Before we discuss this, you should read this article here in the UberQ: “Why Do My Headings/Subheadings Show Up at the Bottom of the Page?”  Make sure you read the part about how eBooks really work, and how to envision the reality of an eBook, behind the screen.

Once you’ve done that, the same reality applies here.  The captions can’t be “required” to stay with the images, not at this point of development, in the technology. To a device, the image is one element, and the captions another; they are not interdependent. We tell the coding to “keep together” the images and captions, but that’s for future-proofing, for the day when the devices and readers do obey that coding.  But for now, that’s really all we can do.

Can't You use a Table, to Make that Work?

I wish we could, but if we put the image in one cell of a table, and the caption below it in an adjoining cell, not only could the problem still exist, (because tables can and do break across screens), but worse, the image itself could then break across screens.  Obviously, that’s not what you want.  Believe me, we don’t want captions coming away from the images, either, and we do everything we can to deter that from happening—but we are not able to guarantee it.

What About Making the Caption Part of the Image?

Well, yes, we can do that—but we recommend against it for two reasons. Firstly, the text (as an image) won’t be as crisp, ever, as real text, and it doesn’t resize, obviously, with the real text, other than when the image is zoomed. Secondly, and more importantly, that would make that text not compliant with accessibility standards—because an eReader can’t read aloud a picture of text—it can only read real text.  If it were you, and you were listening to your book rather than using your eyes to read it, you wouldn’t be very happy with that—now, would you?