$MachineEpsilon gives the smallest machine-precision number which can be added to 1.0 to give a result that is distinguishable from 1.0.
$MachineEpsilon is typically , where is the number of binary bits used in the internal representation of machine-precision floating-point numbers.
$MachineEpsilon measures the granularity of machine-precision numbers.
See The Mathematica Book: Section 3.1.6.
See also: $MachinePrecision, $MinMachineNumber, $MaxMachineNumber.
Related package: NumericalMath`ComputerArithmetic`.