"Upper limit" of PIE

[Jens Elmegaard Rasmussen]

On Tue, 15 Feb 2000, Stanley Friesen wrote:

> [...] [W]e can
> place a fairly secure *upper* *limit* on how old the divergence of PIE is.
> That upper limit is about 4500 BC.
> [...]

Help me, I'm dumb and ignorant, what is up and down in archaeological
dating? Does the quoted statement mean that PIE split up "no later than"
4500 BC, or does it mean "no earlier than" 4500 BC? Is the present moment
the low or the high end of the scale? This is quite honestly meant as no
criticism, but perhaps there are other language-oriented IE-ists who have
difficulties following an argument that points up and down when it means
before and after (in this order or the reverse). Will anyone stoop to
informing me (us) on this important point?

Jens