From Möbius Bands to Klein-Knottles

Year: 2012 Authors: Carlo H. Séquin

Core claim

The twistedness of the two Möbius bands determines the regular homotopy type of an immersed Klein bottle.

Topics

regular homotopy, immersed Klein bottles, Möbius bands, topological classification

Domains

topology, differential geometry, surface immersions, mathematical art, sculpture, visualization

Methods

geometric construction, regular homotopy analysis, decomposition into Möbius bands, shape deformation

Media

paper strips, glass models, computer-generated figures

Paper text

The text below is the locally extracted OCR/Markdown version of the paper. Raw PDF files remain local and are not published here.

Bridges 2012: Mathematics, Music, Art, Architecture, Culture

From Möbius Bands to Klein-Knottles

Carlo H. Séquin

CS Division, University of California, Berkeley

E-mail: sequin@cs.berkeley.edu

Abstract

A construction of various immersed Klein bottles that belong to different regular homotopy classes, and which thus cannot be smoothly transformed into one another, is introduced. It is shown how these shapes can be partitioned into two Möbius bands and how the twistedness of these bands defines the homotopy type. Some wild and artistic variants of Klein bottles are presented for their aesthetic appeal and to serve as study objects for analysis.

1. Introduction

This paper extends the study of the regular homotopies of tori presented last year in Tori Story [17] to the realm of Klein bottles. Topologically, immersed Klein bottles are non-orientable surfaces of Euler characteristic $\chi =0$ , with no boundaries or punctures. Surfaces are in the same regular homotopy class if they can be smoothly transformed into one another without ever experiencing any cuts, or tears, or creases with infinitely sharp curvature; however, a surface is allowed to pass through itself. The paper by Hass and Hughes [9] states (pg.103): Corollary 1.3 (James-Thomas): There are $2^{2-\chi }$ distinct regular homotopy classes of immersions of a surface of Euler characteristic $\chi$ into ${\mathbf{R}}^3$ . This tells us there should be four distinct Klein bottle types that cannot be transformed smoothly into one another. However, I have found no publication that shows representative pictures of those four types. The single-sidedness of these objects also makes it conceptually more difficult to visualize these shapes. But some good discussions and e-mail exchanges with Dan Asimov, Tom Banchoff, Matthias Goerner, Rob Kusner, and John Sullivan helped clarify the situation. All Klein bottle types can be built from fusing together two Möbius strips, and their twistedness plays an important role in the determination of the type of a given Klein bottle. Thus I include this decomposition in the images below and start this exposition with a discussion of these important building blocks.

Figure 1: Deforming a Möbius band of type ML and thereby changing its apparent twist compared to a rotation-minimizing sweep from (a) $+180^{\circ }$ (ccw), to (b) $0^{\circ }$ , to (c) $-180^{\circ }$ , and to (d) $-540^{\circ }$ (cw).

2. Möbius Bands

A Möbius band (Fig.1) can be constructed by taking a rectangular domain – for instance a paper strip – and connecting two opposite edges in reversed order, i.e., after executing a $180^{\circ }$ flip. For any topological analysis the surface is permitted to pass through itself; thus we can always execute one or more of the Figure-8 Sweep Cross-over Moves described in Tori Story [17]. This allows us to add twists in increments of $\pm 720^{\circ }$ to a closed toroidal loop. The same applies to single-sided, non-orientable Möbius bands, and thus there are two regular homotopy classes for immersed Möbius bands, and they differ in the amount of

Séquin

“built-in” twist by exactly $360^{\circ }$ . All the shapes depicted in Figure 1 belong into the same regular homotopy class (ML) with a built-in left-turning twist of $180^{\circ }$ . This twist should be measured when the sweep path forms a planar circle; if the 3D shape of the sweep is changed, the “apparent” twist as compared to a rotation-minimizing sweep may also change. Mirroring, on the other hand, would turn the left-turning Möbius band (ML) of Figure 1a into a right-turning one (MR), and thus cast it into a different regular homotopy class.

Several of the stages of different apparent twistedness depicted in Figure 1 have been exploited by various artists: The almost circular shape (Fig.1a) can be found in a wedding band [21] (Fig.2a). The shape shown in Figure 1b has been celebrated in a sculpture by Max Bill [1] (Fig.2b), and the configurations (1c) and (1d) appear in drawings by M.C. Escher [8][6] (Figs.2c and 2d).

