If you mean stereo pair, then no, you cannot do that.  in fact, it should be a criminal offence to do so.  If you instead want to play the speakers in perfect sync you need to GROUP them, not pair them.

Thank you for your quick response John. Ok so for better sound etc to group them ?  How do I do this. Many thanks 

In the Sonos App, select the system (middle) icon in the bottom set of your screen and then group via the small up arrow in a box. :)

https://support.sonos.com/s/article/3391?language=en_US

Stereo pairing speakers with such different sound signatures would sound awful, hence not possible.

Lol. One can’t say till one has tried. In my outside patio, I have a Bose 251 pair, a Minirig Bluetooth speaker, and a Yamaha Subwoofer and they hang together quite well sonically, played together from one amp. Ditto when I try passive speakers from two makes when I am switching things around once in a while and have experimented. I suspect it looks more a crime aesthetically than sonically:-).

Anyway, the OP issues seems to have sorted itself out.

Your first example isn’t an example of stereo pairing exactly, so doesn’t really apply.  Your second example may apply, but without knowing the how similar the speakers are in signature and capabilities, it’s hard to say. 

I wouldn’t go so far as to say that stereo pairing 2 different Sonos speaker models in criminal, just not the best option.  The two speakers are not going to sound the same at the same volume levels, so it will not be accurate as a stereo pair.  Sure, it may have an appeal to some extent, but only in a way that it sounds different than what you ‘re used to...similar to how it’s appealing to listen to cheap speakers every now and then.  Regardless, if what you have is two different speakers, the better option is just to group them as you wish, and avoid a weird balance of left and right audio channels.

I know it is not the best option, though as I have with my set up, sound level differences can be catered to in a large way via tweaking the various controls, including the balance control. I had to do a lot of one time work before I was able to get the sound across my patio to be consistent and well balanced.

I have found that once sound level differences are dealt with, sonic signatures from a certain type of speakers are more similar than they different from each other.

YMMV.

In the case of Sonos of course, this option simply isn't there, so this is not relevant in the Sonos universe.

Oh, surely not. Frequency and phase responses that match? Between two totally dissimilar speakers? No way. The stereo soundstage could well be ruined.

I can still recall taking delivery of passive speakers that came complete with lab-measured response plots, to reassure the customer that they were a matched pair. My headphones even claim to be matched within +/-1dB (not that my ears are these days).

Lol. I knew this would be controversial. But over the years I found that a lot of audiophile dogma does not stand the test of real world listening in a domestic environment. Or, more correct to say, in my domestic environment, to my ears.

Here is what I think are the reasons why, apart from looking a little weird, stereo sound via a Quad 11L and a KEF Q100 did not sound off. And why Sonos may be being unnecessarily closed about stereo pairing different models.

1. Speakers addressed to a certain market are more similar in how they sound these days than different - after all, they cater to the same market. Any other than large differences in measured spec gets washed out once room acoustics come into play - that impose a common kind of “distortion” on all speakers in a particular place in a given room. Note here that I would not say this applies to say a JBL Flip 3 paired with a Quad 11L. Their addressed markets are too dissimilar.
2. Where room acoustics can be eliminated as in via headphones or in near field listening, the small difference may still matter.
3. What is more relevant to the sound stage are sound levels - these need to be the same, or as close to the same as not to matter. Now, I agree that what this means for Sonos is that it is easy to achieve these when both units are identical. 
4. This common sonic signature thing also is something that is dismissed as a factor by specialist sub makers. Whereas the maker of speakers that also makes Subs will say just the opposite to promote the sale of their Sub.
5. In the case of Sonos, if grouping works as well as it does, why not stereo pairing? After all even grouping works best when sound levels are equalised - for which the recently found trick of using the volume limiting feature makes it convenient to achieve these by tweaking the limits so that sound levels are equalised at the same slider positions across units.
6. While it is possible that stereo pairing won’t sound good across different Sonos units even if sound equalisation is done, there is no way to definitively say it won’t till it is heard - which Sonos does not permit anyone to do. 
7. Yes, this will still look weird if allowed, but perhaps Sonos need to be so anal as to not even offer it as an option. Especially so in the case of units like Play 1 and Sonos One, or in the generations of 5 units after the first generation 5. In this case, stereo pairs will not even look weird. If someone wants to exercise the option of stereo pairing a 5 and a 1, let them, they know it looks weird and are presumably fine with that.

Digressing a little:

Of my three passive speaker pairs, all of the same basic standmount speaker design, one pair - Dali -  has speakers marked right and left for connecting in that manner per the user manual. I have no clue via looking at them what the difference between the two is and it isn't a surprise to me that I hear no audible difference when they are “wrongly” wired. 

On the other hand, this may be proof that my hearing is defective.

The acoustic signatures of Play:1 and One are different. As I understand it, the acoustics of Five were tuned to be similar to Play:5/g2.

However, in both cases the internal computing/networking platforms are quite different. Leaving aside the acoustic aspect there could one day be features which work on the later model but not on the earlier one. This would leave an unmatched stereo pair in a strange position. 

Another key point to consider is regression testing before firmware release. If a unit was allowed to pair with anything else then such testing would become an N-squared problem, instead of order N. After a short while it could simply become uneconomic to support every combination.

It would be interesting to see the results of the Play 1 and One in a strict DBT in a domestic listening environment to see if they can be picked out.

I have no idea about the nuts and bolts that you refer to of course, except for easily understanding that a unit that cannot play on S2 will not pair with one that does. If the problems are N squared as you say, there is obviously a trade off to be considered from a commercial point of view.