A spatial context to aid interpretation is fundamental to all sample analysis: whether those samples are rocks, medical tests, or police evidence, we need to know where they came from to interpret the results. Meteorites are our only samples of a protoplanetary disk; the only surviving physical record of the formation of our own Solar System (including the variety of local stellar sources that contributed to our disk, and the chemical processes that occurred within it); the only record of how differentiation and core formation occurs in planetesimals. And yet we have virtually no constraint on where they come from. Until we get sample-return missions to numbers of asteroids, what we need are orbits for specific meteorites. Unfortunately, out of tens of thousands of meteorites, we have good orbits for only four, and reasonable orbits for a couple more. The aim of this project is to determine orbits for numbers of meteorites. The results potentially have implications for every area of meteorite study, just as a knowledge of spatial context of any other sample has implications for all subsequent analyses of it. Meteoroids produce a bright fireball as they transit our atmosphere. By photographing the fireball from different angles, the atmospheric track of the object can be determined with great accuracy. If material survives to the surface as a meteorite, this allows us to work out what its orbit was before it entered the atmosphere, and also where it landed on the Earth's surface. This technique has been employed a number of times over the last 50 years, all in temperate regions of the northern hemisphere, but although hundreds meteorite falls have been observed, only four were recovered. The poor success rate is down to the difficulty in recovering a small rock in an area of several square kilometres when there is significant undergrowth. Our solution was rather simple. Over the last few decades, tens of thousands of meteorites have been found in the world's deserts. Put a fireball network in a desert and it should be much easier samples. We have designed a fireball observatory that can operate automatically in the harsh environment of the Australian desert. Based on previous fieldwork in this area, looking for old weathered meteorites, we should have about a 70% chance of finding meteorites that we see land. Our first grant, to put a small network of 3 observatories out in the desert, and test both the technology and concept, began June 2005. Two years on, fireball observatories have been built, deployed in the Australian desert, and successfully integrated with satellite internet and solar power. In addition to fully-functioning autonomous observatories, logistics to maintain them in operation, and support regular fieldwork, are in place. The UK now has a fireball camera network operating in the Australian desert. Orbits have been calculated from 22 fireballs - the first orbits to be determined from southern hemisphere fireballs. And most important: at least one of these fireballs has produced a meteorite on the ground. An expedition to recover this sample - and any others that fall in the meantime - will be mounted in the next field season. Our initial trial network has proven a success. We now need the support that will let us translate this success into a growing collection of meteorites with orbits, finally providing meteorite scientists with that most basic information: a knowledge of where their samples come from.