Problem with sequential double IF staining with rabbit antibodies - (Sep/27/2012 )

I am trying to do double staining with two rabbit primary antibodies in a sequential way. Let's call them rb anti-alpha and rb anti-beta. I set up 4 conditions:
(1) No-rb anti-alpha --> anti-rb Alexa488 --> 4% para fix 5 min --> No rb anti-beta --> anti-rb Alexa594
(2) No-rb anti-alpha --> anti-rb Alexa488 --> 4% para fix 5 min --> rb anti-beta --> anti-rb Alexa594
(3) rb anti-alpha --> anti-rb Alexa488 --> 4% para fix 5 min --> No rb anti-beta --> anti-rb Alexa594
(4) rb anti-alpha --> anti-rb Alexa488 --> 4% para fix 5 min --> rb anti-beta --> anti-rb Alexa594

In a sequential staining like this, the most worry comes from the 2nd secondary antibody (594 here) binding to the 1st primary antibody (anti-alpha here). So the fixing in the middle is a measure to prevent this.

The result of this expt was as follow:
(1) No signals in 488 and 594
(2) No signals in 488 and 594
(3) Signals in 488, but not 594
(4) Signals in 488 and 594

I was expecting signals in 594 for condition 2, but I did not get it. I thought it was a technical error, but the same result was repeated 3 times. Since there's no 594 signals in condition 3, there should not be any significant binding of anti-rb 594 to anti-alpha. So that would mean the 594 signal in condition 4 is real. But without anti-alpha, I did not get the 594 signal in condition 2. So I'm very puzzled. Can someone please help? Thanks!

Have you tried doing the 594 ab first rather than the 488?

Haven't tried that yet... I am about to do it this weekend or next week.

Btw, now that I have taken images of everything, there IS actually some staining of 594 on condition 2. But it's extremely weak comparing to condition 4. If the intensity of 594 in condition 4 is 10, intensity of 594 in condition 2 is only 4. Normally, this indicates that the anti-rb 594 is binding to anti-alpha. But comparing condition 3 with condition 4 indicates that such binding is very little.