Success in generating an acoustic interference pattern with two
coherently derived sinusoidal sound sources will depend on many
environmental factors, but the crossovers *per se* can't be the
source of your trouble. The fact that there are two speakers on
each source, on the other hand, may well be very important. If
the speakers are spaced by an interval comparable to a wavelength
of the sound one would not expect ideal double source behaviour
in any event. If everything else was the same, as I infer from
your note, then I guess that this is the source of your problem.
It is important that reflections be supressed, as you mentioned
the building fifty yards away. You did not mention the ground on
which the demonstration was conducted, and that is probably more
important than the distant building. If you can choose an
acoustically dead surface your weak effect may improve quite
appreciably. Reflections from the ground will produce a four
source pattern, with the contribution from the two reflections
being capable of partially cancelling the stronger two source
pattern if you are unlucky enough to choose speaker and observer
heights particularly badly.
It is hard to find an acoustically dead environment. If you ever
have a chance to visit an anechoic chamber be sure to do so. It
is an unforgettable experience which should teach you how unusual
acoustically dead environments must be. I did some experiments
with tympani back in the seventies without having an anechoic
chamber. There were problems. The nearest thing I have found in
Nature to the anechoic experience is during a snowfall of large
clumps with no wind blowing and new snow on the ground. That is
a notable acoustic environment indeed, but an anechoic chamber is
deader still.