Briffa’s temperature reconstruction from Tornetrask (northern Sweden) tree rings is a staple of multiproxy studies, used in Bradley and Jones , Hughes and Diaz , Jones et al. , MBH98, MBH99, Briffa and Crowley and Lowery , Briffa et al. , Bradley, Hughes and Diaz , Mann and Jones , Jones and Mann . Briffa makes an ad hoc "adjustment" to the MXD chronology which has a dramatic impact on the relation of 20th century and medieval levels of the chronology, which then affects all downstream multiproxy studies.
The Tornetrask data was discussed in three publications: Nature(1990), Clim. Dyn. (1992) and Climatic Variations (1996) ("NATO"). Pertinent archived data at WDCP is under series swed019w – ring width and swed019x – maximum density. Esper et al.  uses Tornetrask, but appears to have used independently calculated chronology. Briffa and Osborn  use the Tornetrask ring width series; Briffa et al.  provide no information on sites used and it is unknown whether the matters discussed here apply to this study.
Tornetrask MXD Chronologies
Figure 1 below shows the following Tornetrask maximum density (MXD) chronologies (my emulations of Briffa’s unarchived results using WDCP dataset swed019x archived by Schweingruber): top – mean MXD; middle – MXD chronology using RCS method [Clim. Dyn. 1992]; bottom – MXD chronology using spline method [Nature, 1990]. Figure 1 replicates Clim. Dyn. 1992 Figure 5 to observable detail. Obviously, 20th century levels are not elevated.
FIGURE 1. Tornestrask 65 tree MXD data: top) mean; b) RCS chronology; c) 67% spline chronology.
Low 20th century levels are inconsistent with Hockey Team policy, so Briffa et al. "adjusted" the MXD chronology. The "adjustment" is described in Clim. Dyn. 1992 (see Figure 7), but there is no illustration of the eventual impact on the MXD chronology. I’ve attempted for some time to replicate this "adjustment", which I describe here.
Briffa et al. justify an "adjustment" to MXD data because of a supposedly changing relationship between ring width (RW) and MXD in the past two centuries. The top panel of Figure 2 below shows Tornetrask chronologies – RW in red, MXD in black. Briffa regressed MXD (smoothed 10-year version) against RW (smoothed 10-year version) for 500-1750 and used the model to calculate residuals the entire period (shown in the 2nd panel, smoothed with a 25-year gaussian filter). Briffa et al. calculated the trend of density against RW since 1750 (shown as a red line in the 2nd panel). I find it hard to ascribe particular significance to this "trend" as compared to earlier "trends". Briffa then adjusted the post-1750 residuals for the MXD vs RW regression by this trend, as shown in the 2nd panel (which is scaled to sd units of the residuals).
The third and fourth panels show the impact of this "adjustment" on the "adjusted" MXD chronology. The third panel shows the unadjusted (black) and adjusted (red) MXD chronology, here centered on the 500-1980 mean of the unadjusted chronology. The fourth panel re-standardizes to a 1902-1980 zero (as used in MBH98-99), with a further 25-year smooth, evidencing the profound change in relationship between 20th century and earlier values.
Briffa purported to justify the "adjustment" by showing that the "adjustment" increased the R2 of the reconstruction from 0.503 to 0.555 (see Clim. Dyn. Table 2, columns 2 and 3). His regressions do not demonstrate any attempt to examine significance of any coefficients – I’ve replicated numbers that look somewhat like Table 2, but, when I do so, the t-values for all coefficients other than MXD lack significance. I have not been able to replicate the very slight gain in R2 – obtaining a gain of less than 0.01. While the "adjustment" is offensive because of its ad hoc-ness, at a minimum, it reduces the number of degrees of freedom being modeled and Briffa should show some sort of F-test being met, which naturally is not done. I’m not exactly sure right now how you would properly model this sort of bizarre ad-hockery in a F-test, but intuitively the gain in R2 seems too small to pass any sort of test. (In passing, Briffa and Jones and are both co-authors of Rutherford et al, , with its diatribe against R2 statistics.)
Briffa has not archived any chronologies. only his temperature reconstruction. Measuements (RW and MXD) are available for a 65-tree set at WDCP (swed019w and swed019x). These were used in the Nature 1990 article. The CD92 article used larger 425-core set for the RW chronology (but stayed with the 65-core MXD chronology). No measurement or RW information is archived for the 425-core chronology. In the absence of information on the larger dataset, I calculated a RW chronology using the 65-core set, and visually compared the plot to the illustration of the 425-core RW chronology in CD92, and the differences do not appear to be material to the effect. The RW chronology is used only to calculate residuals and the panel shown here replicates Figure 7 of CD92 to all material detail. (I can’t think of a good reason why a valid effect should not be present in the RW chronology of the 65 cores in question anyway.)
Figure 2. Top panel: Tornetrask. RW (red) and MXD (black) chronologies. Second panel (compare to CD92 Figure 7): Black – residuals after regressing MXD against RW based on 500-1750; red line – "trend" of residuals after 1750; red – "adjusted" residuals. Third panel – "adjusted" MXD chronology after adding back "adjusted" residuals. Bottom panel – MXD chronologies with 1902-1980 zero (additional 25-year smooth). Black – unadjusted; red – "adjusted". Note the change in 20th century relative to other periods.
A small curiousity in Briffa’s temperature reconstructions is the presence of a 1980 value. His regression estimate for the temperature in year N include the following year (N+1) ‘s tree ring values. Since the tree ring chronologies end in 1980, how does he get a 1980 temperature reconstruction value? The upspike at the end of the temperature reconstruction (Nature 1990, Figure 2a) is not present in my emulation. What seems to have happened (and I noticed this because I did it in a first run) is that some software picks up the year 1 (i.e. AD500) value and inserts it for the missing value. It looks for sure that the NA value for 1980 was overwritten by a high value from AD500. Not much turns on this latter error, but the impact of the first "adjustment" looks like it may be quite important in the very non-robust MWP calculations of Jones et al.  and perhaps others.
UPDATE: Here is the text of the Clim Dyn. "adjustment" justification in full (p.116):
"Up until about 1800, the April-August RCS-based reconstruction (primarily dependent on density data, as is evident in Table 2) is consistently warmer (by ~0.25 deg. C) than the RCS ring-width reconstruction for July-August. The density chronology (Fig. 5b) shows a low-frequency decline over the last century which appears anomalous in comparison with both the RW data and the instrumental data over the nineteenth and twentieth centuries. These facts suggest that the density coefficients in the regression equation may be biased as would be the case if the density decline were not climate related (CO2 increases and/or the potential effects of increasing nitrogen input from remote sources, known to have occurred over the present century, may be implicated here.)
We examined the magnitude and timing of the recent MXD decline by regressing the 10-year smoothed RCS density curve against the equivalent RCS ring-width curve over the period 500-1750. The regression equation explains just under 35% of the MXD variance. Using this equation, we estimated MXD values for 501-1980. The residual MXD data (actual – estimated) are plotted in Fig 7. [this corresponds very closely to my 2nd panel] A systematic decline is apparent after 1750. By fitting a straight line through these residuals (1750-1980) and adding the straight line values (with the sign reversed) to the RCS density curve, the anomalous post-1750 decline was removed (cf. Fig. 7). This ‘corrected’ RCS curve was then used along with the RCS ring width curve in a final reconstruction of April-August temperature. The calibration of this reconstruction now explains 55 per cent of the instrumental variance (cf. Table 2 [where the 'uncorrected' reconstruction shows calibrated variance (R2) of 50.3%]. The improvement supports our contention that the anomalous recent density trend was not climate related."