As for the attachment of the lines to the main cord, that is presented by CWarner
(op.cit.) as a Girth Hitch using the bight end of the 2-strand cord. I think that one
can discern this in khipukamayuq.fas.harvard.edu/images/KhipuGallery/VA/VA47076-AS152(a).jpg
esp. in the cord with different colored (blue & brown) strands towards the left.
...
--dl*
====
I have read that in making a pendant cord the khipukamayuq (knot maker/keeper) would select
a cord of the chosen material and correct twist and would stain it to the chosen colours.
This string would then be set in place over the main cord
and then laid up into a two ply pendant cord of the desired twist (Z or S) and showing the desired colour pattern.
I am fairly sure that this is demonstrated in the image referred to by Dan.
If a girth hitch had been used then there would have been two strands of cord around the main cord, plus of course the free end.
...
Derek
No, no, you've got it wrong--get a better magnifying lense or monitor or ... !

Perhaps I should've been more precise: by "Girth Hitch" I mean that the the cord,
which is essentially a 1-strand folded in half and laid up from its center thus into
2-strand cord, is put around the main line and pulled through its eye-end (which
can be seen as a G.H. with parts twisted). This is not only just discernable as I
pointed out in my reference--i.e., esp. in the cord left of center w/blue&brown strands--,
but in the quite clear sketches of such things in
History & Science of Knots(did you not check this?). And this is obviously a much more snug-secure attachment
than would be the reliance on twist to keep tight a turn around something--where
any tension on the pendant would open the turn. (I'm not sure to what your remark
about cord thickness pertains, but I can discern
lay in the turn around the main
line, and shortly out from that line the pendant garners "subsidiaries", one of which
is noticeably thicker than the other & pendant.)
I believe there is a logic to forming the pendants in this manner.
The Khipu were seriously important records, so to loose a pendant because its fastening knot came undone
would be catastrophic. By taking the time to 'make' the pendant on the main cord, its permanence was ensured.
I have also read that the knots recorded in the pendant strings were probably tied, untied and retied frequently,
so whatever knots were used had to be easily made, reliable and easily unmade. This might give some clues
as to their form and method of tying.
... . ... , the record keepers "untied some of the knots which they had in the deposits section [of the khipu],
and they [re-]tied them in another section [of the khipu]" (H. Pizarro 1920 [1533]: 175 and 178).
Well, now, let's not merely mention but use that logic! You have the right idea, and should
see exactly that in the
sort of Girth Hitch I point to. It would be quite a bother to remove
pendants were they attached as you assert: one would have to open the turn and then slide
this widened pendant eye over ALL of the other pendants between it and its new location!!
And how can you imagine a GH being at all liable to coming untied in this context?

In the brief bio of the Harvard researcher it is stated that he has recorded twist
and other fine details of the structures, and I'll guess that he has some closer
photos of the knots, etc.--and sketches and other models to match. (C.Warner's
references include many by Ascher & Ascher, & C.L.Day.) Possibly some of the
current researchers who are reachable for enquiry could provide photos, or
even be moved to newly take them (and host them)?
--dl*
====