[0)] Where there is no diacritic on the vowel, then macrons are used for alpha,
iota and upsilon in the headwords.
1) Vowel lengths are not shown where one is expected to know it, e.g. in the
prefixes of verbs
2) Vowel lengths are not shown in closed syllables - strictly speaking, breve
and macron are being used to show syllable weight, not vowel length.
3) Some diacritics imply vowel length.
#0 and #1 and #3 seem consistent with the 4 entries we've explored, and
this is interesting information, thank you.
About #2:
The entry of ἄγαν mentions that there is variation (always so much
variation we're confronted with for this dialect family!) between ᾰγᾱν
and ᾰγᾰν, in a note after the dictionary entry (you remembered cases
like this for your original email, I take it). It'll be great to see an
example of a superheavy syllable where LSJ doesn't reveal anything but
another dictionary will tell us about a distinction, but if breve/macron
usage there were to only indicate syllable weight (interpreted literally
as a "light or heavy?" bit value for the syllable in question), LSJ
wouldn't have bothered with this annotation.
(About my earlier question: It's dawning on me that perhaps someone just
precomposed whatever was typographically common or attested with some
frequency, with LSJ as a model, for which someone must have made decisions.)
Stephan
