Evolution. In the article “Intron positions correlate with module boundaries in ancient proteins” by Sandro Jose de Souza, Manyuan Long, Lloyd Schoenbach, Scott William Roy, and Walter Gilbert, which appeared in number 25, December 10, 1996, of Proc. Natl. Acad. Sci. USA (93, 14632–14636), the authors request that the following be noted. In analyzing a new intron database from Genbank 96, we found that mistakes had been made in the positions of some introns (31 small errors in 570 positions). Some of these mistakes affected the statistics for Tables 2 and 3. In Table 2, which shows the statistics for the whole data set at 28 Å, there are now 214 intron positions found in the linker regions versus 182 expected, a 17% excess, with P = 0.005. In the set of 20 putative “old” introns, 13 are in the linker regions versus the 6.4 expected, a 100% excess with P = 0.0015. The curves in Fig. 3 and the general argument are not affected.

On redoing the calculation with the larger data set of intron positions, 662, the 28-Å analysis yields a 17% excess with P = 0.002 and a larger set of “old” introns (37 positions) has a 70% excess with a P = 0.004. The peak at 21 Å reaches a χ² of 19, whereas the 28- and 33-Å peaks have a χ² of 15 and 13, respectively.