The delay of one cloud is its distance from the source minus its displacement toward the observer, divided by light speed. Subtracting the direct journey is the essential step. It explains why equal radii can give different delays and why equal delays need not mean equal radii.
Define the clock before measuring a delay
Prerequisites are distance divided by speed, a right triangle and the idea of a coordinate. Take the continuum source as the origin. The distant observer lies in the positive z direction; x and y lie across the sky. A cloud is at (x,y,z), at source distance r=x2+y2+z2. All lengths in the examples are light-days. Delays are days in the source rest frame.
Set the source flash at time zero. Suppose the observer is a distance D from the origin, much larger than the cloud distribution. The direct photon travels distance D. A reprocessed photon first travels r, then approximately D−z. We neglect the cloud’s displacement during the light crossing and take its response time to be negligible.
τ=cr+D−z−D=cr−z.
The large distance to the galaxy cancels. We compare two arrival times at the same telescope, not the source clock with the telescope clock. The distant-observer approximation neglects terms of order r2/D in path length.
If θ is the angle between the cloud position and the positive z axis, then z=rcosθ. The formula becomes τ=(r/c)(1−cosθ). Angles measured from the opposite axis give a plus sign instead. The diagram and angle definition must travel with the formula.
One radius gives a range of times
For a cloud directly in front of the source, z=r and τ=0. A cloud directly behind it has z=−r and τ=2r/c. A sideways cloud has z=0 and τ=r/c. These are geometric limits under a prompt response, not three different gas-recombination times.
Worked example. A ring has radius twenty light-days and is tilted by thirty degrees from face-on. Its nearest and farthest points have z=+rsini and z=−rsini, where i is the inclination. Since sin30∘=1/2, the first and last responses arrive after ten and thirty days.
τmin=cr(1−sini),τmax=cr(1+sini).
Their midpoint is twenty days. The half-width is ten days. A symmetric uniformly responding ring therefore keeps the same mean delay as it tilts, although the spread grows. Unequal near-side and far-side response weights can change that mean.
Analytic teaching example, not measured data. A twenty-light-day ring gives a single geometric delay when face-on and a ten-to-thirty-day interval at thirty-degree inclination. Finite cloud samples and time bins broaden sharp features in numerical plots.
Equal-delay surfaces are curved
At fixed positive delay, write ρ2=x2+y2 for the squared distance from the sightline. The relation r−z=cτ gives r=z+cτ. Squaring and cancelling z2 gives
z=2cτρ2−(cτ)2.
This is a paraboloid around the sightline. A delay selects a curved slice through the gas, not a sphere around the black hole. The zero-delay axis is a limiting case and is not obtained by dividing by zero in this expression.
The observer sees different points on that paraboloid respond together because the first and second path lengths compensate. A time sequence of responses sweeps these surfaces through the source environment.
Practice with geometry
A cloud lies fifteen light-days from the source and nine light-days toward the observer. Find its delay.
Two clouds both respond after twelve days. Must they have the same radius?
A symmetric ring has first and last responses at six and eighteen days. Find its radius and inclination, within the thin-ring model.
A source has cosmological redshift zcos=0.1. Its rest-frame delay is twenty days. What delay would the telescope measure?
Answers with the reasoningShow the answer
The extra path is fifteen minus nine light-days, so the delay is six days.
No. The measurement fixes r−z, leaving different radii and sightline positions possible.
The midpoint gives r/c=12 days. The half-width is six days, so sini=6/12=1/2 and i=30∘.
Cosmic time dilation gives τobs=(1+zcos)τrest=22 days. The coordinate z and the cosmological redshift have different meanings.
Sources
Peterson and Horne, arXiv:astro-ph/0407538v1, Section 2 and Figure 1. Their illustrated angle is measured from the opposite direction; the derivation here explicitly uses the observer-facing positive axis.