Interesting. My variation wasn't that marked. Some of the tests were different, so their scales may have been slightly different (IIRC). On at least one I remember I had when I was umm... 5(?) or 6? then again in mid 20s then again in late 30s - all were within 5 points of each other. I can't remember if it was stanford-binet or something else though. Hrm... no... I think earlier tests were stanford-binet, later were wisc and wj. To that end, the scoring models would have been different on the ones when I was a kid, and I know there's some issues between child scores vs adult scores - not necessarily directly comparable.
Not all IQ tests use the same scoring system. Some use S.D. of 16 others S.D 15 etc.
However, my point was that small variations are not significant. Also, the correlation between IQ scores and intelligence in general is highly debatable - and that's what people are arguing about.