Cavendish's enclosed-globe inverse-square nulls with a sensitivity control

Primary source and reading limits. Henry Cavendish's original manuscripts as edited by J. Clerk Maxwell, The Electrical Researches of the Honourable Henry Cavendish (Cambridge 1879), 'Experimental determination of the law of electric force', arts. 217-235, printed pp. 104-113, with the detector and electrometer descriptions arts. 244-249, printed pp. 119-121; read in the Internet Archive OCR of electricalresea00cavegoog. The original apparatus, procedure, reported observations and sensitivity check are identifiable, but the fractional calibration ratios and the exponent computation are corrupted by the OCR and no numerical exponent limit is transcribed or recomputed here. The account below is the report's descriptions and observations structured by the record contract of Report-backed operational evidence with uncertainty; it is a reported finite-resolution null, not a proof that a true charge is exactly zero.

A conducting inner globe 12.1 inches in diameter, supported by a glass stick coated with sealing wax, enclosed between two tinfoil-coated pasteboard hemispheres 13.3 inches in diameter with a small clearance, the two hemispheres applied to each other to form a complete sphere; a removable wire connected the inner globe and the hemispheres, and could be drawn out by a silk string; linked strings operated the apparatus so that the charging wire, the inner connecting wire and the hemispheres were successively separated (paragraphs 219-222). A coated glass jar and an electrometer monitored and repeated the charging level through a separate electrometer fastened to the prime conductor (paragraphs 224-225). The inner-globe detector was a pair of pith balls about one-fifth of an inch in diameter on fine linen threads (paragraphs 220-222 and 244). The paper-cylinder electrometer of paragraph 248 monitored charging; the later straw electrometer with cork balls, suspended from the prime conductor, served that monitoring role in the subsequent charge-comparison experiments (paragraph 249), rather than replacing the inner-globe detector in the reported null tests. A second experiment enclosed a wooden piece between two wooden drawers forming a hollow box (paragraph 235).

The hemispheres were charged from a Leyden vial through a wire, which was then withdrawn; immediately afterwards the wire connecting the inner globe and the hemispheres was drawn out by the silk string, the hemispheres were separated without touching the inner globe, and the outer shells and charging wire were discharged; a pith-ball detector was then brought to the inner globe (paragraphs 220-222 and 226). For greater sensitivity the pith balls were precharged positively before the charge, the comparison being made against the negative-precharge case at monitored equal charging levels (paragraphs 228-229). For the sensitivity control the hemispheres and their frames were removed, leaving the globe and pith balls. A smaller coated glass element was charged to the monitored level and its charge transferred to the initially uncharged main jar; that jar then charged the exposed globe. After withdrawal of the charging wire the precharged pith balls were compared (paragraph 230). The report's fractional charge comparisons are unusable in this OCR; no calibrated charge value is assigned here.

Several repetitions without perceived pith-ball separation or signs of residual inner charge (paragraph 227); in the precharged-ball comparison, equal small residual separations for positive and negative precharge rather than a detectable charge-sign asymmetry (paragraphs 228-229); in the sensitivity control, the small deliberately imposed charge was visibly distinguished by the same detector (paragraph 230); and a corresponding qualitative null with the enclosed-box geometry (paragraph 235). The reported 'must be inversely as the square' inference (paragraph 232) is recorded as the author's conclusion from the null under his theory, not as a modern theorem.

The read passages supply charge-level monitoring, the precharge comparison, a deliberate sensitivity check and the pith-ball and electrometer construction, but no calibrated charge transfer function, no trial-level sampling model, no confidence interval and no modern SI traceability. Insulation leakage, finite separation time, imperfect spherical geometry, shell holes, accidental contact and outside charges are apparatus-model and systematic conditions. The OCR cannot faithfully recover all fractional calibration ratios or the exponent calculation, so no numerical exponent bound, precision, likelihood or certainty is invented.

An ideal inverse-square enclosed-conductor equilibrium model predicts zero charge in the connected enclosed interior under its idealization, and the reported observations support that model within the apparatus sensitivity and auxiliary assumptions; they constrain specified alternative force laws only after an explicit response and sensitivity model, which the read passages do not supply in numerically usable form. A finite null cannot prove an exact exponent, and the empirical support is a nonlogical relation to the static prediction.