An artist's impression of Kepler-78b. Image: Karen Teramura (UH IfA).

Astronomers have discovered an impossible world – a rocky exoplanet that orbits its star at distance of less than 1.5 million kilometres and swelters at 2,000 degrees Celsius. It is one of a number of such worlds being discovered in the data from NASA's Kepler Space Telescope, but science cannot yet explain how these worlds came to be in such tight orbits.

What is known for sure about the planet, named Kepler-78b, is that it has 1.7 times the mass of Earth and is 1.2 times wider, with a diameter of 15,300 kilometres. Its density is just a tad less than Earth's – 5.3 grams per cubic centimetre compared to 5.5 – meaning Kepler-78b must be made from iron and rock. It revolves around its star every 8.5 hours. In other words, for every 24 hours on Earth, Kepler-78b goes through almost three of its years.

The planet turned up in data from the Kepler spacecraft that, up until this year, had been patiently watching 100,000 eligible stars in the constellations of Cygnus and Lyra for telltale dips in their starlight that signify the 'transit' of a planet across the face of the star. Because the stars are so far away – in the case of Kepler-78b it is 700 light years away – and because the planets so small it is not possible with our current telescopes to directly see these transiting worlds. However, based on the magnitude of the dip in light and how regular it is, scientists can determine the diameter of the planet and how big its orbit is. Its mass measurements comes from measuring the size of the gravitational tug of the planet on the star and, for Kepler-78b, the world's two most sensitive instruments for such measurements were employed – the HIRES spectrograph at the Keck Observatory in Hawaii and a debut-outing for the HARPS-North spectrograph at the Roque de los Muchachos Observatory in La Palma, which is a brand new Northern Hemisphere version of the hugely successful HARPS (High Accuracy Radial velocity Planet Searcher) built by astronomers at Geneva Observatory and based at the European Southern Observatory in Chile. That both instruments were able to produce the same results provides strong confirmation that the Kepler data is correct.

However, Kepler-78b could not have formed in its current location. In the past its star was larger and the planet would have actually been inside the star and incinerated. Therefore it must have moved into its orbit since then, but such a theory is not without its problems.

"It couldn't have formed in place because you can't form a planet inside a star," says Dimitar Sasselov who researches 'super-earth' planets at the Harvard-Smithsonian Center for Astrophysics. "It couldn't have formed further out and migrated inward because it would have migrated all the way into the star. The planet is an enigma."

Comparing two worlds, Earth and Kepler-78b. Image: David A Aguilar (CfA).

Many more impossible worlds have been discovered by Kepler, but Kepler-78b is the first to have its mass measured. Planets migrating from further out to closer to their star is a well-known phenomenon, starting with the discovery of the first 'hot jupiter' exoplanet in 1995, but the migrators tend to be gas giants that move inward while the protoplanetary disc is still in residence around the star and the planet can lose angular momentum to it as it accretes the gas. Astronomers have previously found worlds orbiting close to their stars that were once gas giants, but their outer atmospheres have been whittled away by the fierce heat and force of the stellar wind, leaving behind only a dense core. Could Kepler-78b be an extreme version of one of these worlds?

There is a twist to the tale. In three billion years Kepler-78b will be no more, the gravitational tides from its star having ripped it to pieces. Had our Sun harboured its own version of Kepler-78b around it, the Sun would have destroyed it long ago and we would never know it existed.

The science team using the HIRES instrument were led by Dr Andrew Howard of the University of Hawaii while the HARPS-North team were fronted by Dr Francesco Pepe of the University of Geneva. Both are having papers published together in the journal Nature.