Interesting observation

FLEX 6700 Interesting observation this evening. I have been using a 6 element quad for six meters. I was given a 5 element m2. So I wanted to do A/B test. My house is 70 feet long and runs North-South. The quad is on a 40ft rhon 25 tower on the North end. On the South end I have a 50 ft crank-up with my hexagonal beam. I have about equal runs of LMR-400 feeding each antenna. I took down the hex and put the M2 on the crank-up. When extended the m2 is within a foot of being the same height as the center of the quad. The Gulf Coast 6m net was tonight and the control is in MS (271 deg 172 miles, according to qrz) Band conditions were poor which was good for A/B testing. I was using headphones. Master speaker was set to 0. I opened one panadapter; rec antenna 1(Quad), pre-amp +30, agc med and mid scale. Headphone slider full left. I opened panadapter two; rec antenna 2(M2), pre-amp +30, agc med mid scale. Headphone slider full right. I did the A/B test by muting either the slice in panadaper one or two. I could tell no difference when listening to either one or two. I could hear the control station talking but intelligibility was only about 25% , however, when I had both un-muted intelligibility was near 50%. It was a very notable difference between listening to both vs either by them self. I did this back and forth over about a 15 min time period to minimize the effects of changing band conditions. My noise runs about S-3. My conclusion to this very un-scientific test is that somehow when listening to both, the noise was being cancelled. Anyone have any better explanation? If this is how diversity is going to work, I can't wait. 73 Roger