This article appeared in Cubism For Fun, issue 88, 2012 (possibly with minor editing


Novel Puzzle Rings
By Bram Cohen and M. Oskar van

Puzzle rings [1] are a classical puzzle concept in which a ring which can be worn will,
once removed, easily scramble and become hard to get back into its wearable state.
Nearly all puzzle rings made for the last several hundred years have followed the
exact same design; see Figure 1. We will throughout this article describe puzzle rings
with diagrams showing how the bands are braided on the front, with the assumption
that all bands go straight around the rest of the ring. The quality of the traditional
design indicates that considerable experimentation must have occurred before that
design was settled on as 'best', but then the experimentation was forgotten and only
that one design was passed along between generations. So far no one who is
interested in puzzle rings as puzzles has gone over histori-
cal examples to trace back the history of the design; hope-
fully someone will do that in the future.

In the later 20th century Jose Grant [2] did some experimen-
tation with puzzle ring designs, mostly variants on the tradi-
tional design consisting of either splitting existing bands in
half or adding a fifth band. Several years ago Bram Cohen
begun experimenting with puzzle ring designs, and this arti-
cle will explain his way of thinking about them and give sev-
eral example designs. The methodology of building these
has been that Bram designs them using pen and paper,
then sends them to Oskar van Deventer in
ASCII art (see Figure 2), who then builds them
using a CAD system and prints them on a 3d
printer, usually Shapeways [3].

It turns out that puzzle ring designs are fairly
forgiving, with lots of small braid variations gen-
erally still producing enjoyable puzzles, albeit
usually inferior ones. The trick is to find the
most interesting and difficult designs involving
the fewest number of bands.

The first part of the analysis of a puzzle is which
bands are linked to which other ones. This can
be done by just looking at each pair of bands
without including the others; see Figure 3. If a
band is only linked to one other, that generally
makes the puzzle very easy. To maximize
difficulty each band should be linked to two

Figure 1. Classic
four band design

Figure 2. Sixth Sense in ASCII

Figure 3. Linked rings
(left) and unlinked (right)
This article appeared in Cubism For Fun, issue 88, 2012 (possibly with minor editing)

others, and bands shouldn't be linked to the ones which lie next to them in the solved
state. It may be that for puzzles with large numbers of bands difficulty is increased if
each band is linked to three others, but that hasn't been experimented with much,
and it becomes difficult to make the puzzle scramble-able if the criterion that bands
should not be linked to the ones next to them is kept.

The other main part of puzzle difficulty
is whether for each band the two bands
linked to it would hang in the orientation
they need to be in the solution if you
hold the first band up and let everything
else fall by gravity; see Figure 4.
Puzzles are most difficult when all
bands make the two they're linked to go
the wrong way.

Ideally a puzzle ring should be so easy
to scramble that simply dropping it from
an inch in the air will scramble it. One
way that bands can come apart very easily is if the outermost bands have the pattern
of first going over some bands, then under others, and finally over. That allows them
to hinge outwards. In the Sixth Sense design, the outer bands both hinge out, and
once those come apart the next ones hinge out, making the puzzle want to scramble
very badly; see Figure 5. Another way in which bands can scramble very easily is if a
band first goes over a bunch of bands and then under several more, which allows in
to swivel in place even if it's an interior band. In the Weave Six design (Figure 11) the
outer bands swivel out, and then the next in bands can swivel out, resulting in the
puzzle scrambling by a single big corkscrew motion.

Given the above
criteria we can
analyse the tradi-
tional four-band-
ed puzzle ring
design. Each
band is linked to
exactly two oth-
ers, with the two middle bands not linked to each other.
Every single band tries to make the two it’s linked to go to
the opposite orientation relative to each other that they
need to be in the solution. One might wonder if the two
edge bands could be made to not link to the ones next to
them and have the two middle bands be linked to each
other instead. There is a simple design which has that
property and also the orientations property, but it doesn't
scramble; see Figure 6. In terms of scrambling, both of
the outer bands swivel out and the two remaining aren't
even linked to each other, so it scrambles very easily. It

Figure 4. Directly linked (left) and
misleadingly linked rings (right)

Figure 5. Sixth Sense design scrambles very easily

Figure 6. Impossible to

This article appeared in Cubism For Fun, issue 88, 2012 (possibly with minor editing)

also has a subtle asymmetry, which Oskar likes. The traditional puzzle ring is proba-
bly the absolute best design for four-banded puzzle ring.

