= Solution
In <general relativity>, <light> follows <null geodesics>. A weak <gravitational field> outside an isolated spherical <mass> gives the <Schwarzschild light deflection>
$$
\boxed{\alpha=\frac{4GM}{bc^2}}
$$
for <impact parameter> $b\gg GM/c^2$. At the solar limb this is approximately $1.75$ arcseconds. A Newtonian corpuscular calculation gives half this value; the full result depends on spatial <curvature> as well as the <gravitational time dilation>. The <parametrized post-Newtonian formalism> predicts $\alpha=2(1+\gamma_{\rm PPN})GM/(bc^2)$, with $\gamma_{\rm PPN}=1$ in <general relativity>.
\Image[/past-exam-of-the-relativity-course-of-the-university-of-cambridge/2012/paper-1-light-deflection.png]
{title=Weak-field light deflection}
The <Eddington experiment> compared photographic star positions near the eclipsed <Sun> with reference exposures taken without the <Sun> in that field. Rays bend toward the <Sun>, but their backward extrapolations place the apparent stellar images farther from it. Plate scale, focus changes, <atmospheric refraction> and comparison-field alignment must be separated from a radial displacement proportional to $1/b$. The original report discusses the instruments, plate reductions and competing predicted deflections; the test is a quantitative comparison with uncertainty, not a unique proof of a gravitational theory.<a determination of the deflection of light by the Sun's gravitational field by Frank Watson Dyson and others>[1920, pp. 291–333: eclipse observations, instruments and plate reductions]{ref}
<radio interferometry> permits an independent test without an eclipse. A 2005 <Very Long Baseline Array> experiment observed four compact sources at 43, 23 and 15 GHz. Differential <astrometry> suppressed common antenna errors, while a fit $\mathbf p(\nu)=\mathbf p_0+\mathbf I\nu^{-2}$ separated <refraction> in the <solar corona> from achromatic gravitational deflection. Tropospheric errors and radio-source structure also required calibration. The reported $\gamma_{\rm PPN}=0.9998\pm0.0003$ was consistent with <general relativity>.<Progress in Measurements of the Gravitational Bending of Radio Waves Using the VLBA by Edward Fomalont and others>[2009, §§II–III on observations, calibration and coronal subtraction; §V on the reported result]{ref}
Back to article page
