The uncertainty that accompanies the best estimate of a measurand is usually based on fewer than 20 degrees of freedom, and sometimes fewer than 10. The reason is as follows.

For Type A evaluations of uncertainty, the number of degrees of freedom, v, is related to the sample size, n. Thus, when calculating the mean of a sample, v= n–1. Where measurements are made ‘manually’ (not under computer control), n and therefore ? are likely to be small. Where measurements are computer-controlled and the environment is sufficiently stable, it is easy to amass samples consisting of hundreds or even thousands of values from the same population. We might therefore think that the number of degrees of freedom associated with the uncertainty in the measurand is also very high. However, this is unlikely to be so, since there will probably exist systematic errors that can be corrected for but that will nevertheless leave a Type B uncertainty. Such an uncertainty is generally associated with fewer degrees of freedom. Admittedly, the estimation of a systematic error may also be based on a large number of repeated measurements. The calibration of the 312 -digit DMM by means of simultaneous measurements with an 812 -digit DMM in section 6.1.2 is a case in point. A large number of such measurements could in principle allow us to determine an uncertainty in the systematic error of the 312 -digit DMM that is associated with a large number of degrees of freedom. However, the readings of the 812 -digit DMM themselves have an uncertainty obtained from its calibration report that is likely to be based on fewer degrees of freedom.