RankingsPrimerArbiterPhysicistMathematicianContrarianQESites logClaimsSourcesResearch notes

Tolman-R: Science Contrarian review (Step 2)

Reviewer: Science Contrarian. Inputs read: facts.json, claims.md, deep/ (both deep reads), sites.md "Leads for review", research/ notes, and the clean-typeset Gutenberg edition (research/files/Tolman-1917-Gutenberg-32857.pdf, which I extracted to text and read at the passages cited as [read: Gutenberg, section N]). Tags: [read], [web], [memory] with confidence. Section numbers are Tolman's; PDF page numbers refer to the scan used by Steps 1b.

Source gap carried over: the scan's equations are garbled. Chapters III-IV and X-XI were deep-read from page images in Step 1b. This review adds my own reading of Chapter II (sections 21-22), section 64, section 77, section 103 and section 172 from the Gutenberg edition. Chapters V, VII, IX, XII and XIII still rest on digests plus the passages named here. Where a verdict rests on [memory] alone, it is said.

Context for the reader: this is a mainstream textbook, not an outsider's theory. The contrarian job here is therefore different from the usual: (i) build the best case for the claims, (ii) separate what the experimental record of 1917 actually supported from what the modern record supports, (iii) examine how the book's rejected alternatives (emission theories, the Abraham electron, the ether) were ruled out, and (iv) apply the same rigour to the book itself.

Corrections to Step 1

  1. Claim 6 overstates the book on emission theories. claims.md says "emission (Ritz-type) theories are ruled out by astronomical and Stark-effect evidence". Tolman's text is more careful and in places the opposite. [read: Gutenberg, section 22]: the Stark-effect (canal-ray) evidence is used against the two simpler emission variants (mirror as new source; Stewart's), not against Ritz. Of Ritz's theory he writes that "unfortunately optical experiments for deciding between the Einstein and Ritz relativity theories have never been performed, although such experiments are entirely possible of performance", and "Against the Ritz theory, however, we have of course the general astronomical evidence of Comstock and de Sitter". He calls the list "the whole of the direct experimental evidence against emission theories". So the book rests its case against Ritz on one argument (binary stars) and admits no laboratory test existed. claims.md and facts.json should say this.
  2. "Tolman gives no data, numbers or apparatus for either" is half right. For the binary-star argument he does give the mechanism and the size of the effect: apparent half-periods lengthen and shorten by 2ul/c^2, which he says is comparable to the half-period for most spectroscopic binaries [read: Gutenberg, section 21]. He gives no stellar numbers. For Kaufmann-Bucherer-Hupka he gives no data at all [read: p.56; Gutenberg section 64]. He also cites his own sun-limb experiment (Phys. Rev. 31, 1910), whose apparatus and numbers appear in the Wikisource paper the Investigator read (Lloyd's mirror, expected about 2 fringes, none seen).
  3. The electron experiments are described more strongly than "a triumph". facts.json calls the mass law "confirmed in Tolman's account". The text goes further: the formula has been "closely verified by the Kaufmann-Bucherer experiment" (section 22), and in section 64 the experiments "have in reality shown that the mass of the electron increases with its velocity according to the formula". He also uses the derivation (which has no charge in it) to reject the idea that the electron's charge falls with speed [read: Gutenberg, section 64]. Compare Tolman's own 1910 caveat (Investigator, Wikisource note) that the electron experiments "may not really indicate an increase in the mass". So the author held a more cautious position in 1910 than in 1917, and 1917 does not mention Kaufmann's own 1905-06 conclusion for Abraham's model [web, see claim 2].
  4. The 1 gram = 10^21 ergs "rounding". The deep read treats this as a rounding. In section 77 Tolman gives the exact figure: "1 gram = 9 x 10^20 ergs" (eq. 87) [read: Gutenberg, section 77]. The 10^21 appears only in the Chapter III heuristic passage (p.57 of the scan). Not an error in the book.
  5. Mass-energy evidence in 1917. Tolman states plainly that experimental proof of the mass-energy relation "will be hard to find" and that outside the electron experiments and light momentum "such evidence has not yet been forthcoming" [read: Gutenberg, section 77]. facts.json has this under claim 2 only as an assertion; it should be listed as the book's own admission. He also notes (footnote) that the theory implies "enormous stores of interatomic energy" even at absolute zero.
  6. The Wilson experiment, section 172 (claim 5, digest only). Not mentioned in claims.md. [read: Gutenberg, section 172]: Tolman says his moving-media equations "are in complete accord" with H. A. Wilson's rotating-dielectric experiment, then states that Wilson used a material of permeability practically 1, so only the first-order form Dz = Ez + (V/c)(epsilon-1)Hy was tested, and "it would be a matter of great interest to repeat the Wilson experiment with a dielectric of high permeability so that we could test the complete equation (251)". The original Lorentz theory gave a different equation with a mu factor. So the book itself says its Chapter XII prediction for magnetic media was untested in 1917. This is a "never properly tested" item that Step 1 missed.
  7. Step 1 is right on: the c/2 + c/2 = 4c/5 example (p.70, section 49), the derivation chain in Chapter III, and the equation set (167)-(208) as I re-checked them against the deep read's SymPy results. I did not redo the SymPy checks. I did independently check the lever arithmetic: F1 l1 - F2 l2 = F1 l1 (1 - (1 - V^2/c^2)) = F1 l1 V^2/c^2, correct.
  8. Minor Investigator item. The Investigator names "Ott-Eddington-Moller" for the T0 gamma law. facts.json says only "Ott and Arzelies". Both are cited in the literature; the Investigator's reading of Pouryazdan and Vakili should be preferred for the fuller attribution [read: research/2026-10-04-arxiv-temperature.md].

