Perhaps I used the wrong word..."
Unstable" may not convey well what I mean :
Starting from a loose knot, if
we pull the standing ends and the standing ends only ( and
not the tails ), because the continuations of the tails do not "communicate" more directly with the standing ends ( due to the presence of those intermediate 270-degree sharp turns / collars ), they can remain slack. ( This means that this knot will not
dress itself automatically, as the Zeppelin bend does : at some point during loading,
we will have to pull the tails as well ). In each link, the slack within the continuation of the tail can only be consumed
after the consumption of the slack within the collar. So, the pulling of the standing end does not result in an immediate, directly proportional tensioning of the continuation of the tail. As a result, the pressure from the main bights can force them to bulge, and be dragged into an
oblique angle in relation to the the axis of the loading - away from the optimum
right angle ( where the main forces they had to withstand, as parts of a genuine Zeppelin-like knot, were the
sheer forces
, not the
friction forces.)
On an oblique
rail /
tail pair 
, the side-by-side main bights will
slide towards opposite directions, until the whole knot will settle in a new stable configuration, resembling the spread-out
Cube bend ( M. B18) and the similar B binder :
1.
http://igkt.net/sm/index.php?topic=4122.0 So, surprisingly, it seems that the insertion of those 270 degree collars in between the standing ends and the tails, instead of making the knot
more stable ( as it usually happens when the lines follow more convoluted paths inside the knots ), it disturbs the initial balance of the original Zeppelin bend, and makes it
less stable !
P.S. I have seen that the Zeppelin-like bend shown in Reply#46 behaves in exactly the same way, regarding its vulnerability to
obliqueness during self-dressing... although it is supposed to be more stable, as the links there are the topologically more "complex" overhand knots, rather than the "simpler" slip knots, as in Lucas s Zeppelin-like bend here. However, the twisted, 270-degree collars are present in both knots, and, as I tried to explain above, I believe that this is the real cause of the problem : the continuations of the tails are too independent from the rest of the knot, they do not communicate with their bights very easily. On the contrary, in the
slipped overhand knot Zeppelin-like bend, the continuations of the tails "communicate" directly with their corresponding bights through simple, U turn-like collars - so any slack left in them can be consumed by those bights in no time.
