The only known technical factor that I can envision being relevant is that resistance between amp and woofer(s) should be as small as possible, to optimize damping factor and to reduce cable effects that might result from the relatively large currents that flow at low frequencies. But even that is most likely utterly insignificant with respect to the path through the jumper, assuming the contacts involved are clean and tight.
So my instinct is that it is likely to make no difference whatsoever, although I would connect to the woofer terminals for these reasons. But to phrase Albert's good comment a little differently, "who knows?"
Some minor corrections to some of the previous statements, though:
Electrical signals propagate through wires at, very roughly, 1/2 or 2/3 of the speed of light in a vacuum, the exact speed depending on the dielectric constant of the insulation.
Tweeters draw less current and power than woofer or mid-range elements primarily because there is usually less energy in the music at high frequencies than at low frequencies, not because the impedance of the tweeter and its associated crossover elements are higher than those of the woofer.
The woofer doesn't gobble up energy that should be made available to the tweeter. If the cable is connected to the woofer terminals, whatever instantaneous voltage appears at the woofer terminals will determine the current and power into the combination of tweeter + high frequency crossover + jumper, based exclusively on that voltage and the combined impedance of the tweeter + high frequency crossover + jumper. The jumper impedance most likely being insignificant.
Mechanical vibration has no direct relevance to which terminals are connected to. Back emf produced by cone vibration can be a significant effect at low frequencies, and to the extent that it may make any difference would again suggest connecting to the woofer terminals, to produce the most direct connection between amp and woofer.