Meteorologists have precise scales that measure the intensity of almost all the phenomena in which they are interested - except, it seems, the size of hailstones.

"Hailstones the size of golf balls" was a common phrase in reports of the Selsey tornado last week. So common, in fact, as to suggest that the golf ball has become the standard unit with which to measure hailstones. Or it may be just a journalistic rule of thumb to decide whether a hailstorm is worth reporting; hailstones the size of golf balls are news, hailstones smaller than golf balls are not.

Before continuing with the linguistic analysis, however, it is worth considering how hailstones form, and why they are so often associated with tornadoes. The idea of globules of ice falling from the sky is nothing unusual. Most raindrops start as frozen droplets, which melt on their way down. When there is a strong updraught in the cloud that contains them, however (which is generally true to some extent because of the warm air rising from the earth's surface), droplets of ice will fall a little, then be lifted, allowing more water droplets to freeze around them, then fall again under the extra weight, then perhaps rise again, acquire more ice, and finally fall as hail when they get too heavy even for a strong updraught to support. Tornadoes have the strongest updraughts of all, so tend to produce the biggest hailstones.

But how big? The largest hailstone ever verified weighed 1 kilogram, and fell on 14 April 1986 in Bangladesh. The largest in America weighed 766gm and was 19cm (7.5in) in diameter. They kept that one in a freezer until scientists had verified it as a true hailstone. The report of a hailstone the size of an elephant falling on Seringapatam in the reign of Tipu, Sultan of Mysore, in the late 18th century is either an exaggeration, or it was a very small elephant.

The expression, "hailstones the size of ... " occurs frequently in accounts of storms. Our newspaper database records 47 occurrences, of which "hailstones the size of golf balls" is by far the most frequent, occurring 19 times. The next most frequent are tennis balls and cricket balls (six each), followed by marbles (five, including one occurrence of "large marbles") and pigeon eggs (three). The remainder comprise single occurrences of hailstones the size of basketballs, hubcaps, gobstoppers, hens' eggs, walnuts, melons, onions and footballs, though the last of these occurred only in a work of fiction.

In such a normally precise field as meteorological measurement, this list is disappointingly vague. While "hailstones the size of onions" may have been deemed an appropriate description for a storm that hit St Tropez in 1993, are we to assume they were large Spanish onions, or little more than shallots? Is an onion smaller or larger than a golf ball? And were the "hailstones the size of melons" the size of watermelons or Ogen melons?

In many of these expressions, the hardness or, in the case of onions, the layers structure of hailstones, is also suggested, but the meteorometrical world is clearly in great need of an official scale for hailstone measurement.

A golf ball, incidentally, must have a diameter of not less than 1.68in. There seems to be no internationally agreed definition of gobstopper size.