Feynman/Einstein/etc and Basic Maths Did famous physicists/mathematicians do basic arithmetic (add/sub/div/mult) in their heads, or would they work them out on paper?
Is there a scale of difficulty for this?  I.e. "Erdos could do 5 digit multiplication in his head, but Feynman could at most do 3 digit reliably".
Or do they all do it on pencil/paper?
 A: Does anyone do all basic arithmetic in their head?  No, so the answer to your first question is that they certainly did some basic arithmetic on paper.
As far as a scale of skill in mental math... I would sincerely doubt this exists.  This would have to be similar to a World Records maintaining group, and maintain a record for all "famous" researchers.  Of course, then they'd have to define famous.
I would be willing to bet that you will only find individual stories of the mental math capabilities of mathematicians or physicists, but with no organization maintaining a ranking system or scale.
Related: Euler carried a math problem to 50 decimal places in his head: http://library.thinkquest.org/22494/stories/Euler.htm  Then again, he was Euler. :)
A: It to a large extent depends on how much logically one thinks. One can do the same by breaking a big problem into smaller simpler ones. In Feynman's 'Surely you are joking Mr. Feynman',  Feynman mentions how he defeats a Japanese abacus expert.. simply by breaking a problem  logically into known steps.
