Visions of Infinity
Tiling a hyperbolic floor inspires both
mathematics and art
Ivars Peterson
Even the most brilliant innovators get their
inspiration from somewhere.
For the Dutch graphic artist M.C. Escher,
such a creative impetus came from a particular illustration in a 1957
mathematical article about symmetry. It gave him what he later
described as "quite a shock" and inspired him to create four artworks:
the Circle Limit series of prints.
 |
|
A circular pattern of
triangles with curved sides, similar to this one, inspired M.C. Escher
to create his "Circle Limit" series of prints.
Dunham
|
The illustration showed a curious tiling of
black and white triangles with curved sides. Enclosed within a circle,
the alternately colored triangles became progressively smaller as they
approached the circle's perimeter.
The concept of infinity had long intrigued
Escher, and he had sought to capture this elusive notion in visual
images. One strategy that he employed was to create repeating patterns
of interlocking figures. Although Escher could imagine how such arrays
could extend infinitely, the actual patterns he drew, of course,
represented only a fragment of an infinite expanse.
In another approach, Escher tried to fit
together replicas of a figure, such as a fish, that diminish in size as
they spiral into or recede from a point in the middle of a square or
circular frame. However, he wasn't entirely satisfied with these
efforts.
The mathematical drawing—an illustration of
the so-called hyperbolic plane—that had so startled Escher offered him
a precise, aesthetically pleasing way to depict diminishing figures
within a circle.
The article containing the diagram had been
written by H.S.M. Coxeter, a mathematician at the University of
Toronto. Today in his nineties, Coxeter continues to focus on the
interplay of symmetry and geometric shapes. His research covers topics
such as the mathematics of kaleidoscopic patterns.
 |
|
A hyperbolic quilt
sewn together from pentagons of cloth, with four pentagons meeting at
each corner in the pattern.
Ferguson
|
Coxeter had sent Escher a copy of his
symmetry article as a thank you for permission to reproduce several of
Escher's periodic drawings as illustrations. The men had met in 1954 in
Amsterdam at the International Congress of Mathematicians, where there
was an exhibition of Escher's work. That encounter led to
correspondence between the two, which continued until Escher's death in
1972.
Escher in 1958 mailed Coxeter a print of
"Circle Limit I"—the first fruit of the artist's venture, inspired by
Coxeter, into hyperbolic geometry. In the decades since Escher's
initial foray, the Circle Limit prints have motivated many
mathematical investigations and artistic endeavors. Each year,
mathematicians and computer scientists continue to add to the
literature of hyperbolic tilings and related mathematical topics.
Basic principles
Despite the many manifestations of
Escher-inspired research, it all stems from the same basic principles
of curved geometry.
For example, if you draw any triangle on a
sheet of paper and add up its three angles, the result is always 180
degrees. When you draw a triangle on a saddle-shaped surface, however,
the angles invariably add up to less than 180 degrees.
Just as a flat surface—like that of a sheet
of paper—is a piece of the infinite mathematical surface known as the
Euclidean plane, a saddle-shaped surface can be thought of as a small
piece of the hyperbolic plane. Picturing what the hyperbolic plane
looks like on a larger scale, however, requires some mind-bending
ingenuity.
Freelance mathematician Jeffrey R. Weeks of
Canton, N.Y., suggests making "hyperbolic paper" from a large number of
equilateral triangles as one way to get a feel for the hyperbolic plane.
Taping together equilateral triangles so
that precisely six triangles meet at each vertex produces a flat sheet.
In contrast, assembling equilateral triangles so seven triangles meet
at each vertex produces a floppy, bumpy surface. The more triangles you
use and the larger the resulting sheet, the more closely it resembles
the hyperbolic plane.
A similar construction can be done with
pentagons. Mathematician and sculptor Helaman Ferguson of Laurel, Md.,
has made an intriguingly wrinkly hyperbolic quilt by sewing together
pentagonal patches of fabric so that four pentagons meet at each
corner. It's an unruly blanket that refuses to lie flat, he says.
