IQ's of Scientists

It seems that sixty some years ago, Harvard psychologist Anne Roe gave some IQ tests to 64 noted biologists, physical scientists and social scientists. IQ fan Rodrigo de la Jara has posted some results, including a nice table of results in verbal, spatial and mathematical sections. The highest score among them was a 195 in the mathematical section, which was achieved despite disqualifying all the physical scientists for whom the test was "too easy." Presumably they all or mostly maxed it. The lowest score by anyone was a 121 verbal, which, if adjusted for sixty years worth of Flynn effect reduces to 108.

While this does tend to show that famous scientists indeed do have pretty high IQs (the median combined score was 152), I was more interested in the exclusion of the physical scientists part. So why was the math too easy for them?

Talent might be some of it, but I have to think that training was the key. They were very good at solving math problems because they had had extensive and varied practice. So are other areas of such tests equally affected by such intensive training and practice? It's hard to believe that they aren't. Philosophers and classics scholars have notably high verbal IQs, probably for related reasons.

This reasoning reinforces my prejudice that IQ is mostly a matter of accumulated intellectual skills. These skills may be affected by natural talent, but they are also affected by experience and training.