The volcanism versus impact debate about the K-T boundary runs and runs, as newshounds tend to say. Things are not so evenly balanced for the biggest of all mass extinctions at the end of the Permian. Although signs have been reported, a link with an impacting extraterrestrial body has not convinced a decisive majority. On the other hand, there is a 1-2 Ma mismatch between the well-determined age (around 253 Ma) of the Siberian Traps and previous dates for the end of Permian stratigraphy in sections that have no depositional break with the Triassic. The extinction has all the hallmarks of a catastrophe, by definition a sudden event, so tying down its age and that of a plausible cause is essential. Not being able to do that for the K-T event and the Deccan Traps, and with uncertainties about the relationship of impact rocks to signs of extinction at the Chicxulub site, add fuel to that long-running debate. The accepted “golden spike” or GSSP for the Permian-Triassic boundary is at Meishan in eastern China, and there are other sites in China that run it close. The sections contain several volcanic ash layers, so zeroing in on a date for the extinction would seem straightforward, using U/Pb zircon dating. There is a problem. Some of the zircons in the ashes are xenocrysts rather than having formed during the various magmatic episodes, and they are microscopically indistinguishable from those that should give precise dates. All the zircons also show signs of having lost radiogenic lead during later alteration of the beds. The last could explain the mismatch with the Ar-Ar age of the Siberian Traps, the generally favoured culprits for the extinction. US and Australian geochemists have taken a new tack in dealing with these problems (Mundil, R et al. 2004. Age and timing of the Permian mass extinction: U/Pb dating of closed system zircons. Science, v. 305, p. 1760-1763). They have “aggressively” treated zircon grains to remove outer parts from which radiogenic lead has been lost, so leaving isotopically undisturbed cores of the grains. Their U/Pb data are mainly from a boundary section in central China (Shangsi), dating 8 separate ash layers, plus one from the boundary clay itself at the Meishan GSSP. The dates agree well with the stratigraphic sequence of the ashes, and hare high precision. Judging the actual age of the boundary at Shangsi relies on statistical analysis of the sequence of ages from the different ashes, and gives a date of 252.6±0.2 Ma. That is within error of the accepted Ar-Ar age of the Siberian Traps. As usual, this is not cut and dried, because there are other ages for the Siberian Traps, including one using the same U/Pb zircon method that suggests a 251.4 Ma age. Clearly the mismatches for the end-Permian events will be a meaty bone of contention, when all respected geochronologists turn up for a meeting early in 2005 to thrash out the conflicts that continually inflame their passions.