In which a supposedly experienced and knowledgeable kitemaker (me) takes nearly three years to fix an unstable kite when the cure for its malady has been well known to kitemakers for thousands of years..
I've been attempting to get a single skin Serpent kite to fly satisfactorily, single line without a pilot.
An easy job, because an Octopus kite with the same head and bridling already flew well done by lunch time, All that was needed, I so naively thought, was to make a tail (or tail and whiskers in the Thai Serpent style) with the same weight and drag as the Octopus's tentacles.

How wrong I was, and the mistake was right at the beginning: Thinking of it as a long narrow kite.
I should have been thinking of it as a head (the kite) and separately, a tail.

But first, a brief reprise of kite stability theory and two definitions:
A useful way to think about kite stability is to consider what happens when a kite finds itself leaning a bit one way or the other (that is, its spine or centre line no longer pointing upwards). There is only one force available to correct this; the kite's weight, acting at a point (centre of mass) below where the lift forces that are keeping it up act. There is widespread misunderstanding about this- many kitefliers believe that the wind somehow provides a reference by which the kite "knows" to point itself upwards. It doesn't, and cannot. In the absence of gravity, a kite would be just as happy flying at any point on the rim of a cone with its apex at the kite's tether point and its axis in the wind direction.
This weight force acts to get the kite pointing upwards again, whenever it gets a bit out of sorts, but in this it is resisted by various drag, lift, and inertial forces.
If, as a result of these other forces, a kite is unable to rotate back into the 'pointing up' position, it will move sideways across the sky until either reaching an equilibrium point hanging out to one side or the other or, in the limit, 'finding' the ground. at one edge of the wind window. This is superstability, and it's the curse of soft kites because there is no elegant way to shift their centres of mass rearward, (the primary fix for superstability) while shifting their centres of lift forward (the other cure) is often also constrained.
If the weight force moment is sufficient to correct a lean but not before the kite has moved a significant distance sideways, when correcting, it may overshoot the centre by more than the original displacement. If this overcorrection builds into lateral traverses of ever increasing amplitude, the kite is Volatile unstable.
Perfect stability is when a kite corrects from any lean almost instantaneously and without significant sideways movement, or, if the original lean was caused by a wind shift, when the kite moves with the wind as it shifts around without over or under correction.

30m Serpent slots wobbly head stable.
30m Serpent with diagonal slot tail, superstable.
30m Serpent with slots and whiskers, supersatble.
30m Serpent with tail scoops, superstable.

Now, back to that misbehaving Serpent:
I tried the simplest approach first; a long tapered tail. This was superstable in light winds and volatile unstable in stronger winds.
Thinking that the tail did not have enough drag (the Octopus tentacles, although having the same total area and weight, flutter around a lot, so have more drag), I tried a tail with many small diagonal slots (30m green/purple Serpent with diagonal slot tail). It was terminally superstable.
In the next attempt, I progressively added scoops to the head end of the original straight tail (30m green Serpent with tail scoops). This was by the theory that the drag the slots generated was too far behind the head (very long tails or tails with drogues at their end, act like a long tail that gets caught in a tree- they pull the kite inexorably over to one side). This was also terminally superstable- and the more scoops added the worse it became.
Volker Hoberg than offered the theory that there needed to be some slots in the tail just below the head to release pressure there. This is from knowing that lift generated by the first section of the tail could shift the centre of lift rearward by enough to cause the kite to dive over to one side. (30m purple Serpent with slots and whiskers). This version was much improved, though still somewhat superstable, and volatile unstable in winds of more than 20km/hr or so.

These are just some of the many variants tried, but give the general idea as to how resistant this kite was to 'conventional' cures. It was at about this point I recognised my basic error and started to consider the problem from the point of view of the Serpent's head being the kite, and the tail being, well, a tail.

