If each one is safe if on 2 devices, what makes them become unsafe when put on 1?
There aren't.
It isn't
But why, other than testing aggro?
By exactly the same logic we should not have any ring finals, because if the ring is broken then the circuit becomes 2 radials.
A figure-of-8 actually decreases the chances of a break leading to an overload potential.
Afigure of 8 (meaning a bridge somewhere in the ring) means we can not be sure to which extent the current will flow in every scenario of current draw at any particular times .
The other meaning for figure of 8 is the test we deliberately do on an unknown circuit to see if such bridges exist.
If a ring is designed properly then the loading during is its lifetime is meant to approximate that current draw will not be unduly imbalanced therefore both legs of the ring might be expected to draw very very approximately similar loads.
That is achieved by placing outlets around the ring and not bunched by load x time or by concentrating heaviest loads in the middle one third portion.
Nobody would go to great lengths to actually measure such and make calculations (except for research) but any half decent contractor would take a few mins to make a mental note and perhaps alter things slightly from first idea to make current draw more equal.
That`s the point of having a ring (along with volt drop and R1 + R2) considerations.
We try to avoid much beyond imbalance exceeding 20 to 12 amps on a 32A ring.
The point of this thread on this forum is one ring final circuit (not ring main that`s an incorrect term and applies to something else). One ring final circuit normally consiste of one ring connected at the fuseway.
My ramblings however show thay 2 rings (or more) connected as one ring final circuit are not unsafe and not non compliant but hey yes it is an unusual set up. Just because it is not shown in the OSG as a standard circuit does not make it become "wrong".
Take a spur (spur meaning a branch off the ring, not a S F C U or F C U - again wrong term).
We can spur anywhere off a ring with no more than one twin socket, at any point or at any JB or at the fuseway. That leaves us with a single bit of 2.5 T & E on a 30 or 32 amp fuse or breaker. it passes on short cct and earth fauly but not on overload, the overload protection is provided by the twin socket having a max rating of 13A and the plugtops each having a 13A fuse in making 26A possible (though unlikely) drawn.
Just a note that use of those little "Death Cubes" should be avoided as a rule of thumb ( Cheapo 13a adadaptors 2 way unfused as it means you could have a total of 4 x 13amp plugtops in a twin socket) they could be plugged in at any point on the ring or spur and are best avoided.
Actually, just to be a bit naughty - what is unsafe about a large number of radials, each one having not more than one twin socket and all connected by say one 32A MCB. Not actually unsafe proving the joints are reliable both electrically and mechanically and all volt drop and Zs are catered for.
I wouldn`t like to see it but it would not cause piles of dead bodies to litter the place.