IT WAS AN embarrassing moment, probably the most embarrassing moment experienced by the Australian Bureau of Statistics this century. On a Friday in December 2004, the bureau was forced to reissue a report it had published two days earlier. In deceptively bureaucratic language, the revised version began thus: “Note the following change has been made to this document on 10/12/2004. Original text, published on 8/12/2004: ‘In 2003 there were 106,400 marriages registered in Australia, an increase of 3300 (3 per cent) when compared with 2002, and the first increase in registered marriages since 1999.’ “Correction on 10/12/2004: ‘In 2003 there were 106,400 marriages registered in Australia, an increase of 960 when compared with 2002, and continuing the increase in the number of marriages since the low of 103,130 in 2001.’ “