This is something I've pondered over for a while, as I've seen both used in practise.
Method 1
<ol>
<li>List item 1</li>
<li>List item 2
<ol>
<li>List item 3</li>
</ol>
</li>
<li>List item 4</li>
</ol>
This seems semantically correct to me, since the sub-list is a sub-list of that list item, however it isn't very clean (List Item Content<list right next to it>
).
Method 2
<ol>
<li>List item 1</li>
<li>List item 2</li>
<ol>
<li>List item 3</li>
</ol>
<li>List item 4</li>
</ol>
Cleaner, but it's not directly understandable the list is a sub-list of item 2 (although it's understandable by human inference).
This is purely a concern for semantic markup by the way, both methods present the same content on the page.
So SO, what do you think? Sources where possible are preferable but personal use is good too. I notice that MediaWiki's heirachial TOCs are using Method 1 which encourages me to believe that's the correct use.