RankingsPrimerArbiterPhysicistMathematicianContrarianQEClaimsSources

Fedderson-Daedalus: Science Contrarian review

Reviewer role: Science Contrarian (PhD standard; judges by experiment, not consensus; applies equal rigour to the researcher and to the critics).

Source gap (repeated from Step 1). Both PDFs are Google Books scans with no text layer. "Daedalus-Fedderson-109.pdf" is NOT Feddersen's work: it is Proc. Amer. Acad. Arts & Sciences vol. 25 (1890). Feddersen's own 1857-1862 papers are not in the folder; anything said about them rests on [memory]/[web] only. I re-read page images myself (Rich 1917 pp. 142-155 of the reprint, i.e. everything except the PDF-front pages; Trowbridge/Sabine pp. 122-123 and Trowbridge p. 193). I did NOT read Trowbridge-Sabine pp. 109-121 (their data and photographs), so no verdict on claim 2 rests on their measurements. Rich's paper is the only experimental paper with data in the folder.

What the folder actually is

A 1917 dissertation-paper (D. L. Rich, Univ. of Michigan, Phys. Rev. 10, 140-155) plus two 1890 Harvard items. There is no Feddersen text and no cold-fusion-style controversy: the story here is a small, honest, little-followed laboratory observation (Rich) and a historical null result (Trowbridge). I do not manufacture a suppression story; none is documented in these files.

Corrections to Step 1

  1. Rich pages not read. facts.json says Rich was "read in full" yet also that pp. 13-19 were not read. I read them. They contain the film-camera method (48-pane glass condenser, 25 mH coil, film on a 2000 rpm wheel, about 42,000 mm/s, spark-to-spark separation 4.1 mm [read: Rich p.146]), the Cu/Fe, Cu/Cu control tables, the DC make/break polarity determination and the Fig. 9 spot-shape evidence [read: Rich pp.147-151]. These are the core of the evidence and change the assessment (see claims 4-5).
  2. Polarity ambiguity (flagged in Step 1) is partly resolved, and it exposes a contradiction. Table headings use superscripts for the sign of each electrode at the start of the half-cycle (e.g. "Cu+Fe-" = iron negative). For Fe-Bi, Cu-Fe, Cu-Bi and Fe-Zn the more-trains half-cycle is the one where the first-named-in-order electrode (Fe, or Bi) is negative, consistent with Rich's electron-emission reading (Fe 5.21 vs 2.86 trains with Fe negative/positive; Cu+Fe- vs Cu-Fe+ about 6.2 vs 3.16; Cu+Bi- 2.16 vs Cu-Bi+ 1.06; Fe-Zn+ 3.1 vs Fe+Zn- 1.41) [read: Rich pp.148, 152-153]. But for Cu-Zn the table gives Cu+Zn- 1.27 (11 half-cycles with trains; 21 of 32 half-cycles had none) vs Cu-Zn+ 4.77, i.e. more trains when copper is negative, which is the opposite of the claimed series (Zn should emit more easily than Cu) [read: Rich p.154]. Rich's prose ("zinc anode ... discharge could start from zinc much more readily") glosses over this, and his text also uses "anode" for the electrode that the Fe-Zn and Cu-Bi tables label negative. So the "Fe, Bi, Zn, Cu" order is not cleanly supported by his own tables. Step 1's Cu+Zn- 1.27 vs Cu-Zn+ 4.77 numbers are right but the sign problem was missed.
  3. Denominators. Cu+Zn- 1.27 is computed over the 11 half-cycles that produced trains (footnote), not all 32; over all 32 it is 0.44. Cu+Fe- states 6.2 trains/half-cycle but 205/30 = 6.8 (my arithmetic from [read: Rich p.148]); the stated 30 half-cycles or 6.2 contains an error. Other table totals I rechecked (Fe-Bi, Cu-Bi, Fe-Zn, Cu-Fe+, spark totals) add up.
  4. No Bi-Zn data and no "all possible combinations". Rich says all combinations were used [read: p.151]; summaries are shown only for Cu-Fe, Cu-Cu, Fe-Bi, Cu-Bi, Fe-Zn, Cu-Zn. The Bi-above-Zn placement is inferred only indirectly.
  5. Trowbridge Doppler numbers. Step 1: expected broadening "0.1 m of equivalent spectrum width". The page says "a whole tenth meter", the old unit equal to 1 Angstrom (1e-10 m) [read: Trowbridge p.193]. Check: 6 mm gap, half-period of the 3e-7 s circuit = 1.5e-7 s gives v about 4e4 m/s; Doppler shift = lambdav/c = 5000 A * 1.3e-4 = about 0.7 A, matching his "tenth meter" order. The sensitivity ("1/100 of a wavelength" per Step 1) is hard to read on the scan; taken literally 0.01lambda is about 50 A, which could not detect 0.7 A, so the figure must refer to something else. I could not resolve it; treat his stated sensitivity as unverified.
  6. Step 1's "Rich attributes the order to ease of electron emission" is correct. Its dating (dissertation 1 Apr 1915, printed Aug 1917) is correct [read: Rich p.155].

