The small French town of Kourou is on the northern coast of the South American landmass, at the edge of the Amazonian forest. From the air, the approach to the region is over a dense mat of trees broken only by the occasional river. The person next to me on the plane remarked that the shape of the trees and their closeness to each other made it look like a forest of broccoli. On a sandy beach covered with palm trees sits a modern hotel where the penitentiary once stood (the notorious Devil's Island is visible a few miles offshore). We arrived the evening before launch after an hour-long bus ride from Cayenne airport, in time for a banquet of fresh seafood, baked plantain, and tropical juices as well as more traditionally French dishes. I didn't get to see too much of the town itself, but it has a population of around 17000 consisting mostly of recent immigrants from the rest of South America, the Caribbean and Europe, together with indigenous Amerinds and a village of the descendants of escaped African slaves who developed their own communities. The next morning, a short bus ride to the west of town takes us to the Centre Technique, and the Jupiter launch control room where the representatives of the various agencies briefed us.

Launcher V113 was an Ariane 44L model, and carried the most massive payload of any Ariane 4 to date, with 4.9 tonnes carried to orbit; it was also the first time Arianespace has carried out three launches in a single month. The V113 vehicle used the lightest stages available at the factory, and the fuel in the lower stages was kept at a lower temperature than usual to increase its density and allow a few extra kilograms to be loaded - they ended up with more than 230 tonnes in the L220 first stage. These measures allowed the record payload, and Arianespace president Jean-Marie Luton predicted that eventually we'll see a 5-tonne payload on Ariane 4.

Two satellites were carried, Afristar and GE 5; both were delivered to geostationary orbit. The lower payload is the GE-5 satellite, with C and Ku band transponders to augment the GE Americom system. GE Americom contracted with Daimler-Benz Aerospace/Dornier Satellitensystem GmbH/Friedrichshafen to provide the satellite in orbit. They in turn contracted Alcatel/Cannes to provide the Spacebus 2000 satellite, originally built as a backup for Argentina's Nahuelsat. Dry mass of GE 5 is 769 kg; it carries 950 kg of propellant at launch.

The upper payload was the first WorldSpace satellite, Afristar. Afristar will broadcast digital radio over Africa and the Middle East. Small handheld radios will be able to pick up the transmissions from its three L-band beams; the satellite can carry a large number of radio channels. It will be followed next year by Asiastar and Ameristar. Afristar is a Matra Marconi Space (Toulouse) Spacebus 2000, using a Marquardt R-4D apogee engine. Prime contractor for the combined satellite and comms payload is Alcatel. Dry mass of Afristar is 1205 kg; it carries 1534 kg of propellant at launch. The WorldSpace project is dominated by the personality of its CEO, Ethiopian-born Noah Samara, whose ebullient personality was very much in evidence at CSG during the V113 launch. There was a strong feeling that Afristar was not just yet one more `boring' comsat, but part of a crusade to empower the developing world by providing improved access to information. Samara's mantra is that `people are only as developed as the information they access.' Another of the leading figures in WorldSpace, chief engineer Pierre Madon, was feted for completing a notable career in aerospace which included a leading role in the first French rocket program Diamant and the Symphonie comsat of the 1970s as well as a long career at Intelsat.