Web checks done (budget 6; 3 used)

Claim 1: the two postulates, time dilation, length contraction, clock offset, Lorentz transformation

a. Mathematics. Consistent. I follow the deep read's reconstruction: the light-clock gives 1/sqrt(1-V^2/c^2), the longitudinal round trip gives 1/(1-V^2/c^2) and hence the contraction after the dilation is removed, the clock offset l'V/c^2 is stated without derivation but reconstructs correctly, and velocity addition reduces to Newton at low speed and gives c + c = c, c/2 + c/2 = 4c/5. Two honest weaknesses, both disclosed by Tolman: Chapter III is heuristic ("no stress on mathematical elegance or logical exactness", p.45), and the transverse-length equality is used in section 25 before being argued in section 26. Chapter IV's "by trial" derivation admits the solution is "perhaps not necessarily a unique solution" [read: Gutenberg, section 36 area, text near the F(V) = 1/sqrt(1-V^2/c^2) passage]; uniqueness is a later result [memory, high].

b. Established experiments. Today: Michelson-Morley and successors (isotropy), Kennedy-Thorndike (velocity independence), Ives-Stilwell (1938, time dilation by Doppler shift), muon storage-ring lifetimes, atomic-clock aircraft and satellite timing [memory, high for all; no web retag needed as no verdict turns on a date]. In 1917 the support was narrower. Tolman cites a long list of null ether-drift experiments for postulate 1 [read: p.38] but for postulate 2 (source independence) he himself says the direct evidence is the binary-star argument plus his sun-limb and Stark-effect results, and that Einstein-vs-Ritz optical tests "have never been performed" [read: Gutenberg, section 22]. So the postulate-2 evidence in the book was thinner than the confidence of the derivation suggests.

c. Tested? By whom: for 1917 Tolman (sun limb, 1910, null), Stark and Paschen (canal rays, as read by Tolman), de Sitter (1913), Comstock (1910). Replications and later tests: Tomaschek (starlight, 1924) and Miller (sunlight, 1924) found agreement with Einstein [read: research/files Martinez 2004 text layer, summarised in research note]; Alvager 1964 and Brecher 1977 closed the extinction loophole [web above]. Important qualifier from the Investigator: Fox (1962-65, via Martinez 2004) argued that the Ewald-Oseen extinction theorem makes interferometer tests "like those using Tolman's apparatus", the starlight tests and the binary-star argument non-decisive, because light traversing glass, air or interstellar matter is re-emitted at c relative to that matter [read: Martinez 2004 per research note; Fox primary papers not opened, [memory, medium]]. That means Tolman's own 1910 null result and his 1917 reasoning did not, by themselves, rule out source-dependent light speed. The rejection was therefore premature on the evidence of its day but correct in the end: the extinction-free tests of 1964 and 1977 do rule it out to high precision. That is the honest history: "reached the right answer, partly on arguments later shown to be non-decisive".

d. If true, what would change? Nothing now; it is mainstream. If the strong form were false (source-dependent c, Ritz), Maxwell's equations would not hold in all inertial frames and the Lorentz group would not be the kinematic group.

