Astute readers of The Irish Times will have noticed that there was a striking resemblance between the Weather Eye column one day last week and a somewhat longer article, by the same author, in Saturday's Weekend. Indeed, they were more or less the same.
Gremlins arranged for a feature article on the poor summer to be plucked from the electronic maze, edited to near half its length, and published as a Weather Eye on Friday. Then to everyone's surprise, much the same material, albeit complete, unexpurgated and unabridged, appeared in full in Weekend the following day.
It was, to quote one of Napoleon's generals on the execution of a valued colleague, "worse than a crime; it was a blunder".
But meteorologists are philosophical when things go wrong. Sometimes, it is true, when they hear complaints ad nauseum about their forecasts, they begin to think, like William Cowper, that:
"Tis hard if all is false that I advance;
A fool must now and then be right - by chance."
But then they recover their aplomb and, to try to reassure themselves that their existence serves some useful purpose, they devise objective ways of measuring their skill.
For the very reason quoted by the poet, this is not as simple as it seems. To the man in the street, a forecast that is correct for him is always "good", but to a meteorologist, skill is only in evidence if the prediction is correct, as it were "against the odds".
There are no brownie points, for example, for predicting frost in February; the trick that month is to predict when it will not occur.
One method works like this. Imagine that in a certain place the climatological records show that on average it rains on 12 days in the month of May. A "forecast" based purely on climatological expectations can be devised by randomly selecting 12 dates from the month, and "forecasting" rain on those 12 days.
In due course, the weather forecaster produces his own prediction for each day of May, and at the end of the month, the results of both sets of forecasts are compared with the days in May that really did have rain.
A simple "skill score" is the ratio of the correct forecasts produced by the two procedures. If the forecaster, for instance, was correct on 25 days out of the 31, and the climatological prediction was correct on 15 days, the skill could be represented as 25/15, or 1.66. The greater the value of this ratio, the more skilful the scientific forecasts can be deemed to be.