Comment by algolicious

I'm a little disappointed that Wirth doesn't mention one of the key reasons that financial institutions use decimal floating point: that binary floating point can't produce exact representations of powers of 10, such as 0.1, and you may want to produce exact results for these situations, because fractions of a cent can really add up over time (see e.g. http://en.wikipedia.org/wiki/Salami_slicing).