Verdict. Split. (i) The kinematics and Lorentz transformation as physics: Established (modern experiments above). (ii) The 1917 book's own evidential case for postulate 2: Never properly tested in the laboratory at the time (Tolman's words), with the astronomical argument later argued non-decisive (extinction); it was neither dismissed untested nor tested neutrally. (iii) The derivation in Chapter III: valid but heuristic.

Bad data / critics. Rigour cuts both ways. The emission-theory camp's complaint about extinction is physically sound and was applied by the critics of Tolman's own evidence; the answer came from 1964 and 1977, not from 1910-1917. Ritz died in 1909 and nobody tested his theory in the following years (Martinez: "neither spent any effort in extending it") [read: research note]. So the rejection of Ritz was a social and evidential mix: prestige of Einstein's theory and a single astronomical argument, followed decades later by sound tests. No evidence found of blocked publication; Tolman published his own emission-theory papers in Phys. Rev.

Claim 2: relativistic mass and mass-energy; Lagrangian rebuilding

a. Mathematics. The colliding-spheres argument is consistent but leans on several assumptions the deep read lists: momentum defined as mass times velocity with mass allowed to vary, transverse length invariance, and the footnoted limit of negligible transverse speed [read: p.56]. The step from "speed V" to "any speed u" is a generalization. DeltaE = c^2 Delta m follows by integration of u d(mu) [deep read, SymPy confirmed]. The modern route (energy-momentum four-vector) gives the same relations. The "relativistic mass" language is mostly replaced by invariant mass but the physics (p = m0 u / sqrt(1-u^2/c^2)) is the same [memory, high].

b. Experiments. In 1917: the electron beam deflection experiments. As in Correction 3, Kaufmann first (1905-06) favoured Abraham's model; Bucherer (1908), Neumann (1914) appeared to confirm Lorentz-Einstein; a 1938 analysis showed all of these only demonstrated a qualitative increase, not a discrimination [web, above]. So when Tolman writes "closely verified" and "have in reality shown", he is stating more than the data of 1917 could support. Hupka (1910) and the others are named in the book; I have not read Hupka's data and nothing in the folder gives it [unknown]. Modern evidence is far better: precision momentum-energy tests at accelerators, Bertozzi's 1964 electron experiments, and nuclear mass-defect measurements confirming E = mc^2 to high precision [memory, high; not individually retagged because no verdict hinges on a date]. The Chapter III heuristic that the same gram corresponds to 9 x 10^20 ergs is exact from c.

c. Tested / replicated? Yes, repeatedly, from 1908 (with the 1938 caveat on early precision) to modern accelerator and nuclear data. Tolman's own admission in section 77 that the mass-energy relation lacked direct tests beyond electrons and radiation momentum [read] is accurate for 1917.

d. If true / what changes. Already incorporated into mainstream physics; the only change was historical (Abraham's rigid electron and purely electromagnetic mass give way).

Verdict. Split. m(u) and E = mc^2 as physics: Established. The 1917 experimental ground: Historically superseded but instructive (a real but too coarse set of data, later reanalysed and replaced by decisive tests; the early reading favouring Abraham's model is a case where the first experimental result went against the eventual winner). "Relativistic mass" as a concept: Historically superseded but instructive. Chapter IX's Lagrangian with H/sqrt(1-u^2/c^2) invariant: digest-level plus section 120 and section 143 read; consistent and standard [memory, medium-high].

