, markings and imperf...