It would be a sine wave, if the input was a true square wave
with +/- swing, otherwise it would be a true sine wave with
some DC offset.
The advantage of the bandpass is simply higher Q, but
in theory a lowpass form should be superior since the
bandpass is wasting half of its slopes on the low side
where they do nothing useful. The trick is to use a
lowpass with high Q giving a peak at 1 kHz. The fact
that there is a plateau on the lower side is of no
consequence, since there isn't anything there anyway.
Whether lowpass or bandpass, at high Q there will be
the issue of stability. If the tuning drifts slightly, the
output amplitude may change substantially. This is
one argument for a more-typical flat passband lowpass,
since you can make it less sensitive to drift in components
or the input square wave frequency... if you are willing to
put more stages into the circuit.
The OP doesn't mention where the original square wave
comes from. If it is being generated in a circuit under your
control, you can instead generate at a much higher frequency
and divide it down to 1 kHz. This opens up two new
possibilitues:
1) Use a switched capacitor filter where the high frequency becomes
the filter clock, insuring solid stability even at high Q.
or 2), you can make a cute little D/A type of circuit where you
sum different amounts of the higher components such that
the harmonics cancel. I've seen several designs like this over
the years, but I don't know what they would be called to search
for them in Google. The big advantage is that they work over
a broad range of frequencies, and they virtually elimiante all
harmonics below 2^N - 1 where N is the number of divider
stages in the original (I think!). Basically, a handful of resistors
and a few divider stages is all this takes, and you can follow
it with a simple lowpass to remove the higher trash.
Best regards,
Bob Masta
dqatechATdaqartaDOTcom
D A Q A R T A
Data AcQuisition And Real-Time Analysis
www.daqarta.com
Home of DaqGen, the FREEWARE signal generator