Claim 3: momentum density = energy flow / c^2; stressed bodies; the right-angled lever

a. Mathematics. Internally consistent as far as I can check and as the deep read verified with SymPy: eqs. (168), (173), (204). The lever torque algebra is correct. Tolman's own admissions: his kinetic potential was "not absolutely uniquely determined" and other elasticity theories "may" agree with relativity and known facts [read: p.172]. The deep read's open caveat stands: whether his stress transformation (167) matches the modern tensor boost term by term depends on reference-volume conventions that were not converted (deep read, [memory, medium]). I flag this as unverified rather than wrong.

b. Experiments. The relevant null experiment is Trouton-Noble (1903): a charged capacitor suspended and moving through the supposed ether showed no torque, as this analysis requires [read: p.38 lists Trouton-Noble among ether-drift nulls; deep read, [memory, high] for the equivalence to the lever problem]. Later repeats (Chase 1926, Hayden 1994) [memory, medium; not checked]. These test the absence of rotation, not the transverse momentum of a stressed moving body. Tolman concedes: "At present we have no experiments of sufficient accuracy so that we can investigate the differences between this new theory of elasticity and the older ones, and hence ... have found no experimental contradiction" [read: p.172]. That is an honest statement that the specific content of section 139 had no empirical test when written.

c. Tested? The Trouton-Noble null: yes. The specific stress-induced transverse momentum: no direct test in the sources. The paradox's resolution is still discussed: Shapiro (2012) notes "extensive discussion and indeed controversy" and offers an angular-momentum and simultaneity resolution; the Investigator lists the Am. J. Phys. thread (Nickerson-McAdory 1975, Cavallieri et al. comment) as further reading [read: research note; not opened]. All sources agree the lever does not rotate. The disagreement is over the explanation (energy flow vs simultaneity vs hidden momentum), not the prediction.

d. What would change? Nothing: momentum density = energy flux / c^2 is the symmetry T^{0i} = T^{i0} of the modern stress-energy tensor [memory, high]. Tolman's unsymmetrical t plus symmetric p = t + gu is an older bookkeeping of the same thing.

Verdict. Split. Principle (momentum density = energy flux / c^2): Established (theory plus radiation-momentum evidence). The lever outcome (no rotation, torque balanced by angular momentum of energy flow): Refuted by calculation is the wrong label; better Established by calculation under accepted laws, and consistent with Trouton-Noble (not a standalone verdict on the scale, so I say: Established for the no-rotation prediction, with the explanation still debated). The stress-specific transverse momentum for real materials: Never properly tested (Tolman: no experiment of sufficient accuracy; my worked estimate from the deep read: 10^9 Pa shear at 1000 m/s gives g_y ~ 1.1 x 10^-5 kg m^-2 s^-1, far below anything measured by anyone [deep read, arithmetic]). "Unsymmetrical tensor" as a statement: Not testable as stated (a representation choice).

Claim 4: entropy, pressure, temperature of a moving system; hohlraum momentum

a. Mathematics. Internally consistent given the premise. The key observation, from the deep read and confirmed by my reading of section 103: the factor sqrt(1-u^2/c^2) in T = T0 sqrt(1-u^2/c^2) comes entirely from the chosen kinetic potential H = -E0 sqrt(1-u^2/c^2) (eq. 198) and the definition of T as the force conjugate to S in the Lagrangian. Tolman gives no thermometer argument for the moving-frame value. His kinetic-theory derivation in section 103 defines the thermodynamic scale by a heat-engine cycle and shows T is proportional to the average of m0 u^2 / sqrt(1-u^2/c^2) for a gas at rest [read: Gutenberg, section 103]; that fixes the rest-frame temperature, not its moving-frame transformation. So the Chapter VIII result and the Chapter XI result are not independent evidence for the moving-body law. The hohlraum momentum (4/3) a v0 T0^4 u/(sqrt(1-u^2/c^2) c^2) follows from E, p and the Planck-Einstein momentum equation and agrees with the earlier electromagnetic result Tolman cites [read: p.180]; the 4/3 factor is the same one that appears in the electromagnetic-mass problem and is resolved by von Laue's treatment of the container stresses [deep read, [memory, high]].

b. Experiments. None bears directly on the moving-body temperature law, and neither Tolman nor the Investigator's two 2022-2026 papers cite a laboratory measurement. Radiation is a different matter and it is real data: an observer moving at v relative to black-body radiation sees a direction-dependent temperature T(theta) = T0 sqrt(1-beta^2)/(1 - beta cos theta); this is measured as the cosmic-microwave-background dipole (a few millikelvin for the Sun's roughly 370 km/s motion) [memory, high]. This confirms the radiation transformation and shows that "the temperature of a moving body" is ill-defined for radiation, since the three proposals (T0/gamma, gamma T0, T0) match different angular averages. Pressure invariance and entropy invariance are standard on all sides [deep read, [memory, high]].

