Samandnoah -- you are right that the 2500 hz dip is probably unrelated to the 250 Hz dip. Best guess is that you have the microphone positioned somewhere other than midway between the woofer and tweeter, so that around the crossover frequency you are getting signal from both woofer and tweeter and they happen to be out of phase at the speaker-to-mic distance you are using. One thing the Axiom folks mentioned is that mic placement is one of the critical things to get useful frequency response graphs, even in their anechoic chamber.
JohnK - thanks for the links. They explain some interesting things, particularly why distance seems to be more of a factor in a "wide" room or in a room with treated reflection points than it is in a narrow room.
I'm not sure how to fudge the equations to take into account a less reflective room of the same volume... maybe in the morning when I'm smarter
EDIT -- actually the tvhandbook link does go into some detail there; at least there is a reflection factor in the relevent diagrams.
