Vacation time has caught up before I have completed my remaining draft posts, or completed a more formal analysis of the differences between the three time series for all Gistemp stations and the “recent” and “dropped out” subsets. My gut feeling is that while statistical uncertainty with few stations may be sufficient to explain the difference for 1880-1900, the increasing number of stations from 1895 on and the consistency of this difference, with reducing variability, through to 1925 suggests that this difference is probably real. I probably will not be able to return to this analysis before August, so I’ve posted the three time series as an “aside” post for anyone who would like to examine this data in the meantime (and if you do, please add a comment).
If you post a comment which is held for moderation, please be patient. I hope to have WiFi access most of the time, and to be able to complete and post at least some of the draft posts, but while travelling (a 16-hour ferry crossing and a 1200 km drive) you may have to wait for moderation or a reply to any comments.