Mann has recently stated at [realclimate] see comments that:
No researchers in this field have ever, to our knowledge, "grafted the thermometer record onto" any reconstruction. It is somewhat disappointing to find this specious claim (which we usually find originating from industry-funded climate disinformation websites) appearing in this forum.
This is a strange statement, since Crowley did exactly that, and Mann changed the Crowley version between Mann et al. [EOS 2003] and Jones and Mann [Rev. Geophy. 2004] to use a less-splice Crowley version.
Disentangling Crowley versions is non-trivial. The original series ("version 0") (Figure 1 of Crowley and Lowery [2000]) was not archived at WDCP. Crowly archived two spliced versions at WDCP in conection with Crowley [2000]. One splice version ("version 1") contains a splice at 1965, with proxy values for a composite of 13 of 15 sites (I'll discuss this selection elsewhere) before 1965 and instrumental values from 1965 to 1998; after an 11-point smooth without padding, the spliced and smoothed series ends in 1993. A second version ("version 2") was spliced in 1870, with proxy values from the 13-site composite before 1870 and instrumental records afterwards. Figure 1 shows the difference between the two spliced versions, with the red line showing the additional splice in the version with the instrumental splice commencing in 1870. The blue series shows the 15-site composite (scaled through regression). Taking the 13-site subset changes the relationship of the MWP and 20th century - which is perhaps why the subset is shown rather than all 15 sites (although a different rationalization is provided.)
Figure 1. Crowley Versions. Black - Spliced version with instrumental values spliced after 1965; Red - splice between 1870 and 1965 for version CL2.Jns11. The black version is used before 1870 and after 1965 in this version. Blue - Original data (15-site composite) scaled onto black version by regression.
Version 2 with the splice at 1870 is used in the spaghetti diagram of Mann et al. [2003]. Version 1 with the splice at 1965 is used in the spaghetti diagram of Jones and Mann [2004]. This can only be seen through careful inspection of the diagrams blown up to 400% scale (since digital information for the not provide digital information for the spaghetti diagrams is not provided. The legend to Jones and Mann [2004] says the following:
The Crowley and Lowery [2000] series shown here replaces an incorrect version of the series shown in similar previous comparisons [e.g., Mann et al., 2003a].
Given that Mann had gone to the trouble of changing Crowley versions (with the effect of reducing the splice beginning from 1870 to 1965), it seems odd that he was unable to recall Crowley's grafting in response to the realclimate reader.