Figure 2: Artistic Möbius bands: (a) wedding band, (b) sculpture by Bill (from [1] used by permission), (c, d) drawings by Escher (from [8],[6] used by permission).

If we focus on the edge of the circular Möbius band (Fig.1a) and include a tiny sliver of the band’s surface (a neighborhood) with it, then this edge-band forms a double loop with an apparent twist of $360^{\circ }$ . This double loop can be unfolded into a figure-8 shape without changing the apparent twist as explained in Figure 18 in [18]. This unfolding of the edge will stretch the narrow Möbius band into an extended surface that resembles a two-pouch basket (Fig.3a), but which topologically is still a Möbius band. Further deformations can be applied to this edge band, creating different kinds of “baskets” or “goblets.” In particular, we can further un-warp the Möbius band edge into a circle. This regular homotopy will change the edge band’s apparent twist by $\pm 360^{\circ }$ . If we let the twist add up to $720^{\circ }$ we obtain the Sudanese Möbius band [12] depicted in Figure 3b. Alternately we can let the twist cancel out to zero and then obtain another interesting warped surface - which happens to be equivalent to a Boy surface [2][3] minus a disk (Fig.3c). Some of these shapes play an important role in the analysis of various Klein bottles.

(a)

(b)

(c) Figure 3: Deforming the Möbius band edge: (a) into a figure-8 shape with an apparent twist of $360^{\circ }$ ; or into a circular loop: (b) with $720^{\circ }$ of apparent twist, or (c) with zero apparent twist, when the Boy surface is cut at the red line (photo from the Archives of the Mathematisches Forschungsinstitut Oberwolfach, used by permission).

From Möbius Bands to Klein-Knottles

3. The Classical Klein Bottle

We begin our investigation of the realm of Klein bottles with a formal construction of the well-known “inverted sock” shape. We start with the parameterized, decorated rectangular domain depicted in Figure 4a. As for the case of torus construction, we first merge the two (horizontal) edges marked with parallel cyan arrows to form a generalized cylinder. For now we give this tube simply a round or oval profile (O) (Fig.4b). We then close this tube into a J-shaped loop (Fig.4c), so that its two ends can be joined with the reversed orientation indicated by the labeling and by the two anti-parallel brown arrows in Figure 4a.

Since we are now dealing with single-sided surfaces, I will use a somewhat different coloring scheme from the one I had used for tori. While I maintain the color red for the meridian bands, I split the domain into two halves with different background colors and give their center bands a fully saturated color, either blue or yellow, to mark the parallel parameter lines; these two regions will always form the two Möbius bands into which a Klein bottle can be partitioned. Parallel to these central bands I draw lines of less saturated color (olive, purple) that will wrap twice around the Klein-bottle loop, thereby executing an even number of $180^{\circ }$ flips. Thus a key difference to Tori Story is that not all parallels are the same anymore. Because of the reverse labeling along the two vertical edges of the rectangular domain, there are only exactly two parallels that meet up with themselves; those form the center-lines of two Möbius bands with $\pm 180^{\circ }$ twist. All other parallel parameter lines form double loops with a $\pm 360^{\circ }$ twist.

(a)

(b)

(c) Figure 4: Formal construction of a Klein bottle: (a) its rectangular domain; (b) the domain rolled up into a tube; (c) the tube bent into a $J$ -formation with its two ends lined up in a concentric manner.

For the tube with the O-profile, the closing of the loop is most conveniently done by narrowing one end and inserting it sideways into the larger end of the tube - forming a J-shaped loop (Fig.4c); this configuration properly lines up all the numbered labels. Now the two concentric ends are merged by turning the smaller one inside out. This yields the classical “inverted sock” Klein bottle, named KOJ (Fig.5a). In this case, both “special” parallels form two Möbius bands of opposite handedness. Correspondingly, the KOJ Klein bottle can be partitioned into two Möbius bands (yellow and blue) having opposite handedness as shown in Figures 5b and 5c. We can denote this symbolically as:

$\mathbf{MR}+\mathbf{ML}=\mathbf{KOJ}$

The Tori Story paper [17] spent much effort to analyze the amount of twist built into the toroidal ring, because tori that differ in their amount of built-in twist by $360^{\circ }$ belong to different regular homotopy classes. Interestingly, for the classical Klein bottle twist is a non-issue! By rotating the handle around the symmetry axis of the “inverted-sock” turn-back, any amount of twist can be added or removed [14]. This is a consequence of the fact that in Figure 4a the labeling at the left and right sides of the fundamental rectangular domain can be shifted cyclicly by any arbitrary amount; but there will always be exactly two labels that are lying on the same parallel parameter lines. These two parameter lines then form the centerlines of the two Möbius bands of opposite handedness into which this Klein bottle can be decomposed.

