In Pharr 835 & 836, he writes:
835. Verbs beginning with a vowel formerly proceded by a lost consonant (usually F [vau], or σ), may take the syllabic instead of the temporal augment, as ἑάνδανον, imperfect of ἁνδάνω (σFανδανω) please.
836. When initial σ has thus been lost, the augment always contracts with the first vowel of the stem, according to the rules (584-5); when the initial F has been lost, contraction may or may not take place....
These two rules seem to contradict. By 836, I would think the imperfect of ἁνδάνω (σFανδανω) please would be ἥνδανον, since the augment supposedly "always" contracts when initial σ was lost. Smyth gave me no guidance. Does the "may or may not" clause for F override the "always" clause for σ? Is this another thing I just have to memorize?
Pharr also gives ἕαδον & εὔαδον as alternative 2nd aorist active priniciple parts. ἕαδον is self-explanatory; it follows the same pattern given in 835 above. εὔαδον raises another question; is this derived by ε+F -> ε+υ -> ευ by contraction?