Unfortunately, for old books the modern transcriptions into text are never reliable.
Only the original scanned images can be trusted, and even those can be ambiguous when the scanner had a too low resolution.
The modern text transcriptions are very useful for fast searching, but the results must always be verified with the original scanned images, because the mistakes made by either automated OCRs or by human readers can be shocking, especially for such texts, which may contain unfamiliar Ancient Greek words.
I determined that there is an error in Gutenberg's version of Samuel Butler's prose translation of the Iliad, where somewhere in the catalog of ships Homer's tessarakonta is translated as something other than forty. I think it was fifty.
I briefly looked into the process for correcting Project Gutenberg, but didn't do anything. And I have no idea whether the error was made by Butler (and would appear in print) or by whoever transcribed Butler's work for Project Gutenberg. I would argue that in either case Gutenberg should correct the text, but I can see where opinions might differ if the error is original to Butler.
Only the original scanned images can be trusted, and even those can be ambiguous when the scanner had a too low resolution.
The modern text transcriptions are very useful for fast searching, but the results must always be verified with the original scanned images, because the mistakes made by either automated OCRs or by human readers can be shocking, especially for such texts, which may contain unfamiliar Ancient Greek words.