c. Tested? No direct test of Tolman's T-law. The competing laws were proposed in theory by Ott (1963), Arzelies (1965), Landsberg (1966-67), van Kampen (1968) [memory, high]. Simulations: Cubero et al. 2007 report numerical evidence for a Juttner-distribution temperature that is observer-independent [web, abstract-level]. Heras and Osorno (2022) say neither the Planck-Einstein nor the Ott-Arzelies law satisfies group closure; Pouryazdan and Vakili (2026) find an effective temperature that rises with speed, against Tolman's form; both are theoretical [read: Investigator note]. Planck-Einstein is described as "supported by many authors" by Heras and Osorno, so Tolman's choice was the majority position until the 1960s and remains one of three live positions. The debate is, in the main, about which quantity one means by temperature.

d. What would change? Nothing in verified physics. Whichever law is adopted is a matter of definition, though a consistent definition has to match the operational thermometer. The covariant inverse-temperature four-vector beta^mu = u^mu/(k T0) is the working convention in relativistic hydrodynamics [memory, medium-high].

Verdict. Split. Entropy and pressure invariance: Established (theory; no dispute). Hohlraum momentum with the 4/3 factor: Established within accepted theory (given container stresses). T = T0 sqrt(1-u^2/c^2) as "the" temperature of a moving body: Not testable as stated (definition-dependent) and Never properly tested; it is a derived consequence of a chosen H, not an independent finding. I do not call it wrong. It was the standard view for half a century and has not been refuted by any experiment.

Claim 5: Maxwell-Lorentz electrodynamics from least action; field transformations; charge invariant; EM energy has mass E0/c^2; ether unnecessary

Digest-level for most of Chapter XII; I read section 172 (Wilson experiment) myself. The rest of this verdict rests on the digest and [memory].

a. Mathematics. Cannot verify the Chapter XII derivations from the sources I read; the digest-level summary is standard. Section 172: the algebra from (248)-(249) to (251) is a valid first-order expansion in V/c as I read it; the reduction for permeability near 1 gives Dz = Ez + (V/c)(epsilon-1)Hy [read: Gutenberg, section 172; OCR of the symbols is poor in places, so read the typeset PDF for exact symbols].

b. Experiments. Charge invariance: supported by the neutrality of atoms and molecules to about 10^-21 of an electron charge despite electrons moving much faster than nuclei [memory, medium-high]. EM energy carries mass E/c^2: light momentum and radiation pressure (Lebedev 1900) [memory, high] and later accelerator and nuclear tests. Moving dielectrics: Wilson's experiment was in accord with Tolman's equations only for permeability near 1, as Tolman himself notes; the high-permeability test he wanted was not done in his text. "The ether is unnecessary": the null ether-drift record [read: p.38]; however a Lorentz-type ether with contraction and dilation is empirically equivalent, so this is an interpretation point, not a test. I flag that the Fresnel-drag and aberration results (Chapter V, digest only) are older than relativity and fit both.

c. Tested? Yes for most field transformations (every EM measurement); not tested at the time for the permeability-dependent term of the Wilson experiment.

d. What would change? Nothing now.

Verdict. Maxwell-Lorentz content and charge invariance: Established. Wilson-experiment term for high-permeability media: Never properly tested in the source (Tolman asks for the test) [later literature not searched; unknown]. "Ether unnecessary": Not testable as stated (empirically equivalent to a Lorentz ether; Occam, not experiment). Digest-only caveat applies to everything except section 172.

Claim 6: the evidence base: electron-mass experiments and the rejection of emission theories

a. Mathematics. The binary-star argument is simple and consistent: apparent half-period shifts of 2ul/c^2 would swamp Kepler's laws in most spectroscopic binaries [read: Gutenberg, section 21]. The reverse derivation of the second postulate from the mass formula (Tolman 1910; mentioned section 22) is sound as logic but circular as an argument against emission theories if the mass formula was itself derived assuming the postulates; Tolman notes it starts from "what has thus become an experimental expression" for mass [read: Gutenberg, section 22].

