We investigate the dynamics of stock clustering in the Johannesburg Stock Exchange (JSE), an emerging market, through the financial market crash of 2008. In particular we apply the fully unsupervised parameter free data clustering technique pioneered by Giada and Marsili (2002) to investigate the changing correlation structure of stocks, as well as clustering in daily market-wide activity, in a crisis. We compare our findings with an identical analysis of the London Stock Exchange through the same crisis period.