What Differs Between the Twins Is Not a Lagging Clock but the Length of a Worldline: time dilation read through the Yamagishi zero-tension, and the null hypothesis of instrumental lag rejected by the muon.
The popular line that a moving clock runs slow invites a natural question in whoever hears it: slow relative to what, and is the clock not simply going wrong? This paper states that question as a falsifiable null hypothesis (Paper 89), writes down which data reject it, and then shows that what invited the question was not the physics but the sentence. THIS IS NOT A PAPER THAT DOUBTS RELATIVITY. Time dilation is real, and this paper marshals the data supporting it. NO NEW PHYSICAL LAW AND NO NEW THEOREM.
Null hypothesis H0: time dilation is an instrumental lag. A moving clock’s mechanism is dulled so that its reading falls behind, and the reading regains its correctness once the clock is at rest. H0 is not empty: it has three testable consequences — (a) different mechanisms would lag differently, (b) what is not a clock would not lag, (c) the difference would vanish at rest. For all three the experiment exists.
Rejection (b): a muon is not a clock. Bailey et al. (1977, Nature 268, 301) measured the lifetime of muons at gamma = 29.33 in a storage ring. Against a rest lifetime of 2.1948 plus or minus 0.0010 microseconds, the lifetimes in flight were 64.419 plus or minus 0.058 and 64.368 plus or minus 0.029 microseconds for positive and negative muons. Special relativity predicts 64.373 microseconds, differing from the measurement by 0.71 parts in a thousand (recomputed here; the paper’s own figure is that the time transformation holds to 0.8 plus or minus 0.7 parts in a thousand). A muon contains no oscillator, no spring, no gears — there is no mechanism in it that could lag. Yet its decay ran slower by exactly gamma. With no instrument to lag, instrumental lag cannot account for the fact. The size of the effect is dramatic too: c times the rest lifetime is 658.6 metres, so without dilation the fraction of muons surviving 15 kilometres of atmosphere would be 1.29 times ten to the minus ten; at gamma = 20 the decay length becomes 13156 metres and the survival fraction 0.32, a factor of 2.5 billion (computed here). That muons are observed at ground level is itself a direct consequence of time dilation.
Rejection (a): the mechanism is irrelevant. Time dilation has been confirmed with instruments sharing no principle of operation — caesium clocks (a hyperfine transition), hydrogen masers (a different atom and transition), optical lattice clocks (an optical transition orders of magnitude away), spectral lines (not a clock at all), and muon decay (no oscillator). If the slowing were a dulling of a mechanism, unlike mechanisms would be dulled by unlike amounts; they are dulled by the same amount. What these share is not a mechanism but a worldline, which means the agreement is on the worldline’s side.
Rejection (c): the difference survives the return to rest. Hafele and Keating (1971) flew caesium clocks twice around the world, eastward and westward, and compared them at rest with the clocks of the U.S. Naval Observatory on their return. Eastward the prediction was minus 40 plus or minus 23 nanoseconds and the observation minus 59 plus or minus 10; westward the prediction was plus 275 plus or minus 21 and the observation plus 273 plus or minus 7. The difference was not recovered. A lag would correct itself once the motion ceased; this did not. What remains is not a wrong reading but a difference in the time that elapsed. There is also a refuting apparatus in continuous operation: for a GPS satellite (orbital radius 2.656 times ten to the seventh metres, speed 3874 metres per second) special relativity gives minus 7.2 microseconds per day and general relativity plus 45.9, a net plus 38.6 (computed here from the orbital elements). Uncorrected, the ranging error grows by 11.6 kilometres per day, and this correction has been applied daily for decades.
So what invited the question? This is the second core of the paper. The word slow tacitly posits something to be slow relative to, and no universal clock exists. The most that can be posited is another observer’s coordinate time, and from that observer’s side the first clock is slow in return. So the question which one is really slower has no answer, and what makes people ask an unanswerable question is the form of the sentence. The one who asked did not err — the sentence invited an ill-formed question. This is the category error diagnosed in Paper 118: the coordinate dt is a label pinned to the manifold while proper time dtau is a functional of a curve, and saying the clock runs slow treats dtau as though it were dt, speaking of a quantity that may differ from curve to curve as if it were a scale everyone shares. Popular expositions keep reproducing the confusion not because the physics is unsettled but because they keep using a sentence with the categories crossed.
Put correctly, the question does not arise. Since Minkowski (1908; Paper 108) the correct statement has been that dtau squared equals dt squared minus the spatial part over c squared, and that proper time is the integral of dtau along the worldline. Proper time is the length of a worldline. The two twins joined the same two events by two different paths; the paths differ in length, and that is all. Neither clock went wrong — each measured the length of its own worldline correctly and in full. What differs is what was measured, not the instrument that measured it. For ten years of coordinate time the proper time is 8.66 years at v/c = 0.5, 6.00 at 0.8, 4.36 at 0.9, 1.41 at 0.99, and 0.10 at 0.99995 (computed here). The same restatement answers the objection that motion is relative and the situation should be symmetric: it is not symmetric, because only one twin changes inertial frame, and this is not a matter of description — give each twin an accelerometer and the readings differ. In geometric terms the traveller’s worldline is bent and the stay-at-home’s is straight, and in Minkowski geometry the straight path maximises proper time, the opposite of Euclidean geometry where the straight path is shortest. The bent path is shorter; that is the whole of it.
An honest brake. The processes confirmed to follow time dilation are many: the caesium hyperfine transition, the hydrogen maser, optical lattice transitions, atomic spectral lines, and muon decay. None is known to disobey. But no human has been carried at relativistic speed; the Hafele-Keating aircraft were subsonic and the differences there were of order a tenth of a microsecond. The statement that a person would not age is an induction over processes. The induction is very strong — a living body is made of chemical reactions, chemical reactions are electromagnetic, and that electromagnetic processes follow proper time has been confirmed repeatedly since Ives-Stilwell; to posit a physical process that does not follow proper time is to discard known physics. Even so, an induction is not an observation. What is established is that every process tested follows proper time; that a person would not age is one step beyond that. Not writing that step with the force of an observation is also the honest reply to whoever raised the question. The phrase the Urashima effect bundles two claims — that the proper time is shorter (established, treated here) and that it felt shorter to the traveller (not treated here, outside physics). The latter is the quantity Paper 118 explicitly refused to enter into the physics census, and it is quarantined here as Layer 3.
IMPORTANT (scope): an organisation of established standard facts and a diagnosis of a popular phrasing (Layer 2). Every experiment used is established (Bailey et al. 1977, Hafele-Keating 1971, Ives-Stilwell 1938, Vessot et al. 1980, GPS). NO NEW THEOREM AND NO PREDICTION. What is not claimed: any new physical law (there is none) / any doubt about relativity (this paper is on the side of time dilation) / an observation that a person would not age (that is an induction over processes) / anything about felt time (outside the physics census, quarantined as Layer 3) / that acceleration is the cause (acceleration breaks the symmetry, but what gives the difference in proper time is the shape of the worldline). What remains: when a question keeps arising, suspect the sentence before the physics.
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). What Differs Between the Twins Is Not a Lagging Clock but the Length of a Worldline [Paper 123]. Zenodo.
DOI: 10.5281/zenodo.22055231
ORCID: 0009-0006-8587-3450