CSG is operated by the French space agency CNES. The Ariane launch vehicle was developed by the European Space Agency, ESA, together with CNES, and is operated by the Arianespace company. Travelling further west from Jupiter on the old Route Nationale 1 coast road, we reach the CSG proper, with the entrance guarded by the French Foreign Legion. The launch pads are on the north side of the road, nearer the sea. We first pass the small clearing which marks the sounding rocket launch area ("aire de lancement fusee-sondes"). Here in 1968 was the first launch from CSG, a small Veronique rocket. The area has four launchers, three still in use for small weather rockets and amateur launches. A little further and we reach the Diamant pad. Used between 1970 and 1975, CNES launched several small satellites from here using the Diamant B and Diamant B P.4 vehicles. The other old pad was the Aire de Lancement Europa, some distance to the west. A single orbital launch attempt from here by the Europa vehicle failed in 1971. However, by 1979 the pad had been rebuilt to become ELA 1, the first Ensemble de Lancement Ariane, marking the beginning of Europe's success in the commercial space launch services business with Ariane launches from 1979 to 1989, when the ELA1 pad was retired. Only a water tower marks the spot currently. Next to ELA1 is ELA2, which we didn't get too close to as it was occupied by our fully fuelled V113 launch vehicle. In the early afternoon, the enclosed gantry was rolled back from ELA2 to reveal the Ariane rocket on the pad, and fuelling of the cryogenic third stage began. We were able to observe the rocket from the roof of a nearby building - the third stage was enclosed in insulation, and no venting was visible. The pad is surprisingly close to the Ariane 4 assembly building, containing the V114 launcher now being assembled, and the nearby ESA and Arianespace offices.

The remaining launch site is ELA3, which is spread over a large area between ELA2 and the Diamant area. The rocket and payload are assembled in two large buildings, BIL and BAF, and are then taken out to the pad on a mobile launch platform (Table de Lancement) which travels on a small railway. We saw the launch platform used for the Ariane 503 mission being returned by rail to the BAF building, while a second platform was under construction nearby. The BIL and BAF are to the south of the main road while the pad is to the north.

The Ariane 5 pad itself, ZL3 (Zone de lancement 3) contains only the minimal equipment for launch, to simplify reconstruction if there is a pad accident. A simple umbilical tower is flanked by three large lightning towers which dominate the site's appearance, reminiscent of the N-1/Energiya pads at Baykonur. The main pad has a circular mount for the central core and mounts for the solid boosters on each side, above large flame trenches filled with water (the water suppresses reflection of sound energy from the launch which would otherwise increase the vibration levels inside the payload fairing). The trenches are similar to the ones I saw at Vandenberg's SLC-6 Shuttle pad. A large water tower and liquid hydrogen and helium storage facilities complete the picture.

After our visits to the launch site were complete, and a brief trip back to Kourou, we set off again to the Jupiter control center where we watched the final countdown from the auditorium surrounding the launch control room proper. (We had the choice of going to the outdoor Toucan viewing point closer to the pad, but I decided it would be geekier to get as close to the launch controllers as I could). The launch commentary, with the cultivated and reassuring tones of former BBC man Martin Ransom lending a touch of class, punctuated the countdown as the display screens showed live TV from the pad beside the clocks and status displays, and the controllers pored over their consoles in front of us.

At T-1 minute, the side doors of the auditorium were opened and we rushed out on the terrace to watch. We scanned the dark horizon and didn't know which direction to look, but then 'Allumage!' and a bright light appeared in the distance to our left. Slowly this new star rose into the sky - initially it was pretty much pointlike. It arced over our position and we began to see the trail of fire behind it. About half a minute later, the sound began to reach us as a dull roar which grew to a loud crackling sound. Separation of the strapon boosters was visible as a dramatic flare, and we were able to follow the rocket through first stage separation as it moved down the coast to our right. The disappearance of the first stage plume as followed by a bright flash like a firework as the separation occurred. Trooping back to our seats, we heard the report of second stage separation and settled down to watch the graphic of the long third stage burn. At T+20 min the Afristar satellite finally separated and a huge cheer went up, followed two minutes later by another cheer for GE 5. Afristar and GE 5 were placed in a 200 x 35788 km x 6.5 deg orbit. The launch team put on T-shirts celebrating the month's three launches (I thought the big boss Luton looked a little uncomfortable in such informal attire), and a long night of partying began. The following day, most of the group went off to tour Devil's Island, but I stayed in Kourou to interview space center old-timers and soak up some beach time to brace myself for Friday's 4:30 am homeward wakeup call.

Thanks to Arianespace for making my trip possible, and to Marie-Vincente Pasdeloup, Yves Dejean, Martin Ransom and Pierre Madon for their helpful information.