Séquin

(a)

(b)

(c) Figure 5: The ordinary “inverted sock” Klein bottle resulting from the construction in Figure 4: (a) the complete Klein bottle KOJ; (b) the lower half of it is a right-handed Möbius band (MR); (c) the upper, left-handed Möbius band (ML) shown flipped over.

4. Figure-8 Klein Bottles

When merging the two parallel edges marked by cyan arrows in Figure 4a to form an initial tube, we are not forced to form a round, circular O-profile. Instead we may form a figure-8 cross section (Fig.6a) or an even more complicated, multiply-rolled generalized cylinder, as was discussed for tori [17]. This results in various Klein bottles that may belong to different regular homotopy classes, and which thus cannot be smoothly deformed into the classical “inverted-sock” shape.

(a)

(b) Figure 6: Constructions for figure-8 Klein bottles: (a) figure-8 tube; (b) the tube twisted through $180^{\circ }$ so that the two ends can be merged into a toroidal loop; (c) a new way to line-up the number labels.

(c)

For the tube with the figure-8 profile, there are a few different ways in which we can fuse the tube ends with the required reversed orientation, and they will result in different Klein bottles. For instance, we can bend the tube into a simple toroidal loop (an O-shaped path) and give the figure-8 cross-section a $180^{\circ }$ torsional flip (Fig.6b). This flip can either be clockwise (right-handed: R) or counter-clockwise (left-handed: L) and this will result in two figure-8 Klein bottles that belong to two different regular homotopy classes – for the same reason that the Möbius band shown in Figure 1 cannot be smoothly transformed into its own mirror image. We call the resulting two Klein-bottle classes K8R-O (Fig.7) and K8L-O (Fig.8), respectively. For both classes the two Möbius bands that form the Klein bottle, and which happen to intersect along their center lines, are also shown separately in Figures 7 and 8 (a, b).

From Möbius Bands to Klein-Knottles

(a)

(b)

(c)

Figure 7: Two right-handed Möbius bands $\mathbf{MR}(a,b)$ form a right-handed Klein bottle of type K8R-O (c). (a)

(b)

(c) Figure 8: Two left-handed Möbius bands $ML(a,b)$ form a left-handed figure-8 Klein bottle K8L-O (c).

5. Yet Another Type of a Klein Bottle

But there is even a third way in which the 8-profile tube can be closed into a Klein bottle (Fig.9):

(a) Figure 9: A Klein-bottle based on a “inverted double sock”: (a) without end-caps to show the nesting of the tube profiles; (b) completed Klein bottle of type K8L-J; (c) one of its two ML Möbius bands.

(b)

(c)

Séquin

This new type of Klein bottle is constructed by forming the same kind of J-shaped sweep path as for the classical Klein bottle and fusing the two nested figure-8 profiles by turning one of them inside out. To do this smoothly, we use an asymmetrical figure-8 profile in which one lobe is larger than the other one. As we sweep from one end of the tube to the other one, the larger lobe shrinks and the smaller one grows, so that the end-profiles can be nested as shown in Figures 6c and 9a. Now a nicely rounded, 8-shaped end cap can smoothly close off this Klein bottle (Fig.9b). I have not seen this particular Klein bottle depicted previously. It is rather special, since it does not just have a single self-intersection line like all the other models, but features two triple points. These triple points occur where the self-intersection line of the figure-8 cross section passes through the wall of the other tube near the mouth of the Klein bottle. If we split this Klein bottle into its two constituent Möbius bands, they can be separated cleanly along the closed self-intersection loop passing through both triple points. Each component on its own resembles an “inverted sock” Klein bottle shape with a sharp crease line, which itself exhibits a Möbius twist (Fig.9c).

6. Minimum Energy Klein Bottle

Since it is not clear which one of many possible shapes within the same regular homotopy class should be singled out as the generic representative, we may look for an objective measure for defining the “best” shape. The shape with a minimal amount of total bending energy may be used in this context. This measure, which integrates mean curvature squared over the whole surface, is scale-independent and thus well-suited for this purpose. Lawson [11] has defined the minimum energy forms for many topological shapes. Kusner [10] has conjectured that this Willmore energy [22] is indeed minimized by a Klein bottle described by Lawson. Even though nicely depicted by Polthier [13], this configuration is the most difficult one to understand of all the Klein-bottle shapes discussed so far.

