Underflow describes a condition in computer arithmetic where a computed value is too small to be represented by its Floating-point type, often rounding to zero. This quiet loss of tiny, significant numbers can subtly corrupt further calculations, mirroring Overflow's opposite challenge.