We reexamine the statistics of discordant redshift galaxies in compact
groups. We find that 43 out of 100 groups in the Hickson catalog cont
ain at least one discordant redshift galaxy. We show that, despite the
prevailing impression, all previous attempts have failed to explain t
his large number of discordant redshift galaxies. The order of magnitu
de excess survives all of our attempts to refine the sample.