2.15|

$ cd ~

SICP Exercise 2.15

Eva is correct. Procedure par1 would produce tighter bounds than par2. Using the uncertain values r1 and r2 (having a non-zero width), multiple times increases the uncertainity. Therefore par1 overcompensates.

Using both the uncertain values only once, par2 avoid overcompensating and therefore returns tighter bounds.