So, why would we not want to encode Mayan numerals now? First, a quick overview of Mayan hieroglyphs. I'm not an expert; this is just what I've gleaned from such resources as http://www.famsi.org/mayawriting/calvin/glyph_guide_i.pdf , and I'm sure to make some mistakes. Also, this deals with petroglyphs; for handwritten glyphs in codices, there are similar issues and probably some additional ones.
Mayan hieroglyphs are primarily syllable glyphs, combined into blocks, and usually stacked into a two-column arrangement. However, there are some non-syllabic logograms and ideograms, among which I think the numbers count. Columns typically read like this: AB KL UV CD MN WX EF OP YZ .. .. Note that this 2-by-2 format is not purely a matter of visual layout. The Popol Vuh is only known from the Latin transcription of the K'iche' (although the text itself asserts the existence of an "original", presumably hieroglyphic codex, version), yet it is easy to hear the echoes of this two-by-two (and occasionally three-by-three) layout in its poetic rhythms, even in translation. A single block (just one of the letters above) combines multiple parts, reading in an approximately left-to-right and top-to-bottom order, just as a single Hangul character combines several jamo. However, unlike Hangul, the combining rules were not canonical (or if they were, they were extremely complex). Aesthetic and poetic issues come into play. But unsurprisingly, rules for writing dates, a common use of numbers, were more stereotyped/canonical than other writing. Here are a few intrablock orderings in crude ascii format: 122233 122233 122233 144444 144444 Or: 11122 11122 11122 33333 Or even: 12559 33559 33679 44889 Typically, numbers are rectangular, so they'd tend to fit into the blocks above in positions 1, 3, or 9, respectively. (In modern spoken Mayan languages, numbers go along with numeric classifiers, similar to east asian languages. However, I don't think that those classifiers were written out phonetically, as any reader would know intuitively which one would fit in context.) Furthermore, I think there are special rules for "face glyphs", which can combine incomplete versions of certain other glyphs at the forehead(?), cheek, and chin(?) positions. Also, each of the digits 0-10 has a "face form" unrelated to the standard digit, and for 11-19 hybrid forms are used (for instance 10/3 for 13) I don't speak any CJK language, but I've briefly skimmed over the unicode rules for those languages, especially Korean. As far as I can see, the facilities that have been built there will help with full Mayan hieroglyphs, but it seems likely that further tricks will be needed. Given the complexity of the block-combining possibilities, I don't think it's possible that each glyph element will have multiple code points for each of its possible positions within a block. CONCLUSION: Therefore, I don't think it's premature to encode a single, horizontal-barred version of each of the digits 0-19. As I've said, these glyphs would see usage far, far beyond the rest of the hieroglyphs, including in math teaching and page numbers. Vertical-barred and face-based digits and combining rules can be left for later.
