records often include accidents. i suggest this tells us something about the flexibility of musical practices and the limits of theories. musical ‘hooks’ provide useful test-cases because the hook is normally considered the least accidental part of a song. we imagine it emerging fully formed in a moment of inspiration – the catchy phrase that comes into a songwriter's head – or at least of calculation: ‘i’ll see your vibraphone and raise you a mellotron'. but hooks sometimes incorporate accidents or happen accidentally. if hooks are less than completely determinate, then every aspect of the popular record must be subject to contingency. i argue for theories – and views of musical practices – which begin with the recognition that accidents happen.