There is a reason to preserve page numbers in ebooks. While correct academic quotations can be excellently made without page numbers, quoting an electronic version, the same is not true with retrieving information quoted somewhere else (in an old paper edition). So for example if an existing book contains a sentence like "This topic is discussed in book aaaaa in pages xxx-yyy, (the exact edition is quoted in a reference) how do you find easily the exact range of pages, without page numbers? The same happens when a book has an index: the index item is often not found literally in the text, and a page number is an handy way to find the reference. Of course, the index can be improved (in an *ML edition) with cross-links, but transforming an index into a cross-linked version is a lot of work, and has to be done by an expert, while reading a page to find a reference is much less work, and is usually done by a (relatively) expert. Some just answer: then do an HTML or a TEI edition. This I don want: I cannot, and I do not want to learn, I prefer working with text, and do more texts. And I prefer using text instead of *ML. Moreover if I keep page numbers, conversion to *ML with page numbers will be much easier than having to retrieve the numbers from the images. Some say: page numbers are ugly in txt. It is the same people that want to have an *ML version, so why do they bother? Please take the txt version with number, do your *ML and leave the txt alone. Of course, having page numbers in Tom Swift might be too much. But at least if a book has an index, I believe that page numbers might be useful, even in txt, and we should recommend to keep the information. Carlo