This is stupendously difficult -- Jupiter would appear only one-billionth as bright as our Sun in visible light to an alien astronomer -- but it, too is now teetering on the brink of feasibility.

And unlike the other techniques, it is effective at detecting giant planets far from their suns.

(Obviously it is in fact better at doing so, since they're less drowned in the star's glare.)

In a talk at the Conference, Bruce Macintosh described the first Earth-based attempts to do it, which have been made possible by "adaptive optics" -- a revolutionary new technique in which telescope mirrors have their shapes slightly and constantly changed in response to real-time computerized measurements of the twinkling effects produced by Earth's air, thus cancelling out most of them and improving an Earth-based telescope's resolution by up to 100 times.

This technique, over the past few years, has begun allowing Earth-based telescopes to achieve visible-light resolutions comparable to the Hubble Telescope, and far more cheaply (although, unlike Hubble, they can't observe in ultraviolet or most infrared wavelengths).

He has been using one of the 10-meter Keck telescopes in Hawaii -- because of its huge mirror size, it is actually much more sensitive for this particular job than Hubble.

It is, in fact, following up on a preliminary survey made by the "NICMOS" IR camera on Hubble before the block of frozen nitrogen that it was using as a coolant ran out.

(The next Hubble servicing mission will install an active cooler to put the camera permanently back in operation.)

Recently formed gas giant planets -- because they're still cooling down -- are much brighter in the infrared than older ones like those in our own Solar System. (Jupiter, when 10 million years old, had a temperature of 300-500 deg C, and was "only" 100,000 times dimmer in infrared wavelengths than the Sun.) Right now the Keck Telescope, using its adaptive optics in the infrared, could spot a young Jupiter at 10 billion km from a nearby star, and a young planet 5 times Jupiter's mass at 5 billion km distance. It is, however, very possible that giant planets (although we do now know that they tend to migrate inwards) never form at such great distances from their suns -- which makes it no surprise that Keck, at this very early point in its survey, has thus far only confirmed the existence of one "brown dwarf", an object intermediate between a star and a planet.

(Keck's photos also clearly showed that this object's central star was a close binary, something Hubble had been unable to see.) And Chris McCarthy reported in a Conference poster that his own similar survey of 100 young stars had failed to reveal any planets, leading him to "the early conclusion that giant planets orbiting far from their host star (about 30-100 AU) are quite rare, occurring in less than 5% of systems."

But Macintosh said that in the next few years, slightly bigger telescopes and longer observation times should allow detection of Jupiter-sized planets, at Jupiter's 800 million km from its Sun, around the six nearest stars.

And the earth-based telescopes being considered for the next two decades -- which will use computers to combine the images from dozens of separate mirrors to produce the equivalent of a single mirror with a diameter of 30-100 meters -- should be able to directly see Jupiters around dozens of nearby stars.