As usual with speakers it is more complex than what has so far been stated here.
I think when considering this question, the application and context are a good starting point.
This question involves musical preference and desired spl.
As I have so often stated with electronic instruments, there is no reasonable point of reference, and so many ills will pass.
However if your choice is natural instruments and you are an experienced listener and really know how it should sound then things are more complex. Even then are your preference solo instruments and small ensembles or the whole gamut is crucial to design. If you want to reproduce accurately from solo guitar, to huge orchestras, with a massive pipe organ and massed choirs at concert hall levels, then you have your work cut out. The latter can pack an awesome punch, and sustained at that.
The next issue is what I return to again and again, the power demands are way above sub level. Little happens below 40 Hz most of the time. The real power is 80 Hz to 2.5 KHz and even to 5KHz and beyond. This is the first problem area and most important to get right. This is where the larger speaker has an advantage. To properly reproduce this range at power takes two drivers in each speaker to really handle this without unacceptable levels of thermal dynamic compression and distortion. It requires substantial motor systems in the drivers.
So a bookshelf with one bass mid per cabinet will not suffice. If power demands are moderate they can acquit themselves well. This gets back to application.
The next issue is that fortunately the ear is forgiving in the last two octaves to a point.
The point comes when you actually get used to a properly balanced non resonant detailed bass.
After that you are ruined for anything else pretty much. My point is this insensitivity goes away with training.
So yes, a crossover from sub to mains is a crossover, and for superior results it has to be an integrated crossover and part of the integral design. Any speaker, any sub does not cut it at this level of accuracy.
I believe firmly that all speakers including any subs have to be part of an overall arching integral design.
Here there is also a difference of opinion between the two sides of the Atlantic. In the UK the view is that speakers should where possible be run full range and allowed to roll off naturally and the subs just gently supplement the last octave as a mirror image of the roll off of at least the mains. This plays fast and loose with phase the least.
On the other hand in the US cutting out the mains is recommended more often than not. I certainly lean to the UK view.
The problem is that the UK approach is more costly as cheaper drivers will not perform optimally without this crossover.
I'm wrestling with this right now. For some time I have been working on a cost effective speaker system, and hope to be ready for some construction in the coming weeks.
For a cost effective system it will demand cutting out the lower 2.5 octaves. This then demands the crossover to the sub be treated with as much care and formality as the rest. It is an integrated design and not a pick and choose one.
What I have found listening to other systems is that the sub is in general far too loud and used crudely to mask a whole host of ills. The end result is far from accurate and realistic reproduction, however much it pleases the owners or demonstrators.
In reproducing systems an accurate, detailed, integrated non resonant bass is so rare you can just about forget it! However when you hear it you know it.
The next question is how much bass do you need? Most music really only requires a response to the 40 Hz range. Movies are a different matter. But how many of us have a steady diet of movie sound effects?
In superior systems a sub should have very little to do and require minimal power. In fact I regard a sub as the most dispensable item in good speaker systems.
As you know in my main system I'm without sub. So how do I do this? All the front short lines are allowed to roll of naturally. Both three front pipes have almost identical F3. The large pipes supplement and the LFE channel mixed in and equalized to make an excellent combined electrical and acoustic crossover.
The surrounds are sealed and roll off 12 db per octave at very close to the fronts. The 12 db point is 45 Hz, with a 12 db point around 25 Hz.
The rear backs 3 db point is 35 Hz and 12 db point 20 Hz. This actually gives me sub performance in the rear.
I do use this system extensively in Dolby PL IIx Music. This is where 7.1 has a huge advantage over 5.1. This highly integrated system does realistically put back the ambiance and space in most two channel recordings. I was playing one of my recordings a coupe of days ago. As many of you know I do not play fast and loose with phase in my speaker designs or microphone layouts. In this two channel recording almost all the applause came out of the surrounds and rears as it should have, very little from the fronts.
In my downstairs system, the speakers have a gentle roll off at 53 Hz and a 24 db per octave roll off below 35 Hz. I use low Q coupled cavity subs to supplement and fill in the roll off. These subs have minimal output and require very little power, yet the bass is more than adequate, bearing in mind this is not a full concert hall spl system. Again application, application and application is key.
To sum up a bookshelf system with integrated subs can work well, but not to reference and concert hall levels, that requires more substantial design.
