Instigated (or better inspired) by Dan’s feedback, I would like to expand the discussion of gi gu so as to include a discussion of names of all four (or seven) vowels. First, a few comments on gi gu, the sign for i-vowel. To begin with, as pointed out by Martin, the Ma ṇi’i bka’ ’bum just refers to it as gug, which is, by the way also recorded in the Bod.rgya (s.v. gug). It is also called gug gu (ibid., s.v.), which may be the result of a phonetic extension of gug, and represents a pre-stage of gi gu. If we consider the description of i-vowel in sources such as Rong-zom-pa’s dKon cog ’grel (p. 137.5: a ni bkug pa ni i) and his sMra sgo ’grel pa (p. 427.12: a bkug pa ni i yin no); sMra sgo ’grel pa (pp. 427.24–428.1: i ni bkug pa dang sbyar ro), it becomes clear that gi gu and its variants must have meant “bent” or “curved,” as i-vowel is described as “bent/curved a” (a bkug pa). Under these presuppositions, the possibility that gi gu could have been derived from gri gug “a curved knife,” because of the graphic similarity, seems less likely. Nonetheless it is not impossible, particularly if we consider the tendency to leave out r-subscript while speaking, which can be found not just in today’s Central Tibetan dialect but also even in some archaic sources.
In addition it should be also mentioned that gug “curve” or “bend” is said to be a collective name for the signs of i and u. The term gug is opposed to kyed or bkyed (perhaps meaning) “erect/upright,” which is a collective name for the signs of e and o. See the Bod.rgya (s.v. gug kyed).
Now we come to zhabs kyu, the Tibetan name for the sign for u-vowel. Its etymology should be something like “curve/bend [at the] foot.” As for ’greng bu, the Tibetan name for the sign for e-vowel, it must mean something like “a little stroke that is upright.” What about rna ru or na ro, the Tibetan name for the sign for o-vowel? It seems to mean “antelope’s horn.” See the Bod.rgya (s.v. rna ru).
In general, dGe-chos has pointed out that although it is no longer evident in certain scripts such as Rañjanā script, we can see in most letters of what he calls Māgadhi script that the graphic signs or shapes of letters (yig gzugs) are suppose to imitate or mirror the forms or shapes of the tongue during the articulation of the sounds. As examples, he provides retroflex, dentals, and palatals (dGe.chos-1: 254). Of course, the sMra sgo and its commentary (most probably by Rong-zom-pa) have already described how the “seven verbal impulses” (ngag gi ’du byed bdun) or articulations of the seven vowels (i.e. a, e, i, u, o, ṛ, and ḷ) occur comparable to the “seven physical impulses” (ngag gi ’du byed bdun), such as yielding weapon (mtshon ’debs pa), dancing (gar), working (las byed), and so on. Thus (1) a is called “straight letter” (drang po’i yi ge), (2) e “erected letter” (sgreng ba’i yi ge), (3) i “bent/curved letter” (bkug pa’i yi ge), (4) u “lowered letter” (smad pa’i yi ge), (5) o “raised/exalted letter” (bstod pa’i yi ge), (6) ṛ “tight letter” (sgrims pa’i yi ge), and (7) ḷ “loose letter” (klod pa’i yi ge). For details, see the sMra sgo’i ’grel pa (pp. 422–423). Sanskritists should be able to enlighten us on this (provided such an idea is found in Sanskrit sources).