The origin of several chapters as separate articles is clearly evident in the text, which repeats the same ideas over and over from very similar angles. But it's nonetheless a very interesting book, which takes a number of fairly simple case studies of (mostly structural) engineering failure, and illustrates how they are examples of wider tendencies towards error. My personal favourites were examples from distant history, Galileo and Vitruvius, perhaps because they were less familiar than some of the more recent accounts of bridge collapse.

The final chapter, which was written in 1994, queried whether major bridge failures run in roughly 30-year cycles (Dee, Tay, Quebec, Tacoma and Milford Haven all fitting the pattern) and wondered whether another was due in 2000. Although Petroski suggested that cable-stayed bridges might be the form that would see the next failure, the more obvious example in retrospect is that of the Millennium Footbridge in London. Eerily, even Petroski's predicted probable cause of failure - "an instability" - came true!

This also ties in with an earlier chapter, where Petroski noted that the design climate at the time of the Tacoma Narrows failure "evinced confidence in analytical techniques and a preoccupation with aesthetics". The same was very much true at the time of the Millennium Footbrige design.

Thought-provoking, especially so for bridge engineers, and let down only by the repetition. ( )