One of the largest sources of uncertainty in reconstructing the warming of the past century stems from problems with historical ocean temperature records. Inconsistencies in method or technology or gaps in observation caused by two world wars mean that long-term records of sea temperature need to be interpreted with care. Drawing on two historical sets of ocean temperature observations—one of the sea surface and the other of the upper 20 meters—Gouretski et al. found that the twentieth century saw a long-term trend of ocean warming. The two data sets were mostly collected independently, using different tools and techniques, and were subjected to different processing. The authors suggest that, owing to the distinct ways in which the data were gathered, the presence of a similar trend could mean that it is not a spurious finding.