Abstract :
The computation of the International Atomic Time, TAI, is made in two steps. First, an intermediate time scale, EAL, is established from the data of about 100 clocks (almost all commercially made), in many laboratories. Then TAI is derived from EAL by frequency steering based on the data of a few primary frequency standards. This organization ensures the continuity and the availability of TAI, as well as its accuracy. It was also expected to optimize its stability. However, comparisons with the time scales directly produced by some primary frequency standards in continuous operation (primary clocks) show that EAL may be subject to systematic frequency variations. If it is found that the operation of the commercial clocks is responsible for the variations, should TAI be simply an average of the data of a few primary clocks? The BIH would be reluctant to adopt such a solution because it wishes to maintain as large an international basis as possible for TAI. It tries to identify the source of the systematic variations, in order to use the data of both commercial and laboratory primary clocks, with proper weighting.
