Answer: Klein Bottle Is Non-Orientable and Doesn't Embed in R³

Key Idea / Intuition

The Klein bottle contains a Möbius band as a subspace — just look at the middle horizontal strip. A Möbius band is the prototypical non-orientable surface, and a space containing a non-orientable subspace is itself non-orientable. The punchline is purely topological: any compact surface embedded in $\mathbb{R}^3$ separates it into two regions (by Alexander duality / Jordan–Brouwer), and such a surface must be orientable (the "outward normal" gives a consistent orientation). So a non-orientable closed surface simply cannot live in $\mathbb{R}^3$.

---

Formal Proof / Solution

Step 1: Find a Möbius Band inside $K$

Consider the horizontal strip $[0,1] \times [1/4, 3/4]$ inside the square. The identifications of $K$ restricted to this strip are:

• Top edge of strip: $(x, 3/4)$ glued to $(x, 3/4)$ — no identification needed.
• Left/right edges: $(0, y) \sim (1, 1-y)$, which for $y \in [1/4, 3/4]$ maps $1-y \in [1/4, 3/4]$, so the sides are glued with a flip.

This is precisely the construction of a Möbius band. So $K$ contains a Möbius band $M \subset K$.

Step 2: $K$ Is Non-Orientable

Definition: A surface is orientable if it admits an atlas $\{(U_\alpha, \phi_\alpha)\}$ where all transition maps $\phi_\beta \circ \phi_\alpha^{-1}$ have positive Jacobian determinant.

Since $M \subset K$ is a Möbius band, $M$ is non-orientable: any attempt to consistently choose an orientation around the central circle of $M$ leads to a contradiction (going once around reverses orientation).

More directly for $K$ itself: consider a loop $\gamma$ in $K$ corresponding to the horizontal path $t \mapsto (t, 1/2)$ for $t \in [0,1]$. At $t=0$ and $t=1$, the identification is $(0, 1/2) \sim (1, 1/2)$, so $\gamma$ is a closed loop.

Now transport a local orientation (a choice of basis for the tangent plane) continuously along $\gamma$. The side identification $(0,y) \sim (1, 1-y)$ reverses the $y$-direction. After traversing $\gamma$ once, the transported orientation is reversed. Hence $K$ is non-orientable.

Step 3: Non-Orientable Closed Surfaces Cannot Embed in $\mathbb{R}^3$

Theorem (Classical): If $S \subset \mathbb{R}^3$ is a compact surface without boundary embedded in $\mathbb{R}^3$, then $S$ is orientable.

Proof sketch: By the Jordan–Brouwer Separation Theorem, a compact connected surface $S$ embedded in $\mathbb{R}^3$ separates $\mathbb{R}^3$ into (at least) two connected components, one of which is bounded. At each point $p \in S$, one can choose the unit normal $\mathbf{n}(p)$ pointing into the bounded component. This choice varies continuously (using the local flatness of the embedding), yielding a globally consistent normal field — which is exactly a global orientation of $S$.

Since $K$ is non-orientable, it admits no such normal field, and hence:

$K \text{ cannot be embedded as a closed surface in } \mathbb{R}^3. \qquad \blacksquare$

Summary

| Step | Content | |------|---------| | 1 | $K$ contains a Möbius band | | 2 | A loop in $K$ reverses orientation → $K$ non-orientable | | 3 | Embedded closed surfaces in $\mathbb{R}^3$ have a global normal → must be orientable | | Conclusion | $K \not\hookrightarrow \mathbb{R}^3$ |

Remark: The Klein bottle can be immersed in $\mathbb{R}^3$ (with self-intersections), but not embedded. It embeds cleanly in $\mathbb{R}^4$.