Such constructions are not the only way to
visualize the hyperbolic plane. More than a century ago, French
mathematician Henri Poincaré introduced a method for
representing the entire hyperbolic plane on a flat, disk-shaped surface.
 |
|
Escher's "Circle Limit
IV," as seen in this computer-generated rendition in gray and white,
has an underlying tiling pattern in which four hexagons meet at each
vertex.
Dunham
|
In Poincaré's model, the hyperbolic
plane is compressed to fit within a circle. The circle's circumference
represents points at infinity. In this context, a straight line,
meaning the shortest distance between two points, is a segment of a
circular arc that meets the Poincaré disk's circular boundary at
right angles.
Although this model distorts distances, it
represents angles faithfully. The hyperbolic measure of an angle is
equal to that measured in the disk representation of the hyperbolic
plane. So, a repeating pattern made up of identical geometric shapes in
the hyperbolic plane transforms, when represented in a Poincaré
model, into an array of shapes that diminish in size as they get closer
to the disk's bounding circle.
For his Circle Limit prints,
Escher worked out the underlying rules of these disk models and
developed his own method for constructing a hyperbolic grid, relying on
his skill and intuition to create the geometric scaffolding he needed.
Coxeter's "hocus pocus" mathematical text wasn't much help, Escher
later remarked in a letter to Coxeter. Nonetheless, Escher executed the
drawings with extraordinary accuracy, Coxeter comments.
"The first time I saw a print of 'Circle
Limit III,' I said to myself, 'that is the most beautiful example I
have ever seen of the Poincaré circle model for hyperbolic
geometry,'" says J. Taylor Hollist, a mathematician at the State
University of New York at Oneonta. He's documented many historical
interactions between Escher and scientists.
Besides its beauty, "Circle Limit III" also
presents a puzzle. For some reason, in this particular case, Escher
drew a pattern of lines somewhat different from that in Coxeter's
original drawing. The main arcs seen in "Circle Limit III" meet the
circumference at a specific angle very close to 80 degrees rather than
precisely 90 degrees.
Coxeter was able to demonstrate that each
arc is of a type known to mathematicians as an equidistant curve. It
bears the same relationship to a hyperbolic straight line as a line of
latitude does to the equator on the surface of a sphere.
 |
 |
|
Douglas Dunham has
developed a computer program that can transform one hyperbolic tiling
pattern into another. In this case, he has used the program to convert
the crosses of his rendition of Escher's "Circle Limit II" pattern
(left) into stars (right). (Dunham)
|
When Coxeter worked out trigonometrically
what the proper angle of such a curve in Escher's print should be, he
obtained 79 degrees 58 minutes, again confirming the accuracy of
Escher's draftsmanship.
Useful vehicles
Besides intriguing professional
mathematicians, Escher's Circle Limit prints and their
repeating patterns prove to be useful vehicles for becoming comfortable
with and teaching hyperbolic geometry, says Douglas J. Dunham of the
computer science department at the University of Minnesota at Duluth.
"Even to mathematicians, hyperbolic geometry is not that familiar," he
contends.
Dunham and his students have written several
computer programs to generate hyperbolic patterns, particularly those
made up of repeating motifs colored in various ways.
Mathematicians use a standard notation for
describing a mosaic made up of identical tiles, where each tile is a
polygon with a given number of edges of the same length and the same
number of corners or vertexes. On a flat surface, there are three such
tilings. A tiling in which six equilateral triangles meet at each
intersection is designated {3,6}, one in which four squares meet at
each intersection is {4,4}, and one in which three hexagons meet at
each intersection is {6,3}.
