More Details about Vellum and KENPC

After the recent release of Vellum 1.2.5, we wanted to provide a bit more detail about the change we made that affects books available through Kindle Unlimited.

As we mentioned in the announcement, Vellum 1.2.5 addressed concerns we heard from authors who reported receiving a lower-than-expected page count (KENPC) when using Vellum.

Amazon does not provide any information about how KENPC is calculated, but instead assures that it is “based on standard settings (e.g. font, line height, line spacing, etc.)” Unfortunately, as many authors have discovered, this isn’t true in practice. We’ve seen a lot of speculation about these calculations, and so we thought it was best to share specifically what we changed.

For Vellum’s output for Kindle (only), we removed the specification of the “line-height” property, which affects the spacing between lines of text in the book. If you are familiar with CSS, we essentially made this change:

p {
    margin-top: 0;
    margin-bottom: 0;
    text-indent: 0;
    line-height: 1.4em;

That’s it. And while this resulted in some issues with our drop caps, causing us to adjust other aspects of our CSS in response, it should make very little difference (if any) to file size and delivery costs.

This fix was actually discovered and tested by some of the very authors who first contacted us. Because they could confirm that the change addressed their KENPC discrepancies, and because Amazon’s most recent guidelines advise against declaring line-height for main text, this seemed like the right change to make.

We don’t know whether Amazon is intending to penalize authors who specify a line-height, or whether it’s just a bug in their KENPC calculations (See update below). But given their promises of a “normalized” page count, we think it’s a flaw that KENPC is at all affected by formatting.

Again, our motivation was simply to eliminate the discrepancy authors reported when comparing other formatting methods. We have no interest in further optimizing KENPC, or gaming Amazon’s calculations. We sincerely hope that Amazon delivers on their promise of a page count that is actually normalized.

And while that may happen in a future implementation of KENPC, we didn’t feel like this was something we could wait on. We heard from authors who felt that they couldn’t continue to use Vellum if it resulted in a lower KENPC, and we didn’t think this was acceptable.

We’d like to thank the generous and patient authors who reached out to us, helped find a solution, and then tested our fix in Vellum. And we look forward to when authors can spend more time writing and less time fretting over KENPC. Until KU 3.0, anyway.

Update 10/8: Further testing has indicated that this was likely because a 1.4em line-height is smaller than Kindle’s default, and that a value larger than the default would have resulted in an even higher KENPC. Given this potential fluctuation, as well as the reasons above, we still believe that removing the line-height declaration is the best answer.

See also: Even More Details about KENPC.