[Hadrien]

Yeah sorry Richard, I haven't answered to this e-mail yet. First of all:

I can see why you have chosen a Schoolbook font, in that size, but Schoolbook is rather large in the body, which means relatively few characters per line. This then increases the need for division, if the page is to be justified. Reducing the font size, perhaps choosing one that's more condensed, and increasing the leading, would leave you with an equally legible page.

You can set another font and font size using our Custom PDF feature on Feedbooks.

If you also went for a ragged-right layout, the need for division would be reduced still further. By then there might be so few divisions that it would be practical to eyeball the remaining text.

Not a big fan of ragged-right, it feels like reading a screen instead of reading a book. Still, that's something we could probably add as an option in our Custom PDF feature.

Now I found that the algorithm Feedbooks uses does not observe usual practice in word-division, leading to some odd formations which hold the reader up.

We use LaTeX for our PDF files, so it's the TeX algorithm (like FBReader). Tweaking the settings or extending the hyphenation dictionnary might do the trick.

It's one of those things like em dashes and curly quotes that need to be got right if e-reading is to thrive.

I agree. When em dashes are represented as -- like in Project Gutenberg it's fairly easy to turn them back into em dashes. Curly quotes isn't that easy to fix. Tried a few reg exp patterns (we use reg exp on Feedbooks to turn the text into the right format) but it always ended up pretty badly. Not sure that you can automatically turn normal quotes into curly quotes. It would be very useful if someone found the right patterns to make this work.

I am not for a moment suggesting that you reformat any of your existing ebooks, but these ideas might help improve your books in the future. But even as it is, yours are far better than some of the commercial ebooks I have seen

Tweaking some of the settings for hyphenation or adding a few reg exp would automatically reformat ALL of our e-books. Not a big deal.

There's one thing that everyone seems to forget: the most important thing is the DTD. Plain HTML is far from being the best source format: there's a structure in a book. A chapter is not some text that you align to the center of the page in a slightly bolder font. A footnote is not a link to another page. These elements could be interpreted this way by the reading system, or through the use of a stylesheet. But in the source file, they need to be treated in a different way.