The link to the test survey. It's similar to the .lss I posted in my last post.

After looking at the survey I think that, while you could solve all of the problems with JavaScript, it would be far simpler to manually place he question numbers in the text.

In fact I started doing that by using {Qcode.qseq} and {Qcode.qseq - n}, where n is a constant used for subtracting.

It's a viable alternative when the survey contains up to 20 or 30 questions. In my case there will be about 70 or 80 questions. A lot of javascript tailoring is used which means all the structure is flexible and I won't really know the exact number of questions just until the survey is launched. If I had to add or remove a question at the beginning that would mean all the qseq codes or the manually set numbers would have to be corrected.

That's why the idea of flagging all the non relevant questions for numbering is not bad. In fact this flag should be accessible in the question advanced setting in one of the next LS versions. Users could then exclude numbering from boilerplate questions and in other situations, like mine for example.

Coding a sequential numbering when using conditions is nearly impossible. E.g. when having conditions to show/hide questions on the same page this requires updating the question numbers for all questions on that page if an additional question is shown due to conditions.