Last year we discussed the work of Richard Lenski, who has been breeding E. coli for 21 years in a laboratory in Michigan. Then, the news was that Lenski's lab had caught direct, reproducible evidence of a genetic mutation with functional consequences for an organism. Now Lenski's lab has published in Nature a major study comparing adaptive and random genetic changes in 40,000 generations of E. coli (abstract here). "Early changes in the bacteria appeared to be largely adaptive, helping them be more successful in their environment. 'The genome was evolving along at a surprisingly constant rate, even as the adaptation of the bacteria slowed down,' [Lenski] noted. 'But then suddenly the mutation rate jumped way up, and a new dynamic relationship was established.' By generation 20,000, for example, the group found that some 45 genetic mutations had occurred, but 6,000 generations later a genetic mutation in the metabolism arose and sparked a rapid increase in the number of mutations so that by generation 40,000, some 653 mutations had occurred. Unlike the earlier changes, many of these later mutations appeared to be more random and neutral. The long-awaited findings show that calculating rates and types of evolutionary change may be even more difficult to do without a rich data set."