Figure 10: Lawson minimum-energy Klein bottle: (a) Shell with center portions of Möbius bands [13] (top cut off); (b) an FDM model in which the parallel parameter lines have been hand-colored; (c) the set of circular, but twisted (by $360^{\circ }$ ) meridians.

7. Regular Homotopy Classes

Now that we have examined several different construction schemes for making Klein bottles, we can try to group them into the expected number of regular homotopy classes, and then single out a representative for each class that describes this class in the most understandable way.

Regular homotopy transformations cannot perform mirroring operations; they cannot turn a left-twisting Möbius band into a right-twisting one. Thus we can immediately identify three different classes represented by: $K8L-\mathbf{O}=\mathbf{ML}+\mathbf{ML}$ , $K8R-\mathbf{O}=\mathbf{MR}+\mathbf{MR}$ , and by: $\mathbf{KOJ}=\mathbf{ML}+\mathbf{MR}$ ; the fist two are chiral, and $\mathbf{KOJ}$ is amphicheiral. These three classes are structurally different; they do not depend on any markings of the surface. This exposes yet another difference to the world of tori [17], where there are only two structural classes: one formed by the tori TOO, TO8, and T8O, and the other one by T88 by itself.

Now, where do we find the $4^{\text{th}}$ expected type of Klein bottle? It has to be composed of two Möbius bands of the same twistedness; thus structurally it belongs into the same class as KOJ, and it can only be distinguished from the classical KOJ if we place some markings (e.g., a coordinate grid) on its surface.

From Möbius Bands to Klein-Knottles

Could the Lawson Klein bottle discussed above be the sought-after $4^{\text{th}}$ representative? The Lawson surface (Fig.10) has a single circular intersection line, and this line happens to represent the coinciding center lines of two Möbius bands with the same handedness. Indeed the Lawson Klein bottle does come in two different chiral versions [10]. So we can conclude that they must correspond to the two classes: K8L-Lawson = K8L-O and K8R-Lawson = K8R-O. Similar argumentation also rules out the “double-sock” Klein bottle (Fig.9). It has ${\mathbf{C}}_2$ rotational symmetry and is composed of two Möbius bands of the same type. There are again two mirror-image versions: K8R-J = K8L-O and K8R-J = K8R-O.

In the world of tori, drawing an explicit parameter grid allows us to introduce Dehn twists [6] in the meridial direction (M-twist) or in the longitudinal (“equatorial”) direction (E-twist). The introduction of either one of such twists yields a torus that belongs into a different regular homotopy class for marked surfaces. An M-twist of $360^{\circ }$ takes TOO into TO8, and an E-twist of $360^{\circ }$ takes TOO into T8O. Thus we might also want to try this operation on KOJ. However, in the world of Klein bottles the introduction of Dehn twist leads to inconsistencies in the resulting texture mapping [19]; so this approach is futile.

Another operation that changes the regular homotopy class of a torus is adding a collar as illustrated in Figure 15 in [18]; thus we give this approach a try. In Figure 11a a new texture is introduced with longitudinal directionality; this gives the classical KOJ a pattern in which the arrows are coming out of the mouth (Fig.11b). Now we add a self-intersecting collar around the rim of the mouth (Fig.11c). But this collar can easily be eliminated by inflating the blue/purple part of the tube. The net effect is that the roles of the thin and the thick branches have been reversed, and the directional texture arrows now are pointing into the mouth (Fig.11d). So far, this is the best representation I can offer for the $4^{\text{th}}$ class.

Figure 11: (a) A new texture with longitudinal directionality; (b) classical KOJ with this texture applied; (c) KOJ with a collar added at the mouth; (d) KOJ with inflated blue/purple branch.