Claim 1: oscillatory discharge of a condenser, T = 2 pi sqrt(LC), seen by rotating mirror

Here T = period, L = inductance, C = capacitance. For example L = 25 mH and C = 0.25 microfarad (Rich's own coil and a mid-range jar) give T = 2 pi sqrt(25e-3 * 0.25e-6) = 1.6e-4 s. - a. Math consistent: yes. Series RLC gives damped oscillation when R < 2 sqrt(L/C); that is a standard ODE result. - b. Consistent with experiment: yes; this is the basis of radio. Feddersen's rotating-mirror photographs (1857-1859) confirmed Thomson's (Kelvin's) 1853 prediction [web: https://en.wikipedia.org/wiki/Berend_Wilhelm_Feddersen and search results summarising it; original papers not read]. Rich's own films show trains of 2-5 sparks per train at 4.1 mm separation, which is also direct evidence [read: Rich p.146]. - c. Tested by Feddersen, Lorenz, Trowbridge and later Hertz and countless others; replicated. - d. Nothing changes. - Verdict: Established.

Claim 2: measured periods deviate from LC; cause "dielectric viscosity" of air (Trowbridge and Sabine 1890)

  • a. The 1890 argument by analogy (Laplace's sound-speed correction, hysteresis) is a heuristic, not a derivation [read: p.122]. No equation for the "viscosity" is given in the conclusions I read.
  • b. Real deviations from the ideal formula exist, but the modern causes are spark-channel resistance that changes in time, stray inductance/capacitance and leaky or lossy dielectrics (glass and shellac show dielectric absorption). Air at low fields has almost no dielectric hysteresis. [memory, medium: not web-checked; I judge the verdict not to depend on this]. The remark that waves in glass in a strong magnetic field are "not notably affected", and the hope of discovering Maxwell's displacement currents that way, is a low-statistics, self-described preliminary note [read: p.122]; displacement current is confirmed by Hertz's waves and all of radio, so no real conflict arises.
  • c. Tested only by the authors in one volume; no replication that I can cite. Their data pp.109-121 were not read by me.
  • d. If "air dielectric viscosity" were a real hysteresis, damping and period of every air capacitor would carry a field-history term; nobody has seen it.
  • Verdict: Historically superseded but instructive (reasonable 1890 reading of period discrepancy; replaced by spark-resistance and circuit-parasitic explanations). The magnetic-field-in-glass experiment: Never properly tested (explicitly preliminary).
  • Researcher rigour: the sound-speed analogy is suggestive but unfalsifiable as stated.

Claim 3: metal atoms are shaken, not carried to and fro, in an oscillating spark (Trowbridge 1890)

  • a. Their Doppler formula lambda1 = lambda * V/(V+delta) is fine. My check gives about 0.7 A predicted shift (see Correction 5).
  • b. Consistent with modern spark spectroscopy: lines come from thermal excitation of vapour, whose bulk motion is not locked to the 60 Hz-to-MHz current reversal. Modern spectra do show Stark and thermal broadening comparable to or larger than the expected shift, which makes this null weak [memory, medium; unchecked].
  • c. Tested once, by Trowbridge, with a Rowland concave grating (20,000 lines per inch) comparing a slit-parallel and slit-perpendicular view. Single photograph pair, no replication known to me. This is a real negative experiment, not dismissal without testing, but with an unverified sensitivity figure.
  • d. Nothing changes.
  • Verdict: Plausible, untested in the strong sense (one low-power null, consistent with accepted physics; no independent repeat).

