Jon Babcock wrote:

>
> To restate the question: To identify any Chinese character by means

of

> hemigrams, does it matter how the hemigrams are positioned relative

to each

> other?

To restate the answer, :-) I think it does, although in the great
majority of cases this is superfluous.

This relative rarity may be exploited to come up with a more economic
notation (not specifying the relative position when this is obvious),
but not to free up the system completely from geometric information.

> Please correct me, but it would seem that U+54E1 and U+5504 don't

quite count,

> if their hemigrams are correctly identified, since yuan, U+54E1, is

a huiyi

> graph, composed of 'oral' (Kangxi #30) over 'cowrie' (Kangxi #154)

whereas

> bai, U+5504, is a xiesheng graph, composed of a 'oral' to the left

of a

> shortened form (brachymorph) of 'tripod' (Kangxi #206), in other

words they

> have nothing in common except for the classifier 'oral'.

You are heavily relying on the "etymology" of components, and this
would not be OK for the application I am talking about:
a "decomposed" encoding for computers.

For my purpose, any two components looking the same should be
unified, and any component suitable of having very different forms
should be split in two or more unrelated units. The reason for this
is to avoid occasions for mispelling.

But we are not necessarily talking about the same kind of
application, so we could both be right, in different ways.

But, so, it is perhaps time to ask: why is everybody discussing this
topic? What kind of usage does each one of us have in mind for this?

_ Marco