Like others, I am doubtful that this is the best way to go about it. But to answer your question, the main area where I've found Project Gutenberg's epubs coul

Like others, I am doubtful that this is the best way to go about it. But to answer your question, the main area where I've found Project Gutenberg's epubs could be improved is in their navigation outline (the toc.ncx file). For example, they often use top-level headings for each line from the title page, then put the entire book under the last line. Whereas other books are closer to what you'd expect, albeit at inconsistent levels of detail. For my project, I abandoned their TOC's altogether and created a simpler format. The images are also at a bare-minimum of resolution. In some cases, higher-quality versions are available in the public domain (such as on Wikimedia Commons). Most of the books are also scanned on archive.org, and so can be referenced there in facsimile. These tend to be higher-resolution scans (although those are all monochrome that I've seen). For corrections in the works proper, I have occasionally submitted corrections by email but never received a response. Otherwise, they are perfect, and I thank them for their outstanding work. EDIT: There are also rare cases (I think Seneca was the one I came across) where the id's are not unique across the book, even if they are within the HTML files. I couldn't find anywhere in the EPUB specification that would require this, yet for practical purposes I think they should be made unique across the book, since the division into HTML files is arbitrary. Further to that, there are some PG books that have a unique (serial) ID on every paragraph. Again, this is not required, but it's extremely helpful when it's there (for anchor referencing). It would make the whole library more usable if this were applied consistently, and the serial id's are apparently mechanically applied.