A single precision real is a 32 bit signed floating point representation.
Numbers of type single are stored using 32 bits, they require less memory than numbers of type double, which use 64 bits.
However, because they are stored with fewer bits, numbers of type single are represented to less precision than numbers of type double.