In this answer to 'Why were TECO variables called Q-registers?' I reference documentation indicating that the original Q-registers were separate from the same-named G-registers. (I.e.,
X1 would store text to a different location than
U1 would store a numeric value, and the two separately-stored values could be retrieved with
It appears from AIM-81 that the early PDP-6 versions of TECO had changed this to remove the G-registers completely, storing text or numeric values in same-named Q-registers where writing text would overwrite the numeric value and vice versa. (There were still separate commands for retreiving the text or numeric value, but you had to use the correct one: 'It does not make sense to TECO to "G" a a Q-register given a number by "U" nor to "Q" or "%" a one given text by "X".' (§V p.5))
According to Lars Brinkhoff in the comments on that answer, this combined storage continued to be the case the on PDP-6 versions of TECO and later ITS versions through 1985. Lars also says he has DEC source code with the merged registers. However, §3.3.2 of the 1985 Standard TECO manual says that 'Each Q-register is divided into two storage areas,' making it pretty clear that these versions for "VAX-11, PDP-11, PDP-10, and PDP-8" (I believe that "PDP-10" here means "TOPS-10") use split storage (though they call all registers 'Q-registers,' regardless of whether you are using X/G or U/Q with them).
So what is the time line and family tree of TECOs that use split versus merged storage for the Q-registers? And is this related in any way to expanding the number of registers that were available?
Please give references to the specific information you used to develop your answer.