How old is the reduplicated perfect in Indo-European?

This question has been nagging me lately ever since Phoenix brought up the issue in his latest post Not present heightening after all?. He's been bravely grappling with the topic of Pre-IE, trying to work out where grammatical oddities reconstructed for Proto-Indo-European come from. This requires careful examination of details in one stage of a language to arrive at a plausible earlier stage that can explain better the details in question. This process is known as internal reconstruction. There are a lot of things that still get my noodle tied in a knot myself so it's helpful for brainstorming when someone else asks some pertinent questions in order to resolve issues. The question that Phoenix has got me asking myself is: How old is the reduplicated perfect in Proto-Indo-European (PIE) and from whence does it originate? Reduplication in PIE simply involves repeating the beginning sounds of a verb root as in *bʰebʰore 'she has carried' to convey a special aspect of an action.

Many IEists, like Jay Jasanoff for one, feel that the reduplicated perfect is not really reconstructable for PIE per se but rather that it carried some other function, perhaps a kind of iterative meaning much like its close kin, the reduplicated present (e.g. *di-deh₃-ti 'she gives'). The reduplicated present displays what is often referred to as i-reduplication in order to distinguish it from the e-reduplication seen in the perfect forms. So far, I've been explaining away i-reduplication as a reflex of former schwa in preaccented syllables in early-to-mid Late IE. Simply put, an original *ə would have been 'sandwiched', so to speak, between two consonants resulting over time in the increasing closure of the vowel (aka. increasing rise of the vowel) from generation to generation. A schwa that rises eventually becomes a high-central vowel, /ɨ/, which would be hard to distinguish acoustically from its preexisting fronted counterpart, *i, of known PIE phonology. Hence, I believe that schwa in preaccented syllables eventually merged with *i without much fuss.

I had up to now been explaining e-reduplication in like fashion. I had presumed that both forms of reduplication originally involved a schwa and that perhaps the only original difference between the two reduplications was a matter of syllable boundary. So, perhaps i-reduplication was the result when the schwa was placed in an open syllable while e-reduplication was the result when the schwa was in a closed syllable, or perhaps vice versa. I admit, while clever if I do say so myself, it's a little more ad hoc than I would like. I can't think of a way of proving such a thing.

However, I came to realize that if, for whatever reason, the e-reduplication seen in the eventual perfect forms were not as ancient as the i-reduplicated forms seen in the durative present, then perhaps we could suggest something simpler:

Perhaps while i-reduplication is inherited from Mid IE and has not undergone loss of original-schwa-turned-*i in its reduplication due to an exemption to Syncope grounded in straight-forward phonotactic restraints, reduplication of the *bʰe-bʰor- kind which originally would not have had a perfect function was coined quite late (i.e. during mid Late IE, perhaps concurrent or postdating my Schwa Merger rule when preaccented schwa merged with *i) and was using *e as a reduplication vowel right from the beginning!

In this way, I no longer have to muck around with open and closed syllables or an ad hoc reflex of schwa as *e on top of *i. Instead, preaccented schwa simply becomes *i. End of story. It certainly tidies things up (i.e. Occam's Razor likes it) but I will have to ponder on this some more. Perhaps my readers may be aware of an important detail I've overlooked that negates these possibilities.