Where every line comes from

Nothing is edited: the scan is the scan, the text is the OCR that came with it, and every line has an address.

The scan

Rendered once from the PDF of record, at most 1,400 px wide. A page number here is the same page of the file at the source link.

The text

The OCR published with the item: the Internet Archive's hOCR, its older DjVu XML, or the PDF's own text layer. Where the OCR and the PDF count leaves differently, the Archive's scandata maps them; the two works where that failed say so on their pages.

The address

/d/<work>/p/<page>?l=<line>. A copied link also carries the quoted text, so it still finds the line after a re-OCR.

Corrections

OCR is machine text; read the scan before you quote. Better copies are welcome, see Wanted.