I've learned a few things about the XBee Pro type 1 today. I wrote a utility to aid in range testing and reliability measurements. I've noticed that the signal strength goes to the lowest level as you walk away long before the data stops coming reliably. S/N ratio is a much more important measurement than just RSSI. It's easier to display or record ACK failure count change per minute. This is a more useful number because it measures what you want. The data flow!
The first thing you should do is measure the energy level in each channel at the location where the majority of your equipment is. This is caused by interference from other digital circuits. Most people will use the XBee Explorer hooked to their PC. If you're collecting data than the energy measured here is by far the most important of the 2 locations. Of the 12 Channels I've found 1-2 are usually very noisy. 1-2 are quiet. Pick one of these! It's easy using my Sketch.
I've tested a few antenna designs. Wanted to find something simple and small to go with the wire antenna on the XBee. Did some testing with the bottom end of a Bud can. It's roughly parabolic shaped like a DTV dish. In theory this should work at the proper distance from the element. Maybe, but in practice, in the woods, it is very hard to adjust the distance and aim. It worked better without the can. Until...
One time by mistake I turned it around the wrong way so the convex side was facing the antenna. In theory this should not work, but in practice it does! Remember there are many wet trees, bushes, rocks, terrain, reflectors all around me deep in the woods. What the convex side is doing is spreading out all the energy that would have been wasted going the wrong direction, but at least now in the right 180 deg half, towards the target. Some of these reflected waves find their way to the receiver. No more aiming! Without the can, you have to aim the wire antenna in a horizontal plane. Not left and right, but up and down. There are many lobes with the strongest being not where you might expect. With the convex can, this problem goes away! Try it?
The results would be different in space, or even in the desert.
My conclusion after testing in many locations and with different variables:
It does NOT increase the range, assuming you aim the stock wire properly.
But it makes the aiming requirement of a simple wire go away!
You don't even need a signal or Sketch running to test the remote setup anymore.