ic

Here we offer an account of the remarkable life and work of this robot surveyor out in clear space. We are talking angles, angles that subdivide the sky. Dividing any circle into equal segments of arc is a well-posed task: the length is not important, for there are no ends. You need agree only on the subdivision. The choice long ago in Babylon was 360 degrees, appealing when arithmetic was a skilled craft and maybe clinched by the hint of days in a year. One degree of arc was and is parted into 60 small arc minutes, and one minute of arc is divided further into 60 exquisitely small arc seconds.

Mountaintop observatories at their best see stars as half-second disks of twinkling light. Hipparcos offers a median error of a thousandth of an arc second, possible because it is outside of all the restlessly varying air. With that acuity, you could make out a poliovirus particle across the room.

The first precision sky surveying was done about 1840, before photography, by astronomer Frederick Bessel, along with a master of instrumentation, Joseph Fraunhofer. Between the two of them they built an original instrument based on a telescope eight feet long, with a six-inch split lens. The stars showed up in its eyepiece as doubled spots.

Micrometer screws allowed the patient observer to shift the lens halves until the two image spots coincided. By moving a star image until it coincided with another reference star, any image shift could be measured. By viewing the same nearby star after half a year's wait, Bessel could measure the star's apparent shift caused by his own movement, along with Earth, across the known diameter of our planet's annual orbit.

An analogy is familiar. While you watch from a moving car, the roadside trees shift in direction rapidly, whereas a tree far across a wide field hardly shifts at all. The car would form the base of a triangle by its known motion, and out in the field the tree would lie at the apex. The stars behave the same way, but their shift in direction is amazingly small.

The altitude of Bessel's extreme triangle to the star 61 Cygni was about 300,000 times as long as its base! The center of the reddish splotch of the star shifted only a few thousandths of an inch during one round-trip on Spacecraft Earth. The triangle's apex was 10 light-years away, the first firm milestone among the stars.

This elegant, patient art, called astrometry, remains vigorous. With the right telescopes, its adepts have laid out hundreds of measured triangles to close stars, building the platform for the rickety, many-runged ladder of big cosmic distances. All other longer cosmic distances depend on ingenious but risky assumptions, supplying estimates calibrated by a chain of links to those few but reliable if very pointy triangles.

The European Space Agency moved astrometry out into the cosmos on Hipparcos, built to divide the circle into a billion steps of tiny arcs, about one milliarc second each. Star images were seen at the modest resolution of its 11-inch telescope, but the images were steady. It scanned the entire limpid sky, lived in microgravity too small to distort parts as the probe turned, and kept tight temperature control over the graphite-fiber telescope mounting. This astrometric station was scrupulously monitored and superbly stable.

Hipparcos was given a wonderful legacy, its Input Catalogue, endowing it with the then best of ground astrometric data. From launch late in 1989 to mid-1993, the satellite watched the sky. The first processed results began to appear in the spring of 1997, for each star sighting took meaning only after elaborate comparisons with many others.

Four distinct teams--hundreds of astronomers, engineers and data mavens--worked long at the task. They had been confronted by a sudden, daunting failure on launch; the rocket did not put Hipparcos into the planned geostationary orbit. As Earth then turned beneath Hipparcos, no single ground station could collect the day's stream of data beaming downward by radio. Quickly, however, not one but three ground stations were set up around the world. In the end two thirds of all expected data came in, a triumph of masterful salvage in space.

How does it work? The satellite scanned the sky as it orbited, spinning on its viewing axis. Its long, elliptical Earth orbits took 10 hours, sampling stars in a belt about a degree wide in the sky. Meanwhile the auxiliary star-mapping detectors picked up catalogued guide stars to find approximately where the axis pointed and thus to identify which stars were being seen. Every star signal that entered the optics crossed a couple of thousand slits in the main electronic sensors as the probe spun, the light registering as voltage pulses. Individual stars generated long strings of pulses over time--some 2,000 stars during each 10-hour orbit.

After years of gathering multiple signals, Hipparcos began to pull out the positional information entangled in this huge database. Because of the precision with which the relative positions of the stars were timed, Hipparcos was able to record thousands of tiny parallax-induced shifts for individual stars at different points in Earth's orbit as the satellite rounded the planet again and again.

The Input Catalogue began the whole process; each program star was revisited about 100 times, far overdetermining the numbers needed to locate a star. That prodigious session of years of heavy computation was planned and executed by two firmly independent expert teams. Their iterations led to consistent positions for each of the 100,000 stars, with errors noted.

Named for the star cataloguer and savant of ancient Rhodes (cited as a major source by old Ptolemy himself centuries later), Hipparcos has mapped our local star neighborhood anew. In one decade it has advanced all ground-based astrometry by 100-fold in coverage, making fivefold or 10-fold gains in precision as well. It confirms most well-regarded astrometric distances to a few percent.

Has our locality been mapped for the last time? Hardly. The Cepheid class of very bright variable stars is marked by peculiarly regular light curves: the faster the pulsed cycle, the brighter the star. Thus recognizable from far away by their timed pulses, Cepheids are major sources for the Hubble distance scale. Hipparcos distances move out by about 20 percent the few nearby Cepheids whose distances were calibrated earlier. That implies a slower expansion, hence an older cosmos, comfortably reducing the conflict between redshift age and luminosity age for external galaxies.

A masterstroke of recalibration? Maybe. There is a long, uncertain chain of theory between the distance to a Cepheid and the age of a remote cluster. Nor do all the rich Hipparcos results yet concur. We cannot assume that good calibration is all we need. But one day sufficient benchmarks in many wavebands will indeed pin down the grand cosmic map to microarc seconds, a solid foundation. Patience.