I have just found out, that ID seems to "scan" a page from left-to-right to define the order of what goes where into the index. Is that correct?
If so, it would be logical that ID also scans text frames from left-to-right. That would explain the strange behaviour I posted above.
Can this be altered? Can I tell ID to scan "from top-to-bottom" instead of "from left-to-right"?
ID starts inthe upper left and reads down, then moves right and reads down again. I suspect that it's picking up the anchored frames after the normal text. Is there a reason you have to use anchored frames instead of pasting the text directly?
Yes, the reason is this:
the text portions (that reside within the RTF file) are texts that are used in multiple documents.
For example, it could be a company adress that is used in 100 docs at various positions. In order to reduce maintenance work, (in case the adress changes: e.g. new phone number), the text frame is placed as a file and links to the RTF file. If I placed real / normal text directly, I would have 100 instances of the adress. This would mean much more maintenance work.
Another advantage is, that if the length of the RTF text changes, I can tell the RTF text frame (in advance) to extend automatically (e.g. towards the bottom of the page). This enables a dynamic layout which reacts flexibly to changes.
It's similar to using an image multiple times. But here, it's text. And flexible.