The Hanbury Brown

James Hanbury Brown was a man who understood that the universe rewards patient people. He spent his life looking at stars and listening to light, and on a hilltop in New South Wales in the nineteen fifties, he proved that photons — particles of light, quanta, whatever word you prefer — can conspire to tell you the size of a star without forming an image at all.

The context matters. In 1956, astronomers estimated stellar radii by assuming values for surface temperature and luminosity and solving the Stefan-Boltzmann equation. It was an inference, elegant but fragile, built on the hope that stars behaved like ideal blackbodies. James Hanbury Brown and Richard Twist wanted to measure directly. They wanted to point two telescopes at a star, measure the correlation between the photon arrivals, and read the angular diameter off the correlation function.

The idea was simple enough that everyone laughed. Light from a star is incoherent. The photons arrive randomly, following Poisson statistics. What correlation could possibly exist between two detectors looking at the same incoherent source at a slight angular separation? The conventional wisdom said nothing. The photons were independent. There was nothing to correlate.

But quantum mechanics disagrees.

The Hanbury Brown–Twiss effect — the photon bunching effect — arises from the Bose-Einstein statistics of photons. Photons like to arrive in bunches. Not because they attract each other, but because they are bosons, and the probability of detecting n photons at a given time is enhanced relative to a random Poisson distribution. The excess noise, the bunching, is tiny — on the order of the coherence time divided by the observation time — but it is real, and it depends on the angular size of the source.

A point source produces maximum bunching. An extended source produces less, because photons arriving at the two telescopes come from different parts of the stellar disk, and those parts are statistically independent. The correlation function falls off as the angular separation increases. The rate at which it falls off tells you the angular diameter.

James and Richard built their instrument at Narrabri. Two optical telescopes, mounted on a tower, separated by a baseline that could be adjusted. Photomultiplier tubes counted photons. A correlation computer — a machine in the era before computers were easy — measured the correlation between the two counting streams. And in 1956, looking at Sirius, they measured an angular diameter that agreed with the inferred value from luminosity and temperature. For the first time, a stellar radius had been measured directly.

The story has an ending that feels deserved. The Narrabri Stellar Interferometer went on to measure the diameters of dozens of stars, mapping the atmospheres of red giants, detecting binary companions invisible to other methods. James Hanbury Brown was knighted. And the effect that his colleagues thought impossible — the correlation of incoherent light — became one of the most powerful tools in observational astronomy.

I think about this because it's a lesson in trusting the noise. The bunching is a noise term, technically. It's the excess variance in the photon counting statistics. But James recognized that excess variance was information, not error. And in astronomy, where every photon is precious and every source is faint and distant, the ability to extract information from the shape of the noise is everything.