Note that the plotted frequency deviation is w.r.t. one value of the US Frequency Standard, and other publications detail the changes which were made to that frequency, so that is not necessarily close to the present frequency of TAI.
You need to be careful with statements like this. Note the scale of the frequency plot is multiples of 5e-9 (!), which is around a million times more forgiving than modern TAI. So I'm not sure what it means to claim "not necessarily close to the present frequency of TAI". Perhaps you can clarify this.
As such, the notion of leap seconds was so far away from the existing CCIR Rec that they did not update 374 to specify them. Instead they created a whole new Recommendation 460. That original recommendation makes it clear that as early as 1970 the CCIR recognized the relevance of the difference between atomic time and universal time, and they specified that the broadcasts should follow universal time.
Is that broadcast time or broadcast frequency? One thing to look for in your literature search is when NBS switched to atomic time. I don't have the paper near me right now. There was leading up to that considerable confusion about which time scale one would use in your lab. Hewlett-Packard tried to solve that in one model of their WWVB receiver with an automatic frequency translator between AT to UT. /tvb _______________________________________________ LEAPSECS mailing list [email protected] http://six.pairlist.net/mailman/listinfo/leapsecs