When faced with the fancy and perhaps confusing contraptions presented in the next two sections, how can we figure out to which regular homotopy class a particular surface belongs? Since these objects have no decorations, we only need to figure out into which structural class they belong. First we make sure that the surface is indeed single-sided and has Euler characteristic $\mathrm{EC}=V-E+F=0$ ; i.e., that it is essentially a single, contorted, self-intersecting “tube”, possibly with some “inverted sock” turn-backs – but with no branching. Then we start drawing a set of three “parallel” lanes and wind this “highway” over the surface until it joins itself again. If each of the three lanes joins itself and exhibits a twist that is an even multiple of $180^{\circ }$ , then we have found a set of meridians, and, based on whether it is twisted or not, we can readily determine whether the surface is in one of the K8R/L classes or in KOJ. If only the center lane joins itself, and the other two lanes form a double-loop over the surface, then we have found one of the two Möbius bands. The portion of the surface not yet covered must then form the other Möbius band. We need to determine the twist of both of them! To find a representative meridian strip, we start a second “3-lane highway” roughly perpendicular to the first one and look for a way to let it close on itself after only a single intersection with each of the two Möbius bands. The twistedness of these three bands uniquely characterizes the structural regular homotopy class of the shape in question.

Sequin

8. Fancy Klein Bottles

Cliff Stoll offers a wide variety of Klein bottles for sale [20]. Some of them are very large (Fig.12a). Even more elaborate glass models have been constructed by Alan Bennet for the Science Museum in South Kensington, UK [4]. These range from an “inverted sock” type with a multiply looped (Fig.12b) or helically twisted (Fig.12d) handle to contraptions that nest several such shapes inside one another (Fig.13a). The latter are topologically no longer simple Klein bottles, since they may be of higher genus or they may form multiple, individual, but interpenetrating surfaces. The examples (Fig.13) are shown here mostly for the enjoyment of the reader, but they may also be used as study objects to train the reader’s skills in determining the regular homotopy class of a particular glass sculpture.

(a)

(b)

(c)

(d)

Figure 12: Fancy Klein Bottles of type KOJ: (a) Cliff Stoll with a very large classical Klein bottle (from [20] used by permission); (b, c, d) Klein bottles by Alan Bennet exhibited at the Science Museum / SSPL in South Kensington, UK (from [4] used by permission). (a)

(b)

(c) Figure 13: Beyond ordinary Klein Bottles: (a) three nested Klein bottles; (b) 2-handle Klein bottle; (c) 3-handle Klein bottle; (d) three merged Klein bottles forming a non-orientable surface of genus 6. All glass models are by A. Bennet; they are located at the Science Museum / SSPL in South Kensington, UK (from [4] used by permission).

(d)

9. Klein Knottles

Inspired by the creative bottle shapes shown above, I searched for other intriguing geometries that topologically are proper Klein bottles. Given my investigation of knotted shapes in the past [15] [16], it was natural for me to look for knotted varieties of Klein bottles; I call these geometries “Klein Knottles.” A simple way to enhance the visual complexity of a Klein bottle is to put more than one - but definitely an odd number - of “inverted sock” turn-backs in series (Fig.14a). All of these surfaces belong in class

From Möbius Bands to Klein-Knottles

KOJ; pairs of subsequent turn-backs can always be created or eliminated by doing an inversion of the segment in between using the Cheritat eversion move [5], which is also illustrated in Figure 5 in [17]. Any chain of such turn-backs can readily be deformed into a knotted geometry, such as the simple trefoil shown in Figure 14b, since tube branches are allowed to slide through one another. Of course, any intermediate state in such a knotting process may also be the final target of an artistic representation of a Klein bottle, since there is no limit on the number of self-intersections allowed (Fig.14c).

(a)

(b)

(c) Figure 14: Examples of Klein Knottles: (a) ring of five KOJ; (b) three KOJ forming a trefoil knot; (c) five KOJ forming an interpenetrating Knottle.

We can also form Knottles with figure-8 profiles. The toroidal Klein bottles of type K8R-O and K8L-O can readily be deformed into any knot desired (Fig.15a) – just as long as we make sure that the cross section makes an odd number of $180^{\circ }$ flips overall, so that we obtain the inside/outside switch-over needed to make a Klein bottle. We can also form chains with multiple figure-8 turn-backs; each one of them acts as an inside/outside switchover. We can even make Klein bottles with an even number of such turn-backs, because we can always insert an extra $180^{\circ }$ of twist in the figure-8 profile. Figure 15b gives an example with six turn-backs – each curved segment also effectively adds $90^{\circ }$ of twist into the chain. Figure 15c shows a minimal configuration of this kind with only two turn-backs and with $90^{\circ }$ of twist in each leg.

