Several readers wrote here a few weeks ago asking for my own definition of a "planet." Since I have been critical of the IAU's definition, it's only fair that I stick out my neck. So here's my definition and reasoning. To be honest, I'm not completely satisfied with it. The rancorous debate shows that it's not an easy problem, and there's no obvious "correct" solution. If a perfect definition existed, astronomers would have agreed upon it long ago.
Here's how I see things. In our galaxy we see large clouds of gas and dust. They collapse gravitationally and form dense cores pockets of gas that go on to form either a single star or a multiple-star system. Most of the material ends up in the star, but since the cloud was probably rotating to begin with, some of the material settles into a disk of gas and dust around the forming star. To make a long story short, material starts clumping together inside the disk, and eventually forms objects ranging in size from pebbles to bodies the mass of Jupiter or larger. Since formation is a rather important process in determining an object's ultimate physical nature, planets are objects that form from material in these disks. But where does one set the upper and lower boundaries?
If an object contains between about 13 and 75 Jupiter masses, theorists calculate that it can fuse the trace amounts of deuterium (a heavy isotope of hydrogen) in its core for a few million years. Astronomers call such an object a "brown dwarf." Anything above 75 Jupiter masses is a star, which fuses ordinary hydrogen. Almost all astronomers agree that the 13-Jupiter-mass deuterium-fusion threshold sets a very natural upper boundary for planets. Anything above that limit is a brown dwarf, and not a planet.
Nature forms objects in a continuum of sizes, so any lower limit to a "planet" will be somewhat arbitrary, and this is the cause of the controversy and hoopla surrounding Pluto. Where should one draw that boundary?
I don't like the IAU's "clearing the neighborhood" definition, for several reasons. First, it's easy for an object to clear its neighborhood or control its region of space if it's close to the star, but it's a lot harder if it's farther out, where the volume of space is much larger and orbital periods are much longer. For example, Mercury controls its region of space, but even Earth, which is 18 times more massive than Mercury, would not be able to clear out the Kuiper Belt or control that region if it were moved to Pluto's distance. So the IAU's planet definition is, in my view, unfairly biased toward objects near their host stars.
Second, astronomers will almost certainly find Mercury-sized or even larger objects in the distant reaches of the solar system. At that point, the IAU's definition will face even more criticism than it does now.
Third, it's very likely exoplanet hunters will eventually find two massive planets sharing the same orbit in a "Trojan" configuration. Under the IAU's definition, neither body clears its neighborhood, or even controls its region of space. But everyone would know that such objects are indeed planets.
So I think we need a definition that's based on size, not dynamics. Several planetary scientists have expressed this view to me in the past few weeks. I'm not claiming this is a perfect solution, but the least-arbitrary lower physical boundary is established by the concept of hydrostatic equilibrium. Admittedly, there will be fuzzy objects around the boundary, but at a certain composition and mass, the self-gravity of an object will overcome the material strength of the object and fashion it more or less into a sphere (which can be flattened somewhat by rotation). The IAU's definition of both "planet" and "dwarf planet," and the original proposal of the IAU committee chaired by Owen Gingerich, incorporated this concept.
One reason I think it makes sense is the vast differences in size, mass, composition, and orbit between two of the IAU’s eight "official" solar-system planets: Mercury and Jupiter. About the only thing they have in common is that they orbit a star and are roughly spherical due to hydrostatic equilibrium. Therefore, if both are considered planets, common sense dictates that any object that shares these characteristics should also be considered a planet.
So my definition is as follows (drum roll please):
"A planet is an object that formed from material in a disk around a star, whose shape is roughly spherical due to its self-gravity, and which never experienced or will experience nuclear fusion in its core." This would be the definition I'd tell the public. I'd come up with a more long-winded technical definition for the scientific community.
If you've been patient enough to stick with me so far, you're probably thinking, "By this definition, lots of moons are also planets." Correct. Remember that the two largest satellites in the solar system, Ganymede and Titan, are not only larger than Pluto, they're also larger than Mercury. The only difference is that these worlds directly orbit another planet, not the Sun.
To take location into account, I would subdivide planets into three categories: Primary planets orbit a star directly. Satellites (or "secondary planets") orbit a primary planet directly. Ejected (or unbound) planets were thrown out of their system and are now free-floating in interstellar space. But all of these objects are still "planets."
Further categories can be adopted to describe the physical nature of the planet. Gas giants, for example, would be large bodies such as Jupiter or Saturn whose mass is dominated by hydrogen and helium. Ice giants are large worlds such as Uranus and Neptune whose mass is dominated by ices. Terrestrial planets, or rock dwarfs, would be objects like Mercury, Venus, Earth, and Mars, whose relatively low masses are dominated by heavy elements. Ice dwarfs, such as Pluto and the larger trans-Neptunian objects, are small bodies whose masses are dominated by various forms of ice. New categories can be added for new types of extrasolar planets as they are discovered and characterized (hot Jupiters, for example). But just as dwarf stars are still stars, and dwarf galaxies are still galaxies, all of these objects are still "planets" as long as they meet the criteria described in the definition above.
I think my definition is self-consistent and flexible that is, it can easily accommodate new discoveries inside and outside the solar system and the rules are simple and easy to interpret in the vast majority of cases. A lot of researchers will object because they don't like the idea of basing the definition on formation. They would argue, correctly, that in certain cases, observations won't be able to distinguish whether, say, a 7-Jupiter-mass free-floating body formed from a collapsing gas cloud or inside a disk from which it was later ejected. While acknowledging that argument, I would counter that we can put these objects in their own category until future advances clarify their origin.
Others would say that my definition would mean too many solar-system planets, perhaps devaluing the entire concept of "planet." Such criticisms are also legitimate. But we don't ask students to memorize the name of every city, just the major ones. I would have students learn the eight official IAU planets, Pluto (for historical reasons), the major satellites (our Moon, Io, Europa, Ganymede, Callisto, Titan, and Triton), and any new very large bodies that astronomers find way out there in the trans-Neptunian region. That's not too many, is it?
I never said my definition was perfect, and I welcome comments and criticism. Fire away!