A European army that cannot fly a weather balloon loses more than a forecast. It loses the atmospheric profile on which artillery fire-control corrections are computed, the advice it gives to aircrew and to the operators of unmanned aircraft, and the input to any prediction of how nuclear fallout or a chemical or biological agent will drift. The instrument that supplies all of this is a disposable package of sensors and a radio transmitter, costing a few tens of euros, attached to a rubber balloon by a length of cord, and it is thrown away after every flight. That combination — a trivial unit cost, an absolute operational dependency and a consumption rate measured in thousands of units a year — is what makes continuity of supply a defence-finance question rather than a meteorological one. The instinctive way to pose that question is to ask whether Europe can build its own radiosondes. It can, in four countries. The answer is therefore not where the difficulty lies. The difficulty lies one layer down, at the point where a sonde has to be received, decoded and prepared by a particular piece of ground equipment, and the question worth asking is narrower and more awkward: at which layers of a European upper-air sounding service can procurement and integration actually be separated on the evidence available in public, and where does that evidence still show the installed system tied to a single manufacturer’s family?