Zero became a number, not just a gap

A placeholder for an empty column is ancient. Treating zero as a number you can add and subtract, with rules, is later. The clearest early statement is Brahmagupta's, in 628.

Babylonian scribes had a placeholder in a place-value system more than two thousand years ago, and it was not always written. Maya glyphs used a zero in a calendrical system by the first centuries CE. What Brahmagupta gives, in the Brahmasphutasiddhanta, is arithmetic: a number minus itself is zero, a number times zero is zero, rules for negatives, and an attempt at division by zero that later mathematics had to repair.

The repair matters. He set a nonzero number divided by zero equal to a fraction with zero in the denominator, and zero divided by zero equal to zero. The modern answer is that these are undefined. The advance was to treat zero as something that enters operations at all. Transmission westward through Arabic mathematics, and only later into medieval Europe, is why European records look late.