Poincaré's Ether: A. Why did Poincaré Retain the Ether?
| Scientific Paper | |
|---|---|
| Title | Poincaré's Ether: A. Why did Poincaré Retain the Ether? |
| Read in full | Link to paper |
| Author(s) | Galina Granek |
| Keywords | stellar aberration, action-at-a-distance, absolute rotation, absolute space |
| Published | 2001 |
| Journal | Apeiron |
| Volume | 8 |
| Number | 1 |
| No. of pages | 13 |
Read the full paper here
Abstract
This paper is divided into three parts, in which I suggest five answers to the question: Why did Poincaré retain the ether? These answers are based on Poincaré's own reasoning: the ether was required for the explanation of stellar aberration, to remove action-at-a-distance, to remove absolute rotation and absolute space from physics, to save broken theories and to save Poincaré's conventionalism. Poincaré's first reason can be seen as related to rectilinear and uniform motions. In 1905 Einstein managed to explain aberration without resource to ether. Special relativity crowned the final oblivion of the ether. Poincaré's four other reasons are centered on the solution to the following old problem: the principle of relativity is not valid for rotations and we thus can claim for absolute rotation. Poincaré struggled with this problem and could not solve it without resource to the ether. In General Relativity, Einstein could not solve it without returning to some kind of ether, either. I first discuss Poincaré's reasoning and in a future paper "Why did Einstein come back to the ether?" I discuss Einstein's solution to the problem of absolute rotation and his return to a revised form of Poincaré's ether.
Overview
This is a paper in the history and philosophy of physics rather than in physics proper. Granek, writing from the Department of Philosophy at Haifa University and drawing on her 1998 Hebrew University dissertation on Poincaré's contribution to relativity, sets out to answer a question that the standard textbook narrative usually disposes of in a sentence. Henri Poincaré had, between 1888 and 1900, argued in print that the ether was "a mere invention, a convenient hypothesis that could be omitted" — and yet he never omitted it. Why not?
Granek's answer is that Poincaré did not cling to the ether out of conservatism or failure of nerve. He retained it because the specific problems that had made physicists invent it in the first place were, for him, still unsolved. She identifies five such problems, each of which Poincaré himself names somewhere in his lectures and popular writings: the explanation of stellar aberration, the elimination of instantaneous action-at-a-distance, the elimination of absolute rotation and absolute space, the rescue of "broken theories", and the defence of his own conventionalist philosophy. The first of these concerns uniform rectilinear motion, and Granek concedes it was settled by Einstein in 1905. The remaining four all converge on a single stubborn difficulty: the principle of relativity does not hold for rotations, so absolute rotation appears to survive. That, she argues, is where the ether kept its grip — and it is the same difficulty that, on her reading, later drove Einstein back to "some kind of ether" in general relativity, the subject of her companion paper.
Poincaré's five reasons
To explain stellar aberration
Aberration is the small annual displacement of a star's apparent position caused by the telescope moving while light traverses it, of order the constant v/c where v is the earth's orbital velocity. Fresnel's 1818 explanation required a stationary ether at absolute rest, freely penetrating the pores of matter as Young had proposed — an "ether wind" passing through matter "like wind passing through a grove of trees". Granek's point is that this account is not neutral: if the ether is stationary, aberration theory should in principle disclose the earth's absolute motion to first order in v/c. If instead the ether is dragged along with the earth, "mobile ether is experimentally equivalent to not having ether at all".
Since Michelson (1881) and Michelson and Morley (1887) had returned null results at second order, Poincaré reasoned that what was needed was a theory of aberration that dispensed with the stationary ether and thereby explained the null results in advance. In his 1904 Saint Louis lecture he said as much: theorists "would make a useful work in constructing a theory of aberration which would explain this in advance". He never built it. His nearest attempt — comparing the apparent oscillation amplitudes of two opposite stars — was resolved by appeal to the Lorentz contraction hypothesis making the two amplitudes measure equal, which still leans on the ether. Einstein derived the aberration formula from the relativity principle alone in the 1905 electrodynamics paper, and Granek makes the pointed claim that only once he possessed that formula was he really in a position to abandon the stationary ether. Poincaré's compromise instead was to allow absolute rest while denying absolute space.
To remove instantaneous action-at-a-distance
Poincaré's second reason is stated in his 1900 Paris Congress lecture. Light from a distant star takes years to arrive; during that interval it "is no longer upon the star nor is it upon the earth; it must be sustained somewhere". Granek stresses the more technical form of the same argument: ordinary mechanics works with differential equations, in which the state of a system depends only on the immediately preceding state. Without a medium, the state of the universe would depend on much older states and physics would be committed to finite-difference equations — jumps from one place to another. "It is to escape this exemption of the general mechanical laws that we have invented the ether."