Tails are self supporting; they slope downwards at an angle to the horizontal that exactly supports their own weight, and have three effects: on the kite:
They damp lateral movement that can otherwise build into destructive figure eighting (volatile instability).
Their drag pulls the kite downwind, reducing any inclination to overfly.
Their weight moves the kite/tail system's centre of mass rearward, reducing superstability (tendency to inexorably lean off to one side or the other).
Of these, the last is the least understood, and in my view, the most important. This misunderstanding causes many kitefliers to think that drogues, (which provide only lateral damping and downwind pull) are equivalent to tails- but they aren't, because drogues tend to exacerbate superstability, not reduce it.
I can't think of any other device that simultaneously reduces both volatile instability and superstability- very clever things tails! But to recover from a lean without pulling the entire tail into line behind it, the kite or head must have some freedom to rotate independently. Just a few degrees either way can be enough.

All of which I already knew but didn't consider relevant for some strange reason. I. should have attached the Serpent's head to its the tail less rigidly. Volker's slots (4 of) provided some free movement (hence this iteration was less badly behaved), and the 8 tentacles on the Octopus clearly provided exactly enough, purely by chance. To provide more rotational independence for the Serpent's head I then cut the tail (and the whiskers) off and re-attached them after cutting curvature into the seam, so that the tail drag tension was taken through the centre of the tail not the edges. (30 purple serpent , slots, wobbly head).
It now flies well, is neither volatile unstable nor superstable. The effect of the seam convexity can be easily seen when it's flying; the outer edges of the tail flap a little while the centre line fabric is taut.

1Skin 76 with tail
in stupidly strong wind.
Janggan, 230m tail
Bali 2007.
1773 PearTop with traditional bowtie tail.

1Skins provide a further illustration of this principle: A double streamer tail, weighing about the same as the kite itself, improves their flying in strong winds and turbulent conditions when attached to a single point on the centre trailing edge. (1Skin76 with tail in stupidly strong wind). When the same tail is attached instead by a long "V" bridle with one leg to each outer trailing edge, the 1Skin becomes diabolically superstable in light to mid range winds and violently volatile unstable in strong winds- because it can no longer correct in response to any perturbation without pulling the tail into line behind it.

And, the Janggan (Janggan, 230m tail, Bali '07) deliberately has its tail attached to the head by its full width in order to cause the regular figure eighting which is their desired characteristic- in fact this is just (controlled) volatile instability. In light winds they are superstable (fall off inexorably to one side or the other)- but the Balinese choose not to fly them in light winds. If their tail was attached instead at a single central point, the Janggan would fly as though it was immovably fixed to the sky in all flyable winds.

Tails with single attachment points (allowing the kite to self correct have been used for all of European kite history at least (Old Peartop, traditional "bow tie" tail, 1773), and probably back to the first ever kite. There's nothing else that so comprehensively cures the ills of a not-quite-right kite, or, for the first kite, enabled a leaf being used to float a fishing line out to unexpectedly launch itself into the air and stay there.

5ingle skin update:
As above, the 30m Serpent is now flying satisfactorily..
And, confirmation that all the single skin show kites: Serpent, Octopus, and Centipede, are stackable.
The Aeolian bridle system is now proven: It substantially reduces the requirement to adjust the bridles for different winds.
The 2015 6sq.m 1Skin bridling approach by which, when flown in stronger wind on lighter wind settings, the centre leading edge collapses in before the shoulders so as not to cause diving off to either side is now being used on 3sq.m 1Skins (after 79) to beneficial effect as a range extender.
But, after extensive testing, the lever/pulley bridle system appears to be a failure. Wind conditions are sometimes encountered which cause kites so rigged to luff irretrievably. Perhaps this can be fixed by constraining the range of automatic angle of attack control available, but the remaining benefit may not be worth the complexity.
1Skins (and single skin show kites) are not yet suitable for kite tourist use; they have to be adjusted to suit the wind range and 1Skins still seem to require occasional tuning as the fabric and bridles stretch and settle.
There's still work to do therefore, fortunately, because I'm enjoying this development immensely, failures of understanding like the above notwithstanding.