> We know that Sanskrit /u/ is not a normal reflex of any laryngeal
And yet we also know that Sanskrit /u/ is sometimes illogically found where
we do expect a laryngeal - e.g. roots ending in laryngeals take -au not -a
in the perfect 1/3sing parasm. Also note that in the dual final -u: (lke
final -i: and final -e) does not cause elision of a following a-, as if
something invisible were in the way.
Peter