FLOAT
FLOAT[(width,decimals)|(bit_precision)] [UNSIGNED] [ZEROFILL]
You can specify a level of precision for this data type. It
may be from 0 to 24 for single-precision floating-point numbers, and
from 25 to 53 for double-precision floating-point numbers. To make it
ODBC-compatible, you can use the second syntax that specifies the
precision only in bits. If you don’t give a precision with
FLOAT, a single-precision floating point is
assumed. If the SQL mode for the server has the
REAL_AS_FLOAT option enabled,
REAL is a synonym for FLOAT.
This is set with the --sql-mode option when
starting the server.