This meshed with the electromagnetic world-picture dominant between 1900 and 1905, under which all forces were ultimately electromagnetic. Lorentz proposed in 1900 that gravitational attraction propagates at the velocity of light; Poincaré concluded in 1905 that gravitational attraction is carried by a wave travelling at c. The velocity of light thereby became common to gravitation and to electromagnetism, and instantaneous action at a distance was excluded.
To save broken theories
Here Granek reconstructs Poincaré's 1900 parable of the cloudy planet. Its inhabitants, permanently unable to see the sky, first suppose their planet immobile. They then have to treat centrifugal forces as real interactions between bodies — but these forces do not fall off with distance, they grow without limit. To rescue the hypothesis they "imagine some kind of a very subtle environment, analogous to our ether" exerting a repulsive action. Further anomalies accumulate: cyclones always turn the same way, breaking the symmetry that an immobile planet would demand. Entities multiply until "a long-awaited Copernicus would arrive and sweep them all away with a single blow".
Read allegorically, the nineteenth-century ether is exactly such a rescue device for the electrodynamics and optics of moving bodies, and the awaited Copernicus would be Einstein. Granek's historical finding is that Poincaré did not accept this reading: as late as 1912 he thought Copernicus had not yet arrived, did not cast himself in the role, and — importantly — did not cast Einstein in it either. He did not believe Einstein had eliminated the ether. So on his own accounting the ether was still doing indispensable rescue work for three theories: stellar aberration, the mechanics of uniform rotations, and action-at-a-distance interactions.
To save conventionalism
Poincaré's conventionalism holds that "the earth turns round" and "it is more convenient to suppose the earth turns round" have one and the same meaning. If only relative motions exist, neither of the rival conventions can be truer than the other — only more convenient — and the convention of a stationary earth is ruinously inconvenient, since the stars would have to sweep out in 24 hours a circumference light takes centuries to cross.
Granek argues that this does not in fact eliminate absolute motion, and she extends Poincaré's own parable to show why. Suppose the cloud-dwellers develop a dynamics equivalent to Newton's, build a spaceship, rise above the grey sky and film the cloud layer from outside. The two conventions were empirically equivalent only for beings confined beneath the clouds. Quoting Ben-Menahem, the equivalence "is internal, and unlikely to persist if an external point of view becomes possible". A being outside could simply decide whether the planet rotates with respect to absolute space. To block that conclusion one must still posit a medium analogous to the nineteenth-century ether, so that daily rotation can be referred to the ether rather than to absolute space. The ether, on this reading, is the price of conventionalism.
Assessment
The paper's distinctive contribution is methodological: it takes Poincaré's stated reasons seriously as reasons, rather than as symptoms. The usual story treats the retention of the ether as inertia in a great mind. Granek instead shows that each of Poincaré's five motives is a genuine explanatory debt, and that four of the five are debts about rotation rather than about uniform translation — a point that the standard "Poincaré nearly got to special relativity" framing obscures entirely. The observation that Poincaré separated absolute rest from absolute space, retaining the first while denying the second, is a real piece of exegetical work, as is the claim that possession of the aberration formula was the precondition, not the consequence, of Einstein's abandonment of the stationary ether. The extension of the cloudy-planet parable is a neat philosophical argument in its own right: internal empirical equivalence is a fact about observers' access, not about the world, and it dissolves as soon as access improves.
The difficulties are of two kinds. First, several of the paper's load-bearing claims are asserted rather than argued from the sources. That Einstein was "very likely influenced" by reading Science and Hypothesis before 1905 is offered without documentary support here; that Einstein could only abandon the stationary ether "after he had possessed the aberration formula" is a claim about order of discovery that the 1905 paper's own structure does not settle, since the aberration section comes late in a paper whose kinematic foundations come first. Second, the promised third leg of the argument is missing. The abstract announces five answers and says the paper is in three parts, but the running text presents four headed reasons; the elimination of absolute rotation and absolute space is folded into the discussions of broken theories and conventionalism rather than treated on its own. A reader looking for the direct engagement with rotation — the pivot of the whole thesis — is referred forward to the companion paper "Why did Einstein come back to the ether?".
Third, and most consequential for readers of this wiki, the paper's central historical claim about general relativity is stated but not defended. That Einstein "could not solve it without returning to some kind of ether" refers to his 1920 Leiden address, in which the metric field is described as an ether devoid of mechanical properties and states of motion. Whether that verbal concession is the same thing Poincaré needed — a medium with respect to which rotation is defined — is precisely the question at issue, and it is deferred. It also sits uneasily with the standard general-relativistic treatment of rotation, where the Sagnac effect and the Foucault pendulum are handled by local inertial frames determined by the global mass distribution, without any medium at rest. Granek does not evaluate whether Mach's principle discharges the debt Poincaré thought only an ether could discharge. Within its declared limits, though — reconstructing what Poincaré himself thought the ether was for — the paper is careful, well sourced to the French originals, and persuasive.