While floating-point numbers are a good fit for engineering purposes, they fail in monetary calculations due to their limited precision. For example, some numbers with a finite representation in decimal notation have only an infinite representation in binary. That often results in a rounding error, which can accumulate over time: