Coordinates label a patch rather than define spacetime

A smooth four-dimensional manifold has neighborhoods mapped one-to-one onto open subsets of real four-space, with continuous inverse maps. On overlapping neighborhoods, the coordinate transition maps and their inverses are smooth. Standard spacetime manifolds also satisfy Hausdorff and countability conditions that exclude pathological identifications. A chart supplies labels and a way to use multivariable calculus; by itself it supplies neither a distance nor a preferred time direction. Those structures require a metric and, for a consistent future designation, a time orientation.

An atlas is a compatible collection of charts covering the manifold. A sphere cannot be described by one ordinary latitude–longitude chart without failures at poles or a seam. A circle similarly has no globally one-to-one real angle coordinate: the values differing by a full turn represent the same point. Embedding a circle or sphere into a larger Euclidean space is a convenient visualization, not part of the definition of an abstract manifold. In particular, bending in an embedding and intrinsic spacetime curvature are distinct ideas.

Consider the circle using an angle θ in the open interval (−π,π). The coordinate u=tan(θ/2) gives one real label for every point except the point at θ=π modulo a full turn. Its inverse is θ=2 arctan u. The divergence as that omitted point is approached does not create a missing physical piece of the circle; a second chart can cover it. The open interval matters: including both endpoints as different labels would destroy the one-to-one property.

For a circle of radius a, its line element in u is regular at every finite u. The radius belongs to a chosen metric, while the chart relation itself is dimensionless. This example shows why a vanishing or diverging coordinate component near a patch edge cannot by itself diagnose intrinsic curvature. In spacetime, polar origins, symmetry axes, and some horizon coordinates exhibit analogous failures. A true metric degeneracy within a claimed valid chart is different and cannot be repaired by a smooth invertible coordinate transformation there.

Coordinate domains also determine which events a calculation describes. Rindler coordinates cover one wedge of flat Minkowski spacetime rather than its entirety. A chart ending at a horizon can therefore conceal an extension even though curvature is zero. Conversely finite coordinate labels do not guarantee geodesic completeness or regular curvature. One should inspect the map, the metric’s nondegeneracy on its domain, and invariant or extendibility information separately. Local chart validity is not a proof of a globally adequate coordinate description.

The vector lab makes the angle seam visible. Reset and set both controls to angle 180 degrees and magnitude one, predicting components (−1,0). Reset and set angle −180 degrees and magnitude one, predicting the same components. The two endpoint labels describe one geometric direction. This finite-dimensional picture contains no manifold atlas, transition Jacobian, or spacetime metric. It illustrates the labeling issue only; reaching the end of an angle slider should not be interpreted as reaching a physical boundary.