If I've read your post correctly, you probably got destructive interference at the measured frequency:
Bass is omnidirectional, and reflections off nearby walls will (depending on phase relationships) either reinforce or cancel subsequent output from the bass driver at the same frequency. You were probably measuring at a frequency where out of phase reflections caused a null in response. Google "Quarter wave cancellation" for a good explanation of this phenomenon.
From my own set-up/measurement experience, I'd note the following:
This effect will be audible below app 200ish hz, and will dominate the in-room response below +/- 125hz in most untreated rooms. If you use Hemholtz devices (bass busters) you might get some relief from this until you get down to +/- 75hz, where these start to fail.
The best way to level match a sub is with a real time analyzer (RTA) and/or EQ. I use Audyssey which combines RTA, crossover, and EQ in the digital domain. It is automatic and (for the recent XT32, at least) tremendously effective.
Velodyne makes a stand-alone version called the SMS-1 that also works pretty well. However, its automatic mode is useless and the much better manual mode requires a fair bit of time and effort to get right. If you use it strictly for set-up, it will help you place your subs for smoothest FR and you can remove the device from the chain after you've optimized placement. Alternatively, you can leave it in the signal path and EQ the subs if you wish.
Re: "replacement" vs "augmentation", well....
If you run your mains full-range and cross at 40hz (augmentation) without any room EQ, you're still going to face the issue of room effects. Even if you use the SMS-1 to move the sub(s) around until you find the best position for minimizing these effects (per the above), you'll probably be left with pretty significant room induced irregularities, particularly above the x-over point. If you cross the subs higher, you can use optimized placement to extend the benefits of smoothed FR up to the new, higher x-over point.