(a) Figure 15: Klein Knottles with figure-8 cross sections: (a) K8R-O wound into a trefoil knot; (b) six K8J turn-backs with suitable twist; (c) two K8J turn-backs with $90^{\circ }$ twist each.

(b)

(c)

Séquin

10. Conclusions

Klein bottles are fascinating geometric objects. Most people are familiar with only the “inverted sock” type (KOJ). Many mathematical texts also refer to “the other” Klein bottle – the twisted figure-8 shape – which actually comes in two different chiral versions (K8L-O, K8R-O). Only a few papers or web pages also present Lawson’s minimum-energy Klein bottle, which also comes in two chiral forms that belong to the same two regular homotopy classes. But there is also the possibility of forming turn-backs with a figure-8 profile and I have not seen this one depicted before (Fig.9).

Of course, there are infinitely many possible geometries that constitute Klein bottles, i.e., single-sided surfaces with Euler characteristic zero. Even with marked surfaces, each of these surfaces can be smoothly deformed into one of the four representatives presented in this paper. Finding the most elegant transformation that will actually accomplish this reduction is a much harder task; it deserves more study [19]. In the meantime the reader may simply enjoy the beautiful glass models created by Alan Bennet and Cliff Stoll.

Acknowledgements

I would like to express my thanks to Dan Asimov, Tom Banchoff, Matthias Goerner, Rob Kusner, and John Sullivan for stimulating discussions and various illuminating e-mail exchanges.

This work is supported in part by the National Science Foundation: NSF award CMMI-1029662 (EDI).

References

[1] M. Bill, Endless Ribbon (1953-56). Middelheim Open Air Museum for Sculpture, Antwerp. [2] W. Boy, Über die Curvatura integra und die Topologie geschlossener Flächen. Math. Ann 57, pp.151-184, 1903. [3] Boy Surface at Oberwolfach. - http://www.mfo.de/about-the-institute/history/Boy-Surface [4] A. Bennet, Surface Model. - http://www.sciencemuseum.org.uk/objects/mathematics/1996-544.aspx [5] A. Cheritat, The torus inside out. - http://www.mat.univ-toulouse.fr/~cheritat/lab/e_lano.html [6] Dehn Twist: - http://en.wikipedia.org/wiki/Dehn_twist [7] M. C. Escher, Möbius Band, I (1961). In The Graphic Work of M. C. Escher. MacDonald, London 1967. [8] M. C. Escher, Möbius Band, II (1963). In The Graphic Work of M. C. Escher. MacDonald, London 1967. [9] J. Hass and J. Hughes, Immersions of Surfaces in 3-Manifolds. Topology, Vol.24, No.1, pp 97-112, 1985. [10] R. Kusner, Comparison Surfaces for the Willmore problem. Pacific J. Math. Vol.138, No.2, pp 317-345, 1989. [11] H. B. Lawson, Complete minimal surfaces in $S^3$. Ann. of Math., Vol. 92, pp. 335-374, 1970. [12] D. Lerner and D. Asimov, The Sudanese Mobius Band. SIGGRAPH Electronic Theatre, 1984. [13] K. Polthier, Imaging math: Inside the Klein bottle. - http://plus.maths.org/issue26/features/mathart/index.html#LawsonKlein [14] C. H. Séquin, Twisted Prismatic Klein Bottles. The American Mathematical Monthly, Vol.87, No.4, pp 269-277, April 1980. [15] C. H. Séquin, Tangled Knots, Proceedings of “Art+Math=X” Intnl. Conf., Boulder CO, June 2005, pp 161-165. [16] C. H. Séquin, Knotty Sculptures. Knotting Math and Art, USF, Tampa, November 2007. [17] C. H. Séquin, Tori Story. Bridges Conf. Proc., pp 121-130, Coimbra, Portugal, July 27-31, 2011. [18] C. H. Séquin, Torus Immersions and Transformations. UCB Tech Report (EECS-2011-83). [19] C. H. Séquin, Regular Homotopies of Low-genus Non-orientable Surfaces. UCB Tech Report (EECS-2012-in preparation). [20] C. Stoll, ACME Klein Bottles. Home page. - http://www.kleinbottle.com/index.htm [21] Unknown, Möbius Wedding Band. - http://scn.wikipedia.org/wiki/File:M%C3%B6biusWeddingBand.JPG [22] T. J. Willmore, Note on embedded surfaces. An. Stiint. Univ. “AI. I. Cusd” Iasi Sect.I, a Mat. Vol.11, pp 443-496, (1965).