Probably shouldn't continue this as its so off topic, but....
You are theoretically correct, but practically there are problems with norming and IQ tests. I have read that they suffer from limited sample sizes at the high end, ceiling effects where they cannot capture the high end, and fat tails/right skew/bump (to a very minor degree because of these limitations). There are some extended IQ tests which try to measure IQ up to 200 or so, compared to normal tests which only go up to 160 or so, but obviously it's very hard to get many people this high up to calibrate a test. Tests are also only correlated by about 0.6 - 0.8, between themselves, so I think you have to understand a whole bunch of limitations exist in trying to compress human intelligence into a nice neat normal distribution, and you might end up with a little bump at the end of your pretty normal distribution, when you start apply an IQ test in the real world.