Mikej - That really must have been a while ago then!!  :->  <G>

When I joined LCR, back in '68 (and has been confirmed by a colleague who'd joined previously in '61), the lines from Herstmonceux Castle (HCX086/087, ISTR!) in rural Kent, were already being used for the GTS system!
That was the home of the Royal Greenwich Observatory (between '57 and '88) since it lay directly on the 0-deg Longitude line - so important in relation to exact timing!!
Some short time before 1988, the RGO moved to Cambridge, but rather than reproduce this "antiquated technology", it was decided to create a new system of time determination which didn't rely on any circuits from the new RGO site. Besides, BT weren't going to be able to supply inter-exchange copper circuits any more, with digital (ISDN) about to take over that communication role!]

The two lines carried 'BT Tone' continuously, with gaps in it to signify the incident of the pips. Tone detectors on the two lines saw these "holes" and then operated relays which permitted "station tone" to be output through the GTS equipment. I think it was at the suggestion of Wally Forsyth (one of the Senior "Engineering types" of the day) that "time gates" were introduced to prevent some of the all-too-frequent 'line-breaks' from generating erroneous pips: these gates opened up 35-seconds before each quarter-hour "pips" were due and shut again just 5 seconds after their incidence.]
There was indeed a feed of BT's 1Kc/s tone into the building, but I'm 99% sure that it wasn't 'that feed' that was used for the pips themselves. I do, however, remember Robbie Roberts (one of the TOMs)  requesting to check the frequency of 'Station tone' directly against that fixed-frequency feed from BT on a few occasions - and without the aid of any Frequency counters either!

The new Time system was instead determined from the (then relatively new) GPS system.
LBH generated a "Master" 10MHz reference pulse-train, which was disciplined by 3 GPS satellite receivers (to provide 'arbitration' in case they ever saw - or generated - any differences!) and was then used to sync to "the accurate time" as generated by the National Physical Laboratory (in Teddington) with their rubidium crystal atomic clock. The NPL (still "listening" to the pips received from the LW transmitter at Droitwich) would advise LBH if the leading edge of the final (long) pip drifted from the exactitude of their clock, and request a slight correction from the BH Time system.
This 10MHz reference was used to generate not only the 48KHz AES-3 Word-clock signal (for synchronisation between various bits of digital equipment) but various other references, such as the 32KHz clock used by the NICAM coders. Even the 1KHz "station tone" - although ISTR that the frequency thereof was perhaps set to 1001Hz (or thereabouts) since accurate 1000Hz tone was not "liked" by the NICAM coder equipment of the day (it clashed with the sampling points on the waveform, and produced digital errors) was also directly related to it. The various "intelligent" digital clocks scattered around the building, as well as in the studios, were similarly linked to this reference frequency. [The older 'pulse-driven' clocks - maintained separately by House Services - were then being "withdrawn" as the elderly pulse-driver equipment (which couldn't be tied to this new digital system) began to fail.]

Indeed! I remember using the 'over-ride' key situated on the OBs desk on such as Bank Holidays. Ah! Such memories.

I shall prompt Alan again about when the GTS lines changed: and maybe one or more of the old LCR lags will remember when they changed (if their memories are still that sharp!!

As a by-the-bye, were you aware that Mark Rivers is now a Councillor member of Luton Borough Council?