The machine behind the world's methane, free airwaves for satellites, and the narrow escape from a test case

Methanogens are the only living things known to make methane, and they have been doing it in mud, sediment, guts and hot springs for much of Earth's history.[5] To pull energy from almost nothing, they run a chemistry that sits at the very edge of what thermodynamics permits, stitching hydrogen and carbon dioxide together into methane. A team centred on the Microbes-for-Climate cluster at Philipps-Universitat Marburg has now imaged the enzyme assembly that manages the electrons for that reaction, and it is one of the largest soluble protein machines ever described.[1][2]

The structure, reported in Nature on 8 July, is a ring roughly fifty nanometres across, built from 252 separate protein chains and carrying more than 600 electron-shuttling cofactors, among them iron-sulfur clusters and flavins.[1] Two six-sided rings of hydrogen-splitting units are joined by carbon-dioxide-reducing modules into a closed loop, so that electrons can travel around the assembly rather than down a dead-end chain.[1] The researchers reconstructed it with cryo-electron microscopy, a method that flash-freezes proteins and rebuilds their shape from thousands of images.[2]

The loop performs a trick called electron bifurcation, a mechanism of biological energy coupling described only in the last two decades.[6] An enzyme takes a pair of electrons and splits them, sending one downhill to an easy destination and using that release to shove the other one uphill to a reaction that would otherwise never run.[6] In this machine the same circuit links the last step of methane-making back to the first, which is how the microbe balances its energy books while barely breaking even.[1] Jan Schuller, who led the work, framed it plainly, saying nature has built elaborate molecular machines to wring energy from extreme conditions.[2]

The images also caught the machine improvising. In cells starved of hydrogen, about one in five of the assemblies had swapped its hydrogen-using part for a module that draws electrons from formate instead, a small organic acid.[2] Because the team also imaged the complex inside intact, frozen cells and not only in a test tube, that flexibility appears to be how the microbe actually behaves, an inference the authors ground in measurements from whole cells.[1]

The stakes reach well beyond one microbe. Methane is the second-largest human-driven contributor to warming after carbon dioxide, and traps heat roughly eighty times more effectively over a twenty-year span.[3] Biological sources dominate the budget, with agriculture and waste the leading human inputs, all of them ultimately the work of methanogens.[4] None of this is an emissions fix, and the authors do not claim one. Knowing the exact wiring of the machine that makes the gas is a starting point, both for imagining ways to slow it and for reading the ancient carbon cycle it has quietly run for billions of years.[5]

A hiker past the last cell tower, a soil probe in a distant field, a shipping pallet halfway across an ocean: the Federal Communications Commission wants each of them to reach a satellite using the same cheap radio that already runs a home Wi-Fi router.[2] On 16 July the agency circulated a draft notice of proposed rulemaking that would open more than 225 megahertz of unlicensed spectrum for direct-to-device satellite links, the kind of connection that skips the cellular network entirely.[1] Commissioners are scheduled to vote on 6 August to formally ask the public how it should work.[3]

The proposal reaches into three slices of what engineers call the Part 15 bands, the unlicensed airwaves that anyone may use without a license so long as they do not cause harmful interference.[4] These are the frequencies that already carry Wi-Fi, Bluetooth, smart meters and garage-door remotes.[2] Under the plan a certified device could transmit to a satellite at its existing power ceiling, roughly four watts, without new hardware.[4] Any satellite service built on these bands would run on an unprotected basis, meaning it must shield existing users from interference and accept whatever interference it receives in return.[4]

The plan's most unusual feature is its licensing, or the near-absence of it.[4] Rather than force each ground device to file as an individual earth station, the commission floats a "license by rule" approach, in which certification alone authorizes the link.[4] That extends a shift the agency began in 2024, when its Supplemental Coverage from Space framework first let satellite operators borrow mobile carriers' licensed airwaves to fill coverage gaps.[4] The new step drops both the licensed spectrum and the carrier partner, and points the same logic at the vast installed base of unlicensed sensors and radios already in the field.[6]

The economics are the sharpest part of the story. Access to these bands would be free, a contrast with the licensed market, where SpaceX has committed about 19.6 billion dollars to direct-to-device spectrum, Amazon is buying Globalstar for some 11.6 billion, and Rocket Lab is acquiring Iridium for roughly 8 billion.[5] The commission's chairman has framed the move as a way to draw more investment into a sector that has already seen more than 40 billion dollars in spectrum deals.[5] The clearest beneficiaries are small satellite Internet-of-Things ventures that today operate on case-by-case waivers, and commercial space-station builders that could add a Wi-Fi access point to a module without a bespoke authorization.[6]

Who might lose is less settled. The 2.4 gigahertz band sits close to frequencies watched by radio astronomers, and a slice near 5.8 gigahertz already carries amateur-satellite downlinks, so incumbent users have begun warning of crowding and interference, particularly on the satellite-to-ground path.[4] The draft leaves the hardest technical questions open, seeking comment on power limits, beam footprints and combined emissions that would decide whether millions of new links can coexist with Wi-Fi networks and federal radar.[4] Those open technical questions, more than the vote itself, will settle whether the freed spectrum becomes a usable commons or simply added noise that no one is obliged to quiet.[6]

The machinery of mass litigation in England rests on a trade. When many claimants raise the same legal question, a court can bundle them under a group litigation order, enter them on a shared register, and pick one claim as the test case that settles the common issue for all. AXA Insurance and Guardian Royal Exchange Assurance had sat on such a register since 2003, part of a long fight to recover corporation and advance corporation tax levied in breach of European Union law.[1] They expected to inherit a favourable ruling won in the group's Prudential test case. On 27 July the Supreme Court told them they could not.[3]

The rule at the centre of the case is a single line of the Civil Procedure Rules. Under rule 19.23, a judgment on a group-litigation issue is binding on the parties to all other claims on the group register at the time it is given, unless the court orders otherwise.[1] That default is the whole point of the device, the court explained, because it buys finality and efficiency by stopping every claimant from relitigating an issue already decided.[5] The dispute turned on the four words at the end, and on how rarely they can be invoked.[1]

The test case went stale because the underlying law shifted. The Prudential ruling had recognised a common-law claim to interest on tax paid before it was due, resting on an earlier House of Lords authority, Sempra Metals. Later decisions overruled that foundation and established that no such common-law remedy exists.[3] Holding AXA's group to the old ruling would have required the Revenue to pay out tens of millions of pounds on a legal basis now known to be wrong.[3]

The court, in a judgment given jointly by Lady Rose and Lord Richards, dismissed the appeal and upheld the decision to order otherwise.[2] Its reasoning ran by analogy to issue estoppel, the doctrine that stops a party reopening a point already decided between them, and the justices noted that even that rule bends where its rigid application would work injustice.[1] The holding is framed as a tightly guarded exception rather than a general licence to escape a bad test case.[1]

"Having regard to the importance of not undermining the efficacy of the GLO regime, there will need to be some exceptional quality about the circumstances before it becomes appropriate for the court to consider whether to exercise the power."[1]

A second strand of the ruling drew a cleaner structural line. The limitation question the appellants relied on had never been a binding group issue at all, the court held, because the test-case judgment had decided it on facts specific to those claimants rather than as the common point of law.[1] That leaves two distinct routes off a group register. An issue may fall away because it was always too fact-specific to bind, or a court may decline to apply a genuinely common ruling in the exceptional case where the law beneath it has collapsed.[5] For the mounting number of group actions in the English courts, from consumer claims to data-breach suits, the practical signal is that the binding default is strong and the way out is deliberately hard to reach.[4]