In other words, what is the longest possible wavelength which is physically detectable, in principle? Or, the slowest frequency? 

I know 3-30 hz are used for submarine communications, because ELF penetrates seawater, so those are certainly detectable.  Below that, I don't know of any principle that would forbid detection, but I imagine constructing a transmitter antenna would be impractical.  IIRC long bolts of lightning propagate ELF.  Also disturbances in Earth's magnetic field.  

One principle that forbids detection, I think, is the uncertainty principle. The detection time needs to be longer than the wave period (or half, to the order of magnitude). I wonder if there are other fundamental limits. 

Yeah, I was wondering if QM principles would impose a limit.  If you need a charged particle to accelerate, say in an antenna, it seems like there could be some minimum energy below which electrons would simply not flow back and forth and no emf wave propagated.

And there's size problems maybe, if you had a bar of metal that was light-hours in length and trying to sustain some very low energy flow.  Seems like you'd have random noise wipe out any attempt at signal.

Any help from physics-trained members, like @swansont, would be welcome.    

45 minutes ago, TheVat said:

Yeah, I was wondering if QM principles would impose a limit.  If you need a charged particle to accelerate, say in an antenna, it seems like there could be some minimum energy below which electrons would simply not flow back and forth and no emf wave propagated.

And there's size problems maybe, if you had a bar of metal that was light-hours in length and trying to sustain some very low energy flow.  Seems like you'd have random noise wipe out any attempt at signal.

Any help from physics-trained members, like @swansont, would be welcome.    

You’re in a regime where classical physics gives a good description if it involves an antenna, but I think the “time of detection” concept has a classical application. If you don’t emit or detect for long enough the sine wave will be truncated, so it will have higher-frequency harmonics and you might detect that, or detect nothing at all.