We have performed evaluations of the major systematic errors in NIST-7
with an overall uncertainty of less than a part in 10(14). The comple
te evaluation process has been separated into two parts. With a comput
er-controlled, digital servo system and some new measurement technique
s, we now perform core evaluations (second-order Zeeman and Doppler sh
ifts, cavity pulling and phase shift, line overlap and some electronic
shifts) with an overall uncertainty of less than one part in 10(14) i
n just a few days of measurements. The complete evaluation of all smal
l and subtle effects in both the physics and electronics requires a fe
w hundred days of data. But, these small effects are not variable at t
he 10(-14) level and their infrequent evaluation does not detract from
the operational accuracy of the standard.