Historical research guide · 1134 · 1920

A Determination of the Deflection of Light by the Sun's Gravitational Field, from Observations Made at the Total Eclipse of May 29, 1919

F. W. Dyson · A. S. Eddington · C. Davidson

In one minute

This is the measurement that made general relativity a public fact. Einstein's 1915 theory says a light ray grazing the sun is bent by 1.75 arcseconds, exactly twice what you get if you treat light as ordinary matter falling under Newton's law. The only way to see it is to photograph the star field around the sun during a total eclipse and compare it with the same field photographed at night months later. The Royal Society sent two expeditions to the eclipse of 29 May 1919, one to Sobral in northern Brazil and one to the island of Principe off West Africa. The paper is unusually candid about its own instruments: three sets of plates gave three answers, and the authors argue at length about which deserves the most weight. The best set, the 4-inch lens at Sobral, gives 1.98 arcseconds with a probable error of about 0.12. The authors' conclusion is that a deflection takes place and that it is of the amount Einstein's theory demands.

Connection

Chapter 4 teaches that mass tells spacetime how to curve, and that the curvature is a real, measurable geometry rather than a way of speaking. This is the first time anyone weighed that geometry against a rival theory with an instrument, and it is the historical anchor under every later metric measurement the site cites. It is also a model of transparent publication: the discordant plate set is printed, discussed and kept in the paper rather than dropped.

From the source

The expeditions were designed to discriminate between three possibilities: that the path of light is uninfluenced by gravitation; that light is subject to gravitation as ordinary matter is under a strictly Newtonian law, giving an apparent outward displacement at the sun's limb of 0.87 arcseconds; or that the ray follows Einstein's generalised relativity theory, giving 1.75 arcseconds, exactly double the Newtonian value.Part I, Purpose of the Expeditions, section 1
The plates from the 4-inch lens at Sobral, which the authors judge the most trustworthy for the superiority of the images and the larger scale of the photographs, gave 1.94 arcseconds from declinations and 2.06 arcseconds from right ascensions, a mean of 1.98 arcseconds with a probable error of about 0.12 arcseconds.Part V, General Conclusions, section 39
The Sobral astrographic plates gave 0.93 arcseconds, discordant by an amount much beyond the limits of its accidental error; the paper prints that number, explains at length why it attaches little weight to it, and does not remove it from the record.Part V, General Conclusions, section 39
The observed deflection cannot be refraction by coronal matter: to produce it that way the sun would have to be surrounded by material whose refractive index at a height of one solar radius equals that of air at about one seven-hundredth of an atmosphere, a density the authors call out of the question.Part I, Purpose of the Expeditions, section 4
The authors' conclusion is that the results of the expeditions to Sobral and Principe can leave little doubt that a deflection of light takes place in the neighbourhood of the sun and that it is of the amount demanded by Einstein's generalised theory of relativity, as attributable to the sun's gravitational field.Part V, General Conclusions, section 39
What to watch, stated by the authors in 1920: the third prediction of the theory, a displacement to the red of the Fraunhofer lines on the sun of about 0.008 angstroms in the violet, was not confirmed at the time of writing, and they say so plainly rather than setting it aside. They also call for the eclipse measurement to be repeated, noting that the unusually favourable conditions of 1919 will not recur and that fainter stars further from the sun will have to be photographed.Part I, section 4; Part V, General Conclusions, section 39, closing paragraph

The way in

https://archive.org/details/philtrans06337895

Historical source-access record. This guide links to the original; it does not reproduce the paper text.

TEXT. Read on 2026-09-11 from the Internet Archive scan of the original Philosophical Transactions printing, item philtrans06337895, which is the full 43-page paper. Reproduced below: Part I, Purpose of the Expeditions, sections 1 to 4 complete, and Part V, General Conclusions, section 39 complete, including the radial-displacement table. Parts II, III and IV, which are the instrument descriptions, the plate measurements and the reduction tables for Sobral and Principe, run to thirty-five pages of plate constants and are not reproduced; their results are carried in the claims. The scan is optical character recognition of 1920 letterpress and it garbles some words and drops some figures. Obvious character errors have been corrected against the surrounding sentence and the arcsecond notation is reset here as a number followed by a double prime. One figure is genuinely lost: the sentence reporting the Principe deflection reads, in the scan, THE DEFLECTION OBTAINED WAS, then a blank where the numeral stood, then the probable error. That number is therefore not stated in this sheet or in its claims, because it was not legible in the source read.

How to cite it

F. W. Dyson, A. S. Eddington, C. Davidson (1920) A Determination of the Deflection of Light by the Sun's Gravitational Field, from Observations Made at the Total Eclipse of May 29, 1919 doi:10.1098/rsta.1920.0009

Where it sits in the curriculum

The metric, warp drives and wormholes