In order to measure the capability of a process, taking into account the location of the process average, we need to calculate the distance to the nearest specification limit. This is the smaller of the following two values:

This gives the distance in sigma units between the process average and the nearest of the two specification limits. If this value is negative, then the process is centred beyond one of the specification limits, and something is clearly wrong. If it is positive, but less than 3 sigma units, there is a good chance that some of the output will be non-conforming.

If the process has more than 6 sigma units elbow room, and the distance to the nearest specification limit is at least 3 sigma units, then we can be confident that the majority of the output will be within specification. These are the minimal criteria for which we can expect the process to meet the specifications. In practice, because we cannot guarantee that the output will be normally distributed, and because the '3 sigma limits' only cover between 99% and 100% of the output, it is desirable that the elbow room for a process should be considerably more than 6 sigma units.

It should be stressed once again at this point that these measures, and those on the following pages are only valid if the process is both stable and predictable, i.e. it is in a state of statistical control.