[cybmole]

normal is not the same as h1.

ti avoid confusion if the epub has an expllit toc , as an html page, I would delete that, so the only toc is the one which sigil will build from h tags.

that should go thru conversions ok.

AFAIK - calibre on default settings does not create TOC . It does try to tag chapter headers via heuristcs & structure detect, but it does not make a new toc unless you tick force generate....

i think the roman numericals thing is not relevant. once the h2 tags exist ( verified in sigil) then I know of nothing in calibre that will then remove them.

PS after reading a related thread - it could just be that mobi / kindle does not like/support nested headers.
so quick fix would be to change all h1 & h2 to h2 tags. You can still have different styles for what used to be h1 & h3, you'll just have to define & code them


you need to prcoceed carefully step by step.
1. be happe with epub toc as displayed in calibre viewer.
2. put all calibre settings back to default ( i.e. you don't want to have heuristics mess with headers if they are already sorted ) - convert your epub to mobi. - view mobi in calibre viewer.
your toc should be as before - i.e. you should be seeing identical toc in both formats, with same viewer. can you get that far OK ?

PS I am not hure how calibre viewer handles a choice of TOC i.e. if there's hard coned html one AND an ncx file both present. that's why I prefer to get rid of the hard coded one early in the process.