Who is to say the two transformers will be exactly identical? The primary inductanc eon one will not necessarily be the same as the other. For instance, with no-load on the secondary, the primary takes a small current because basically it acts like a big inductor across the AC voltage. It might be 10 henries and at 50Hz, the impedance will be: -
\$2\pi f L\$ = 3142 ohms - this will take an RMS current from a 220V AC supply of 70mA.
Bigger transformers will generally have bigger magnetizing inductance but just because two transformers look the same it doesn't mean they'll have the same inductance - one transformer could have 10H inductance and the other might be 5H.
The result is that two-thirds of the primary voltage will be across the primary with the higher inductance. The cetre point of the two series connected secondaries will also be offset by the same amount. Remember this is on no or light loads.
This is important to remember if you think that the centre point of the two secondaries could be used as a "mid-point". You might be lucky but more likely it'll be out by at least 10% or more. Why do I say 10%? A mains transformer doesn't really take care to control its primary magnetizing inductance and as there are no air-gaps to "manage" the flux in the core, the resultant inductance can vary substantially.
One of the mechanisms that can control flux density is the actual core saturating a little bit - both cores will be roughly operating at half flux level and although it is a benefit not to have saturation, this in turn can exacerbate the primary inductance differences even more.
OK I've exhausted the problem of the primaries not being well-matched and this basically means you can't use the centre-point of the two secondaries for anything meaningful because it will be unpredictable as to what percentage voltage it has on relative to the full voltage developed across both secondaries.
As load current increases this becomes less of a problem because the dominant primary current starts to become the load current reflected onto the primary and this will be substantially greater than the magnetizing current.
As for other reasons, Spehro's answer covers this.