The Wigner angle is an area – velocities do not add plainly because the place they live in is curved.
To say that in relativity velocities do not simply add makes the matter sound smaller than it is. This paper shows in numbers that the reason they do not add is the curvature of velocity space. NO NEW PHYSICAL LAW IS CLAIMED – nothing here contradicts the Lorentz transformation. The hyperbolic character of velocity space, rapidity, the Thomas-Wigner rotation and gyrogroups are all standard.
First, along one line the rapidities add exactly. Composing 0.5 with 0.5 gives 0.8, 0.8 with 0.6 gives 0.9459459459459459, and 0.99 with 0.99 gives 0.9999494975001263: Einstein’s rule and the hyperbolic tangent of the sum of rapidities agree to order ten to the minus forty-one (computed here). The rapidity is the hyperbolic distance from rest, so lengths along a geodesic are being added, and of course they add.
Second, off that line a rotation is left over. Composing a rapidity of 1.0 along x with 0.9 at an angle alpha gives not a pure boost but a boost times a rotation. The composed rapidity is 1.849219714502761 at thirty degrees, 1.431199273508851 at ninety, and 0.5695508468232462 at one hundred and fifty.
Third, and this is the core. That rotation angle is the area of the rapidity triangle. Take three points on the unit hyperboloid – rest, after the first boost, and after both – and the three sides have the respective rapidities as lengths. On a surface of curvature minus one, Gauss-Bonnet makes the area equal to pi minus the sum of the three angles. Sweeping alpha from fifteen to one hundred and seventy degrees, the Wigner angle and the angle defect agree: at fifteen degrees both are 0.0848772541736125232, at ninety both are 0.38510252707757708, and the worst difference across the table is 2.612 times ten to the minus forty (computed here). The leftover rotation is not an effect bolted on afterwards. It is precisely the angle defect of a triangle in a curved space.
The same Gauss-Bonnet is working a second time inside the corpus, the first being the earth’s surface (Paper 27). There positive curvature gave an excess; here negative curvature gives a defect. The theorem is the same and only the sign differs. So the Thomas-Wigner rotation is not a strange phenomenon peculiar to relativity: it is the holonomy of a closed circuit in a curved space, and as a theorem it sits where Foucault’s pendulum turning on a sphere sits.
Fourth, the separator is non-collinearity. At alpha equal to zero the triangle collapses to a segment and both area and rotation are exactly zero. At a thousandth of a degree the value is 5.69524130474 times ten to the minus six, at one degree 0.005695078292, and at ten degrees 0.0567892433037 (computed here). Move the angle and the curvature begins to bite. The first section could say they add not because velocities are special there, but because a one-dimensional submanifold has no area.
Fifth, the brake. They add differently is not enough. The composition is neither commutative nor associative. With u equal to (0.8, 0) and v equal to (0, 0.6), composing u then v gives (0.8, 0.36) while v then u gives (0.64, 0.6); the speeds agree at 0.8772684879784524 and only the directions differ, the discrepancy being exactly the Wigner rotation. Taking w as (0.4, 0.4), the two bracketings differ by 0.0600488 (computed here). Under this operation the set of velocities is not even a group – the accurate statement is that the failures appeared because something that is not a group was being written as one.
Sixth, the one half of the low-speed limit is not an adjustment. At ninety degrees with both speeds equal to v, the ratio of the Wigner angle to half v squared is 1.00502935547 at v of 0.1, 1.00005000292 at 0.01, 1.0000005 at 0.001 and 1.000000005 at 0.0001, converging to one (computed here). The one half carried by the leading term is the same one half that appears in the spin-orbit coupling. It is not a fudge factor but the leading term of an area, for the same reason that the area of a triangle begins as half the product of two sides.
The census. Galilean velocity lives in Euclidean space, curvature zero, and forms an abelian group. Rapidity is geodesic length in hyperbolic space, curvature minus one, and is additive only along a line. Relativistic velocity consists of points of that same space, curvature minus one, and is neither commutative nor associative. All three have been called adding velocities. The separator is the curvature of the space the set lives on.
IMPORTANT (scope): a census of established standard material (Layer 2). What is not claimed: any new physical law (there is none) / anything contradicting the Lorentz transformation (nothing does). What is added: an independent numerical verification that the Wigner angle equals the hyperbolic area, naming the separator as non-collinearity, writing addresses for three additions, and noting that the same Gauss-Bonnet works a second time inside the corpus. What remains: velocities do not add plainly, and not because the rule is odd. It is because the place where velocities live is curved. And in a curved place, going once around a triangle turns you – on the ground and in velocity space alike, one theorem is saying one thing.
On the making of this work: The ideas and content of this work stem from the author’s own considerations. Assistance from an AI (a large language model) was used for structuring, English translation, and checking the algebra. Any remaining errors or misinterpretations are solely the author’s. Feedback and corrections are sincerely appreciated.
Reference
Yamagishi, Y. (2026). The Wigner Angle Is an Area [Paper 33]. Zenodo.
DOI: 10.5281/zenodo.22090872
ORCID: 0009-0006-8587-3450