SchusterAndHemsalech · Source catalogue

A six-page abstract of a measurement of how fast metal vapour spreads across an electric spark gap. Sparks from six Leyden jars (0.033 microfarad total, charged by a Wimshurst machine with twelve 62 cm plates, 13 inch sparks) between metal electrodes about 1 cm apart were focused on the slit of a spectroscope (Steinheil 60 degree prism) and recorded on photographic film held flat on a spinning disc (discs 33 and 22.2 cm diameter, 10.8 cm annulus exposed, typically 120 turns/s, about 100 m/s film speed, up to 170 turns/s possible; made by the Cambridge Scientific Instrument Company). The method follows Dixon's work on explosive waves; the authors' own rotating-mirror attempts over several years failed. Because the film moves, spectral lines of the metal appear inclined and curved (the slope measures how the emitting particles move away from the pole), while air lines stay straight but slightly widened. Zinc gave the sharpest results (lines at 4924.8 and the blue triplet led by 4810.7 Angstrom). Table I gives mean velocity V1 (pole to a point 2 mm away) for zinc against number of jars (2, 4, 6) and gap (0.51, 1.03, 1.54 cm): values range from 400 to 1526 m/s, with 6-jar values near 416-545 m/s for gaps of 0.51 and 1.03 cm and two values marked doubtful (435?, 492?) at 1.54 cm. For the shortest gap velocity falls as capacity rises (814, 556, 416 for line 4925), which the authors call 'very curious' and contrary to expectation (more jars, higher temperature). At 1 cm no marked capacity effect; at longer gaps sparks are irregular and no firm conclusion is drawn. The 4925 doublet line gives a smaller apparent velocity than the 4811 triplet line (the p.3 prose says 'uniformly higher velocity deduced from the double line 4925', which contradicts Table I and the following sentence; the table and next sentence are taken as correct); the authors hesitate to attribute this to different molecular mass and suggest it may be a measuring artefact of the unresolved doublet, while noting the photographs hint that the doublet molecule may truly move faster, and defer the decision to higher dispersion. Other metals: aluminium over three times the zinc velocity (not claimed accurate); magnesium scattered too widely to measure; zinc and cadmium nearly identical; bismuth (taken with 2 jars instead of 6), despite high atomic weight, had some lines giving 1420 m/s, others like zinc and cadmium, and one line (3793) slower; mercury unsatisfactory (tried via amalgamated zinc/cadmium poles). They conclude there is no simple law linking velocity to atomic weight. Interpretation: Feddersen (1862) held that torn-off metal particles take no part in the discharge. The authors propose instead that the first discharge goes through air, the heat volatilises metal, and the later oscillations of the Leyden-jar discharge pass through metal vapour rather than air. Supporting experiment: inserting a coil of wire (self-induction) in the jar circuit makes the air lines nearly vanish, leaving only metal lines, explained as a longer discharge giving the vapour time to diffuse into the gap; Hemsalech is studying this further. They liken the vapour spreading to gas flowing into a partial vacuum rather than thermal diffusion, so the measured speed approximates the speed of sound in the metal vapour, giving V = 80 sqrt(T/rho) (V in m/s, T absolute temperature, rho vapour density relative to hydrogen; ratio of specific heats differences neglected). Worked case in the paper: cadmium V = 560 m/s, rho = 56, so T = 2700 K (reconstruction check: 80*sqrt(2700/56) = 555 m/s, consistent). From this they conclude the cadmium molecule in the spark is not much lighter than the mass measured near the boiling point (that is, still essentially a normal cadmium vapour molecule, not split into smaller pieces). Final section: slit focused directly on the film without the prism shows a straight image of the slit (taken as the initial air discharge) followed by curved bands extending from both poles. Approximate timings for a 1 cm gap: air luminous about 5e-7 s; cadmium vapour reaches the gap centre in about 6e-6 s; oscillation period with six jars and minimal self-induction about 2e-6 s; some luminosity at the centre persists about 1.5e-5 s. The authors say the oscillation period was too short for best results (later oscillations may disturb the motion of vapour not yet far from the pole) and hope to use higher capacities. Further observation on p.6: the metallic vapours remain luminous in the centre of the spark for a longer period than near the poles. The vapours 'then begin to diffuse' after the air glow, so the 6e-6 s may be counted from when diffusion begins rather than from the start of the discharge. Hull (1906) and Royds (1908) later challenged reading the slopes as bulk vapour velocity [read via Investigator notes in research/, not from this document].

Key claims

• Metal vapour torn from spark electrodes reaches the centre (5 mm) of a 1 cm gap in about 6e-6 s (cadmium), at mean speeds of roughly 400-1500 m/s depending on metal and line.
• Metal spectral lines photographed on a moving film are inclined or curved, and the inclination measures the velocity of the emitting particles away from the electrode; air lines stay straight.
• The first discharge of a Leyden jar passes through air; subsequent oscillations pass through the volatilised metal vapour, contrary to Feddersen's view that vapour particles take no part in the discharge.
• Inserting self-induction (a coil) in the spark circuit suppresses the air lines and leaves the metal lines, because the lengthened discharge allows vapour to fill the gap.
• The measured vapour speed approximates the speed of sound in the metal vapour, V = 80 sqrt(T/rho), giving about 2700 K for cadmium at 560 m/s.
• Spark-gap molecules of cadmium are not much lighter than normal cadmium vapour molecules (no extensive dissociation).
• Vapour velocity shows no simple relation to atomic weight (bismuth, a heavy metal, is fast on some lines; aluminium is over three times as fast as zinc; zinc and cadmium are equal).
• Different spectral lines of the same metal (zinc 4925 vs 4811) may give different velocities, possibly indicating different emitting masses, but the authors say this is unresolved and may be a measurement artefact.
• At small gaps the zinc vapour velocity decreases as capacity is increased, contrary to the expected temperature effect (the authors do not call this established).
• Timeline of a 1 cm, six-jar spark: air luminous about 5e-7 s, oscillation period about 2e-6 s, vapour luminous at centre up to about 1.5e-5 s.
• The metallic vapours remain luminous in the centre of the spark for a longer period than near the poles (p.6).
• Bismuth was measured with 2 jars rather than 6, with line 3793 slower than zinc and cadmium; mercury was tried via amalgamated zinc and cadmium poles; magnesium was too scattered to measure.

Relevance: This is the only document in the project and so the whole basis for it. It is an early (1899) application of time-resolved spectroscopy to the spark: a precursor of spark-discharge plasma diagnostics and of the Schuster-Hemsalech line of work (Hemsalech continued it with self-induction and the 'flame/spark' spectra). Of interest for how experimenters of the period inferred particle motion, temperature and the sequence of air then metal-vapour conduction in an oscillating Leyden-jar discharge. The paper makes no unusual or fringe claims; the authors repeatedly flag uncertainty (doubtful values marked with a query, 'not claiming accuracy', 'we reserve the decision').

Mainstream conflict: Interpretation of line slope as vapour velocity is contested (Hull 1906 Doppler limit; Royds 1908). The measurements themselves are not in conflict with mainstream science. Later recasting of spark light as plasma emission at much higher temperatures is a [memory] point, not in the source. The authors report a counter-intuitive result (velocity falling with capacity at short gaps) without claiming it as established.