I'm converting from markdown to epub. I'm trying to understand how Pandoc decides when to make a pagebreak. Here's the bit that has me confused:

I had a large markdown file, that I broke up into 5 smaller files. There is one h3 header and 10 h4 headers. What I expected was that in the epub either the different source files or the different header sections would be emitted as separate xhtml files. What I have tho is that all 5 files are combined into one xhtml. I made a test doc with the same structure and got separate files for each header section. I can't figure out why I get different results. Any ideas what I could look for?

The 5 files look like this:

1. file 1

    ### h3 header

    text

    #### h4 header 1

    text

    #### h4 header 2

    text

2. file 2

    #### h4 header 3

    text

    #### h4 header 4

    text

3. file 3

    #### h4 header 5

    text

    #### h4 header 6

    text

4. file 4

    #### h4 header 7

    text

    #### h4 header 8

    text

5. file 5

    #### h4 header 9

    text

    #### h4 header 10

    text