I believe that the place I've heard that 10 point rule come from is the Gifted Development Center. I've never seen any actual studies done to support this. The GDC, I believe, came up with that based on what they see with kids whose parents bring them in for testing. I'd have to expect that there is some bias there, though, in that they are quite expensive and something like 84% of the kids they assess wind up with at least mildly gifted IQ scores. Thus, I'd suspect that parents who are willing and able to spend $1-2K on testing for multiple kids are pretty darn sure before putting out the money skewing the results toward more of the kids they test being gifted. This is, of course, just my own conjecture.
I'd be curious to see if there were actual studies done separate from a self selected group like the GDC has.
eta: this is all anecdote as well, but this issue was discussed a few months ago here as well with various people chiming in as to their experiences with IQ similarity or variability among their own kids:
http://giftedissues.davidsongifted.org/BB/ubbthreads.php/topics/142624/1.html