b. Experiments, and how they were weighed. See Corrections 1-3 and claim 1. Points for the strongest honest case: (1) Tolman's own 1910 emission paper called the emission idea "very simple" and explicit on the first postulate; he ran his own limb experiment and reported a null. (2) In 1912 he said Ritz's theory "does not seem to have received the critical attention which it deserves" and that a definite experimental decision between Einstein and Ritz was "a matter of the highest importance" [read: Investigator's Wikisource note, [web] https://en.wikisource.org/wiki/Some_Emission_Theories_of_Light]. (3) In the 1917 book he drops the open tone and lists Ritz as ruled out by astronomy. (4) Fox (1962-65) and Martinez (2004) argue the early tests, including Tolman's apparatus and de Sitter's argument, were inconclusive through extinction. (5) The tests that actually settle it came in 1964 (Alvager) and 1977 (Brecher) [web]. The weakness of the emission camp: it was never developed into a complete theory after Ritz died in 1909, and the extinction-free evidence is against it by large margins.

Electron experiments: first result (Kaufmann 1905-06) went against the eventual winner; the early confirmations (Bucherer 1908, Neumann 1914) were later judged too imprecise to discriminate [web]. The record shows no suppression or blocked funding; it shows criticism of Kaufmann by Planck, Bestelmeyer and Ritz [web], and later reanalysis.

c. Tested / replicated? Both tested, eventually decisively; the 1917 statement of their status was ahead of the evidence.

d. What would change? Nothing; already absorbed.

Verdict. Electron-mass evidence as of 1917: Historically superseded but instructive. Rejection of emission theories: as stated in 1917 (astronomy only, no optical test): Never properly tested in the laboratory at the time, not dismissed without testing (Tolman and others tried; the tests were later argued non-decisive); as of today: Tested, not supported (Alvager 1964, Brecher 1977).

Investigator's findings kept in the record (whatever the verdict)

  • Gutenberg typeset edition with clean equations (claims 1, 2, 5): used above.
  • Tolman 1910 "Second Postulate of Relativity": his own sun-limb null, his statement that emission "would be very simple", and the caution that the electron experiments "may not really indicate an increase in the mass" (claims 1, 2, 6).
  • Tolman 1912 "Some Emission Theories of Light": Stark/Paschen canal-ray arguments against two emission variants; the call for a sunlight Michelson-Morley test; Ritz "deserves" attention (claim 6).
  • Martinez 2004: early evidence "misconstrued as telling against it more conclusively than actually was the case"; Fox's extinction critique; Carmichael 1913 wrote there was then no undoubted experimental evidence for or against the postulate "despite Tolman's inclinations"; Tomaschek and Miller 1924 (claims 1, 6).
  • Heras-Osorno 2022 and Pouryazdan-Vakili 2026: moving-body temperature is still contested; neither cites a measurement (claim 4).
  • Shapiro 2012 and the Am. J. Phys. thread: lever paradox still debated in explanation, not in outcome (claim 3).
  • Caltech and OAC Tolman papers (12 boxes of lecture notes and manuscripts); NAS memoir (not read): historical record, no gap. Tolman's career continues through Manhattan Project advising; there is no anomalous silence to analyse. The Wilson-experiment lead the Investigator named is addressed above (Correction 6).
  • Still not looked up (unknown): Fox's primary papers, Hupka's data, the Wilson high-permeability repeat, the 1909 Lewis-Tolman paper.

Summary table

Claim Strongest evidence Verdict
1 Kinematics, Lorentz transformation modern: Ives-Stilwell, muons, clocks; 1917: nulls only for postulate 1 Established (modern); book's own case for postulate 2: Never properly tested in the lab in 1917
2 Mass increase, E = mc^2 accelerators, nuclear data; 1917 electron data too coarse Established (physics); 1917 evidence Historically superseded but instructive
3 Momentum = energy flow / c^2; lever Trouton-Noble null; no stress-specific test Established (principle, no-rotation outcome); stress-specific effect Never properly tested; unsymmetrical tensor Not testable as stated
4 Entropy, pressure, T, hohlraum no test of T-law; CMB dipole for radiation S and p Established; hohlraum Established; T law Not testable as stated / Never properly tested
5 Maxwell-Lorentz from least action charge neutrality, radiation pressure; Wilson (mu near 1) Established; high-mu Wilson term Never properly tested; ether Not testable as stated (digest-level)
6 Evidence for mass law; emission theories Alvager 1964, Brecher 1977; Fox's extinction critique 1917 status Never properly tested; today Tested, not supported