Incorporating data from a primary frequency standard into a time scale

We describe a new time-scale algorithm called AF1. It is designed to combine data from a primary frequency standard with time-difference measurements from an ensemble of cesium clocks and hydrogen masers. Each clock in the ensemble is characterized by a frequency which is estimated from the periodic time-differences and a frequency aging that is estimated using data from the primary frequency standard. The author compares the estimates of the algorithm with the data from our primary frequency standard, NIST-7