Personally, I attribute its use (at least in science / engineering) to the numpy, scipy, matplotlib combo. It makes prototyping fast, flexible, and rivals (surpasses IMO) Matlab.
Spoken like someone who hasn't written nearly enough Fortran.
.. or COBOL
and IPython, it's a import part of NumPy/SciPy/MatplotLib family, offers a great many convenience features, such as tab-completion of python functions and a good help system. not to mention the new ZeroMQ messaging design(http://ipython.scipy.org/doc/nightly/html/development/messag...).