float(n) is an approximate numerical data type. n (which must be between 1 and 53) specifies the number of mantissa-bits. If not specified, n defaults to 53. real is a synonym for float(24). double precision is a synonym for float(53). n between 1 and 24, float(n) requires 4 bytes for storage. For n between 25 and 53, it requires 8 bytes of storage.