Sorry, do you mean to say the word starts with the character LATIN SMALL LETTER N WITH TILDE, or literally LESS-THAN SIGN? The former does indeed sound like a mess, but the latter might be deliberately mutilated (or amputated) as part of how the dictionary templates manage homonyms—typically in the format headword<!>integer.

EDIT: You were not talking about the stupid thing and were, in fact, talking about words that start with ñ. There's nothing specific in my code that discriminates against non-ascii text that I know of. However, there are headwords in Lilitika that start with Latin-1 characters, and to reply to this comment from Izo:

quoting Izambri, Duke, the Findible League:

Now I realize that dictionaries arrange entries this way:

A
B
C
...
a
b
c
...

So in a double alphabetical order, giving priority to words beginning with upper case.

Is that how things have been working since the first day? Just wondering. No problem with that.

...the Lilitika dictionary does not do this, either. So I'm thinking maybe the site started using a binary collation order in the Annie db, against my wishes.