> Note too that it is highly likely that in the past rape of lower casts was accepted in some form, which would bring any selected IQ genes to the lower casts.

There is preliminary evidence from genetic sequencing that Indian castes have been highly endogamous for at least 3000 years.

India is a vast place with an enormous population, and there's resistance to continuing this research, for fairly obvious reasons. But continue it will, you can't un-sequence a genome after all. So we may yet see a robust answer to this sort of question.

Surely once caste is established intelligence can't be selected for any more strongly than under any other conditions - the rules of the game are fixed at that point.

Your final point about selection via coercion holds more water, but one has to believe that the subjects would be explicitly chosen for their intelligence, rather than for any other features, which seems difficult to support.

As a rule all women breed (there are exceptions of course). In some societies high status males breed with multiple females and low status males have none. In most societies today marriage is 1:1, though there is lots of cheating but in general it means that some women will breed with low status males just for someone - thus all genes get carried on.

As I said I assume the powerful selected politically powerful mates for their girls, which might or not be intelligent as well. However it is possible that high status was high intelligence. The lower castes not following that.

Look, the above is all just barely possible. you will have to work with me here...

There is a negative correlation between IQ and fertility.

I haven't seen the science so I might be wrong, but from what I can tell this is correlation not causation. Which is to say intelligent people [in modern times] are less likely to have kids (birth control) and not that intelligent people are less likely to be able to have kids if they want to.