A radio telescope that wants to see the cosmic dark ages cannot be built on the ground, and it cannot be launched in one piece either. China's Hongmeng mission, one of four space science programmes named for the 15th Five-Year Plan period, proposes to answer both problems with the same trick: ten satellites flying in formation behind the Moon, with separations ranging from 100 metres to 100 kilometres. The reasoning behind each of those numbers is worth following, because none of them is arbitrary.
The signal is a shadow, and it arrives stretched
The dark ages run from the moment about 380,000 years after the Big Bang when protons and electrons combined into neutral atoms, to the formation of the first stars roughly 100 million years later. There were no stars in that interval, so there is nothing to see in the ordinary sense. What the hydrogen gas does instead is absorb microwave background photons at its 21-centimetre hyperfine transition, cutting a faint dark line into the background spectrum.
Expansion does the rest. A line emitted at 21 centimetres arrives stretched by the time it reaches a telescope today; the Beijing Daily account of the mission notes that after 13.7 billion years the wavelength exceeds 2 metres, and that the standard cosmological model places the dark-age absorption band at 15 to 20 metres, with the later dawn era, when stars began forming, around 4 metres. The dark-age band carries information about primordial density fluctuations and the expansion history; the dawn band records when the first stars formed and how their energy was distributed.
That band is where the engineering problem starts. Earth's ionosphere refracts, reflects and absorbs at these wavelengths, and human radio traffic adds interference on top. The signal is already weak next to foreground emission from the Milky Way. A ground-based dish, however large, is looking through a wall that flickers.
Why ten spacecraft, and why 100 kilometres
Aperture has to scale with wavelength. The source puts the comparison plainly: Hubble works at 2.4 metres in visible light, the Webb telescope at 6.5 metres in the infrared, and China's FAST dish has a 300-metre effective aperture for the 21-centimetre hydrogen line. At wavelengths around 20 metres, resolving power of any use means an aperture of tens of kilometres. No fairing holds that.
Very-long-baseline interferometry sidesteps the fairing. Separate telescopes, separated by large distances, record the same wavefront and their signals are combined afterwards, so the array behaves like one aperture as wide as the spacing between elements. Hongmeng applies this in orbit, with the shortest satellite-to-satellite baseline at 100 metres and the longest at 100 kilometres, and with orbit control intended to change those spacings so that the array covers the observing band more completely.
Ten spacecraft, rather than two, follow from the same logic. A pair of telescopes gives one baseline and one orientation; ten give 45 distinct pairings, so a single pass yields many samplings of the sky at once instead of one. The spread of baselines matters as much as the longest one: short spacings at 100 metres are what pick up the broad, low-contrast features of the dark-age absorption, while the 100-kilometre spacings supply resolution. Working the diffraction limit, λ/B, a 100-kilometre baseline gives roughly 40 arcseconds at 20 metres and roughly 8 arcseconds at 4 metres. That is coarse next to optical astronomy and entirely normal for a first survey of a band that has barely been observed.
Why the far side, and why a relay
The lunar far side is the quietest radio location within reach. It is shielded from Earth's transmitters by the body of the Moon itself, and the source describes the electromagnetic environment there as exceptionally clean. Under the plan, one main satellite and nine sub-satellites would be launched together as a single stack, captured by the Moon into a 300-kilometre target orbit, and then separated into the formation. The array observes while it is behind the Moon. When the formation comes around to the near side, the main satellite downlinks the recorded data to Earth.
That relay arrangement is a scheduling constraint, not an afterthought. A spacecraft on the far side has no line of sight to Earth, so the observing cadence and the downlink cadence have to be designed around the same orbit.
What else the array is for
The same instrument supports work beyond cosmology: the plan lists a first high-resolution survey at ultralong wavelengths, a three-dimensional map of the interstellar medium in the Milky Way, study of the solar system's immediate surroundings, and observation of ultralong-wave activity within the solar system to explain how space environments interact. A low-frequency array behind the Moon serves several fields with one payload.
Where the engineering gets hard
Interferometry needs the relative positions of the elements and their clock offsets known well enough that phase errors stay small — a general requirement for any VLBI array, not a criticism specific to this one. Holding ten free-flying spacecraft in a controlled formation across 100 kilometres, while tracking their geometry and timing to that standard, is the difficult part of the design. The published account describes the baselines, the orbital control measures and the observing strategy; it does not set out the metrology, the propulsion budget or the data rates, and the mission is a plan rather than a flown instrument.
What the plan does make clear is the chain of reasoning. The 21-centimetre line arrives at tens of metres, so the aperture has to be tens of kilometres. Nothing that wide fits in a launcher, so it is synthesised from separated receivers. Separated receivers need a radio-quiet site, so they go behind the Moon, and something has to carry the data back around the edge. Ten satellites, 100-kilometre baselines and the lunar far side are not three separate ambitions; they are one answer to a single wavelength problem.