Claim 4: a spark gap between unlike metals starts more readily on one polarity (partial rectification) (Rich 1917)

This is the one anomaly-type claim in the folder, and it is the best-evidenced part of it. - a. Mathematics is almost absent; reasoning is verbal. The logic (higher sparking potential on one half-cycle leads to fewer trains per half-cycle, fewer sparks per train, odd/even bias) is sound qualitatively [read: pp.144-145]. Statistics: counts of about 28-33 half-cycles per polarity per film, no error bars or significance test; I treat 2:1 ratios as large but not quantified. Odd/even counts are confounded by flux change of the transformer (Rich says so himself, p.145) and Rich admits that "not much reliance can be placed" on them [read: pp.151, 154]. - b. Polarity and electrode-material dependence of sparking is accepted in general (cathode work function, oxide films, field enhancement, rectifying contact). Rectifying behaviour of unlike-metal gaps was also noted by Guthe's coherer work (1901) [read: p.154, Guthe Ann. Phys. 4, 762]. However, the explanation fails a quick check: work function values for polycrystalline metals overlap and are not in Rich's order: Fe about 4.7-4.8, Bi about 4.3, Zn about 3.6-4.9, Cu about 4.5-5.1 eV [web: https://en.wikipedia.org/wiki/Work_function]. Bi lies below Fe, yet Rich finds Fe the easiest. In air, oxide layers and surface condition (Rich cleans/uses "various specimens" but gives no cleaning protocol [read: p.155]) probably dominate. So "ease of electron emission" is an unproven reading of an apparently real asymmetry. - c. Tested by Rich (Univ. of Michigan, supervisor K. E. Guthe) over two years with many film photographs [read: p.148]; the like-metal Cu-Cu control is symmetric (2.03 vs 2.24 trains/half-cycle; 3.97 vs 3.93 sparks/train) [read: p.149], which is a good control against supply asymmetry. Polarity was fixed with a DC make/break ballistic voltmeter kick [read: p.150], which is a thoughtful method, though transferring DC make/break polarity to AC half-cycles by the spot-shape signature (Fig. 9) assumes the signature is polarity-linked. Opposing earlier null results (Righi 1876, De La Rue & Muller 1878, Peace 1892, Carr 1903 per Step 1 and Rich's introduction) were at ordinary gap lengths; Rich used 0.1-4 mm and slightly overvolted gaps. I found no replication [web: none sought beyond the searches above; unknown]. Do not state it was replicated or refuted. - d. Modest: a unlike-metal gap acts as a weak rectifier, as known for point contacts and coherers; a quantitative table by metal would be an addition to a text, not a revolution. - Verdict: Never properly tested (the effect is plausible and reported with a sensible control, but there is no neutral replication on record; it was not rejected, only overlooked). The mechanism "electron emission order" is Plausible, untested and in tension with work-function data.

Claim 5: the order Fe, Bi, Zn, Cu for easy spark initiation, electrode material not negligible

  • a. Internally inconsistent as published: the Cu-Zn table gives more trains with copper negative (4.77 vs 1.27), contradicting Zn above Cu if the other tables' polarity rule is applied; prose uses "anode" in a way that flips table labels for Fe-Zn and Cu-Bi (see Correction 2). The Bi-Zn relationship has no data. The ordering is a total order from five pairings, with no consistency check.
  • b. Not predicted by work functions (claim 4b). Whether it matches later data on sparking potentials in air: unknown; no source checked.
  • c. Tested only by Rich; no replication on record that I know. Rich's data are the single source.
  • d. As in claim 4.
  • Split verdict: "Material matters, not negligible": Never properly tested (see claim 4). "The specific order Fe, Bi, Zn, Cu": Internally inconsistent as stated (Cu-Zn reversal and the anode/negative wording), pending a re-check of the original paper's labelling.

Closing contrarian points

  • Strongest honest case: a two-year, many-film study with a clean like-metal null control and a 2:1 to 6:1 asymmetry in trains per half-cycle across several pairs is better evidence than the 1870s-1900s nulls it opposes. A single flash film on a modern fast camera with fixed gap, cleaned and conditioned electrodes, and a signal generator instead of a leaky transformer would settle it in a day.
  • Against it: tiny samples, no error analysis, a leaky transformer with hidden asymmetry (partially controlled by Cu-Cu), uncontrolled surface oxides, hand-labelled polarity, and the Cu-Zn contradiction. Neither Rich nor the critics ran anything resembling a modern blinded replication.
  • Nothing in these files documents suppression; the claim was simply unrepeated.