-t versus no consomant in 3p sg verb forms in common IE

[Anthony Appleyard]

  It is usually said that the 3rd person singular in IE ended in -ti
(primary) or -t (secondary). But Greek has the forms {luei} = "(he)
releases)", {elue} = "(he) was releasing".
  {elue} may well < I.E. {eluet} : notice that in Greek it adds an "n
ephelkoustikon" if the next word starts with a vowel, which doesn't
happen in the similar imperative form {lue) = "release!" which didn't
lose a final consonant; as if the n-ephelkoustikon replaces an earlier
etymological t-ephelkoustikon.
  But re Greek present {luei}: did this form come from *{lueit}?; or
perhaps it never ended in a {-t} in the first place. IE *{lueti} would >
Attic Greek **{luesi}.
  Perhaps in early IE times the final -t was only present when the verb
had no noun subject, and ultimately derives from an adhering postposited
pre-IE subject pronoun.