The same notation applies to regular tilings
of the hyperbolic plane. A tiling where four pentagons meet at each
vertex is labeled {5,4}. In general, for polygons with p
sides, meeting q at a vertex, the result is a hyperbolic
tiling when (p - 2) multiplied by (q - 2) is greater
than 4.
 |
 |
|
A simple Celtic knot
pattern (left). Such a woven pattern can be converted into an array of
interlocking rings, then depicted using hyperbolic geometry (right). (Dunham)
|
Escher's "Circle Limit IV," which features
interlocking devils and angels, is an example of a {6,4} tiling. In
other words, the underlying hyperbolic grid consists of hexagons that
meet four at each vertex.
Dunham has developed a computer program that
transforms a hyperbolic Escher design from one tiling pattern to
another. For example, he can transform the {8,3} pattern of crosses in
"Circle Limit II" into a strikingly different {10,3} tiling, where the
central figure is a star.
The same pattern can be transformed into an
array of starbursts with any number of rays, Dunham notes. Indeed,
there is an infinite number of hyperbolic tilings available for such
transformations. The use of different motifs and color schemes
increases the possibilities even further.
At a conference on mathematical connections
in art, music, and science held last summer at Southwestern College in
Winfield, Kans., Dunham described a novel extension of some of these
ideas. He had worked with Celtic knot patterns, used centuries ago in
Ireland and elsewhere to decorate religious texts. Such knots are
formed by weaving a ribbon into an alternating over-and-under pattern,
then joining the two ends to form a continuous band.
Dunham now has a computer program that
generates intriguing hyperbolic versions of these ancient designs. He
can also construct examples in which rings interlock in the
over-and-under pattern characteristic of Celtic knots.
 |
|
A pattern based on
transforming a computer-generated rendition of Escher's "Circle Limit
III" so that a pentagonal shape appears in the middle instead of a
square.
Dunham
|
Interestingly, Escher himself incorporated
an intricate pattern of interlocking rings within a circular frame in
his last woodcut, "Snakes." Dunham can show that this pattern is
closely related to a hyperbolic variant of a Celtic weaving. Although
Escher and Dunham approached these patterns from different
perspectives, their intellectual common ground is apparent.
In a 1960 essay later translated and
published in the book The Graphic Work of M.C. Escher (1961,
Duell, Sloan and Pearce), Escher noted, "The ideas that are basic to
[my art] often bear witness to my amazement and wonder at the laws of
nature which operate in the world around us.
"By keenly confronting the enigmas that
surround us, and by considering and analyzing the observations that I
have made, I ended up in the domain of mathematics," he continued.
"Although I am absolutely without training or knowledge in the exact
sciences, I often seem to have more in common with mathematicians than
with my fellow artists."
Constructing Escher
Nowadays, mathematicians, computer
scientists, and others have a variety of speedy computer-based methods
for generating hyperbolic patterns and tilings. M.C. Escher didn't have
such technology at his disposal. Neither did Henri Poincaré and
other 19th-century mathematicians who drew various pictures of the
hyperbolic plane. They relied on the traditional tools of
geometry—compass and straight- edge—to create their diagrams.
 |
|
A geometric
scaffolding of lines and points outside a circle aids the construction
of the hyperbolic pattern within the circle.
Goodman-Strauss
|
In an article to appear in the January 2001 American
Mathematical Monthly, however, Chaim Goodman-Strauss of the
University of Arkansas in Fayetteville suggests what these procedural
details might have been. He offers techniques and instructions for
drawing by hand some tilings of the Poincaré model of the
hyperbolic plane.
"Remarkably, this may be the first detailed,
explicit synthetic construction of triangle tilings of the hyperbolic
plane to appear," Goodman-Strauss notes.
His directions are built out of basic tasks
familiar to a student of Euclidean geometry: bisecting a line segment,
drawing a parallel through a given point, drawing a perpendicular
through a given point, constructing a circle through three given
points, and a handful of other operations.
A suitable combination of those activities
enables one to construct, for example, the hyperbolic line that passes
through two given points. To achieve this, one must create a geometric
scaffolding of lines and points outside a Poincaré disk to guide
the drawing of arcs and points within the circle's boundary.
