A clustering in August and early September of fatal accidents, killing 417 passengers, has raised questions about air safety. However, there is no common thread to the accidents, other than human error in various degrees (see ASW, Sept. 19).

Despite the wreckage, Paul Hayes, director of safety for UK-based consultancy Airclaims, argues that a spate of accidents occurring in close succession may mask the longer-term trend toward improved safety. The distribution of accidents is not random, and the occurrence of accidents within a relatively short period of time, by itself, tells us nothing about air safety, he says.

On the other hand, the number of recent accidents that were preventable may be exceedingly relevant. Clearly, there was a breakdown somewhere in the checks and balances. As one might say, accidents are not accidental, and corrective actions will doubtless be warranted when the investigations are complete.

To be sure, Hayes' characterization of the number of deaths in the past 12 months as "unremarkable" is open to challenge by the bereaved, but he is talking about the total number of deaths in the context of losses over many years.

Here is his take on the recent spate of accidents:

"In the last 12 months there have been 19 accidents where passengers have been killed on revenue passenger flights. Perhaps surprisingly given the recent publicity about air crashes, this is almost exactly the number of losses predicted by the long-term trend. It is certainly not an exceptional number and would probably be considered a reasonable result if it were not for the small number of such accidents recorded for the 2003 and 2004 calendar years (see ASW, Jan. 5).

"As [the graph below] illustrates, it is more that the last two to three years have been exceptionally 'good' than the last 12 months have been exceptionally 'bad.'

"On a calendar year basis, so far in 2005 there have been 14 fatal accidents to passengers. It is entirely possible, but probably unlikely, that there will now be no more fatal accidents for the rest of the year. If this was to be the case, 2005, despite the bad experience of August, would still turn out to be a 'good' year. Not quite as good as 2004 and 2003 perhaps, but still better than any previous year and well below the trend. If accidents were evenly distributed, 14 accidents in nine months would still be on target for 19 accidents for the whole year. Once again, precisely the number predicted for 2005 by the long-term trend.

"As with the number of accidents, the number of passengers killed in the last 12 months, at 787, is also unremarkable and, again, is more or less what might be expected based on the long-term trend.

"There have been 659 passenger fatalities so far in 2005, with most of these occurring in August/September. This is certainly more than for the whole of 2004 and 2003, which saw 347 and 579 fatalities respectively but is still a good way below 2002's total of 782.

"The recent spate of accidents has put aviation safety firmly back on the agenda and has probably left many members of the public believing that, once again, airlines are becoming less safe. But, although some individual years may be better or worse than others, airline safety continues to improve." (Contact: Paul Hayes, e-mail: paul.hayes@airclaims.com)