The Holistic Ring (see Figure 7) is another four-banded design. In it, unfortunately,
each band is connected to the ones next to it, but they are each linked to two others,
and the orientations property holds for all bands. What's unique is that none of the
bands individually hinge or swivel, but the puzzle as a whole can scramble in one big
holistic move for a very subtle reason, hence the name.

When you get to
more than four
bands there's an
additional design
requirement to
look out for. If you
design things
wrong, the bands
can get stuck around a linkage be-
tween two other bands; see Figure 8.
There is a tendency for the bands to
be able to pass through each other
and get out of this state, but that has a
tendency to feel like it was accidental
or cheating, and isn't very aesthetic.
Ideally the bands passing through
each other should be either clearly al-
lowed or disallowed. If you make the
bands thick enough the passing through can be clearly disallowed, but that has a
tendency to make the puzzle's motion feel very restricted. If the puzzle were made
sufficiently elliptical as a bracelet instead of a ring then it would be very easy to pass
bands through each other and one could go the opposite direction of designing a
puzzle where that phenomenon was forced to happen as much as possible, but that
has not yet been explored.

The Weave Five design (Figure 9) has the property that each band is linked to
exactly two others, no band is linked to either of the ones next to it, and the
orientations property holds for all but the middle band. Because there are an odd
number of bands, it was impossible to achieve the orientations property for all of
them. It has a subtle asymmetry. The outer two bands hinge out, and the rest of the
puzzle then scrambles easily. It also has none of the bands passing through each
other phenomenon. It is much harder to solve than the traditional design.

The Sixth Sense
(Figure 5) is a gener-
alization of the tradi-
tional design. In ret-
rospect it is surpris-
ing that no one
found it before, al-
though it was discov-

Figure 7. Holistic Ring

Figure 8. Bands passing through each
other; middle band stuck (left), bottom
band stuck (right)

Figure 9. Weave Five
This article appeared in Cubism For Fun, issue 88, 2012 (possibly with minor editing)

ered via a painstaking iteration process with the
obvious pattern only being noticed at the end. If you
remove the outer pair of bands, or the middle pair of
bands, or the other two bands, the remaining four
bands form a traditional puzzle ring.

This pattern can be generalized to any number of
bands, with an interesting difference in flavour
between even and odd numbers; see Figure 10. Four
produces the traditional design, three results in a
design which several people have found independ-
ently, and five has not been made yet although is
undoubtedly fun. Odd numbers have rotational
symmetry, while the even numbers are asymmetric.
Larger numbers work as well although it seems likely
that at some point the puzzle won't be able to
scramble well any more. The Sixth Sense is an
extremely difficult puzzle, for which I still don't have a
completely reliable set of steps to solve it even after
solving it hundreds of times.

The Weave Six is a puzzle where each band links to
exactly three others, no bands link to the ones next to
them in the solution, and there is no bands passing
through each other; see Figure 11. Unfortunately to
get those criteria to be met and make it easy to
scramble it was nec-
essary to drop the
orientations require-
ment, which is fol-
lowed by exactly zero
of the bands. It has a
clear asymmetry. It
scrambles easily by
swivelling. It's a fairly
easy puzzle to solve
once you know to start at one edge and work your way to the other, and that the next
band to place is always the one not linked to the last one positioned, but nearly
impossible otherwise. This leads to a pleasant and insightful solving experience. The
weave pattern obviously generalizes to any number of bands, but the same solving
approach applies to more bands, so they don't really add anything to it, and smaller
numbers of bands can be solved in much easier ways, so the version with six bands
is clearly the best one.

[1] Wikipedia, ”Puzzle Ring”,
[2] Jose Grant, “Puzzle Rings”,
[3] Oskar van Deventer, Bram Cohen, “Puzzle Rings”, Shapeways,
[4] Oskar van Deventer, “Puzzle Rings”, YouTube playlist,

Figure 10. Canonical
3-banded (top) traditional
4-banded (middle) and
generalized 5-banded

Figure 11. Weave Six