The Van Cittert

Field note. The date is not important. What matters is that I was standing in the noise and the noise was teaching me something.

Every radio astronomer learns, sooner or later, that the sky is lying to you. Not maliciously — the sky is just doing its job, which is to scatter photons in every direction, every which way, from sources that have no obligation to be neat. A star is not a point source in the sky. It is a Lambertian emitter, an extended disk, a collection of independent atomic transitions throwing photons into the sky like dandelion fluff caught in a wind you cannot see.

And then the van Cittert-Zernike theorem arrives like a revelation, or a punishment, depending on your temperament. It says, in effect, that the mutual coherence function measured in the far field of an incoherent source is the Fourier transform of the source's intensity distribution. Incoherent source. That's the key. The source does not care about phase relationships between different points on its surface. Each point emits independently. But you — you, with your pair of telescopes separated by a baseline — you measure the phase difference between the waves arriving at those two points. And that phase difference, that whisper-thin quantity, is the Fourier component of the brightness distribution.

I keep thinking about this because it feels like a violation of the natural order. You take a source that has deliberately discarded all phase information — thermal emission, spontaneous emission, the great cosmic shuffle — and you recover structure from it. The theorem is a bridge between chaos and order, and it works because the far-field propagation itself does the Fourier transform. The wave equation, under the Fraunhofer approximation, is a Fourier transform. The atmosphere doesn't know it's computing.

What makes this theorem useful, rather than merely beautiful, is that it is invertible. You measure the complex visibility at a set of baselines. Those are Fourier samples. You invert the transform, and you get an image. The resolution is set by the longest baseline. The fidelity of the image is set by how well you've sampled the Fourier plane. If you've only measured along a line, you'll reconstruct a line. If you've measured in a circle, you'll get circular symmetry. The Earth's rotation fills in the gaps, slowly, patiently, like a painter who works in layers.

There is a practical consequence that every student of interferometry discovers on their first real dataset. The theorem assumes an incoherent source. Real sources are not incoherent. Masers are coherent. Pulsars are coherent. Some fraction of the emission from every active galaxy is at least partially coherent. When you apply the van Cittert-Zernike theorem to these sources, you get something that looks like an image but isn't — a corrupted Fourier transform of something that was never the Fourier transform of an intensity distribution in the first place. The theorem is a lens. Lenses distort. The question is whether the distortion is useful or fatal.

I write this because I've seen too many people treat the van Cittert-Zernike theorem as a result. It is not a result. It is an invitation. The invitation is to think about what coherence means, to think about how waves propagate, to think about why the far field of a chaotic source is ordered. The answer, always, is the wave equation.