dicreased Gain in bandpass

Hi all!
I am designing a standard second order bandpass for the range [0.5-5] Hz :
**broken link removed**

High-pass section:
C1 ,C2 = 1 microF
R1 = 450 kOhm
R2 = 225 kOhm

Low-pass section:
R1,R2 = 48 kOhm ≈ 47 kOhm
C1 = 1 microF
C2 = 0.5 microF

I am using an LM386, and non electrolitics capacitors.
The input signals is suppose to be a square wave of max amplitude of 2V.
The power supply +Supply is meant to be 9V and +Supply/2 4.5 (a 9V battery and a voltage divider made by two equal resistors).
The gain is unitary. The signal I am using to test is very small (from the frequency generator of the oscillator).

I have done it and tested with an oscillator and a frequency generator (integrated in my amplifier).
Varying the frequency it looks like the passband is working (at 1 kHz the signal is almost completely attenuated and there is no more square wave).

Now we go to my questions:
The amplitude is reduced .
1) How can this be possible if my gain is unitary ?
I don't have a graduated oscilloscope so I cannot check how much is the dicrease, but it looks like 1 Volt/Div.
Is there any explanation for this ? The behavior appears even if I remove the High-part section.

2) Is the filter really working ? I mean, can be that the attenuation is an effect related to the tesion decrease ?

3) Is there anything besides the AC-DC switch in the oscilloscope I should change when I play with AC signals ?

Thank you in advance.

The LM386 is a power amplifier. Why not use a standard op-amp.

This same thread with the capacitors in the lowpass filter mixed-up is at www.allaboutcircuits.com .
He might try using 741 opamps but I explained that some do not work from a supply as low as 9V.