Goodman-Strauss worked out the method by
extending his expertise in Euclidean geometry to encompass the types of
curves and angles necessary to represent hyperbolic structures. He
admits that there's probably nothing original in his contribution.
"Surely, this was all well-known at the end of the 19th century, just
as it has long been forgotten at the dawn of the 21st," he remarks.
Nonetheless, reviving long-lost construction
techniques has value. Such exercises offer an illuminating window on
not only Escher's art but also the remarkable work of earlier
mathematicians who explored non-Euclidean geometries.
"It is wonderfully satisfying to make these
pictures by hand, patiently, with pencil and paper, compass and
straightedge," Goodman-Strauss adds. "I encourage you to test this
theorem for yourself!"

References:
Dunham, D. 2000.
Hyperbolic Celtic knot patterns. In Bridges: Mathematical
Connections in Art, Music, and Science; Conference Proceedings, R.
Sarhangi, ed. See http://www.sckans.edu/~bridges/.
Goodman-Strauss, C. In
press. Compass and straightedge in the Poincaré disk. American
Mathematical Monthly.
Hollist, J.T. 2000. M.C.
Escher's association with scientists. In Bridges: Mathematical
Connections in Art, Music, and Science; Conference Proceedings, R.
Sarhangi, ed. See http://www.sckans.edu/~bridges/.
Further Readings:
Coxeter, H.S.M. 1996. The
trigonometry of Escher's woodcut "Circle Limit III." Mathematical
Intelligencer 18(No. 4):42.
______. 1998. Angels and
devils. In Mathematical Recreations: A Collection in Honor of
Martin Gardner, D.A. Klarner, ed. New York: Dover.
______. 1969. Introduction
to Geometry, 2nd. ed. New York: Wiley.
Dunham, D. 1999.
Transformation of hyperbolic Escher patterns. Visual Mathematics
1(No. 1). Available at http://www.d.umn.edu/~ddunham/isis4/index.html
or http://members.tripod.com/vismath/dunham/index.html.
______. 1999. Artistic
patterns in hyperbolic geometry. In Bridges: Mathematical
Connections in Art, Music, and Science; Conference Proceedings, R.
Sarhangi, ed. See http://www.sckans.edu/~bridges/.
Escher, M.C. 2000. The
Magic of M.C. Escher. New York: Abrams.
______. 1967. The
Graphic Work of M.C. Escher. New York: Ballantine Books.
Hargittai, I. 1996.
Lifelong symmetry: A conversation with H.S.M. Coxeter. Mathematical
Intelligencer 18(No. 2):35.
Levy, S. 1993. Automatic
generation of hyperbolic tilings. In The Visual Mind: Art and
Mathematics, M. Emmer, ed. Cambridge, Mass.: MIT Press.
Schattschneider, D. 1990. Visions
of Symmetry: Notebooks, Periodic Drawings, and Related Works of M.C.
Escher. New York: W.H. Freeman.
Escher's artworks are
featured at http://www.worldofescher.com/.
Information about H.S.M.
Coxeter can be found at http://www.math.toronto.edu/~coxeter/,
http://www.science.ca/scientists/Coxeter/coxeter.html,
and http://www.georgehart.com/coxeter.html.
Helaman Ferguson has a Web
site at http://www.helasculpt.com/.
He describes his hyperbolic quilt at http://www.helasculpt.com/gallery/hyperbolicquilt/index.html.
Additional information
about hyperbolic tilings can be found at http://aleph0.clarku.edu/~djoyce/poincare/poincare.html.
Sources:
Douglas J. Dunham
Department of Computer Science
University of Minnesota-Duluth
10 University Drive
Duluth, MN 55812-2496
Web site: http://www.d.umn.edu/~ddunham/
Chaim Goodman-Strauss
Department of Mathematics
University of Arkansas
Fayetteville, AR 72701
Web site: http://comp.uark.edu/~cgstraus/
|