The TAR was, in my opinion, wrong to say anything about the precedence (or lack thereof) of the warmth of the individual year 1998.
The reason is that all reconstructions have very wide uncertainty ranges bracketing individual-year estimates of part temperature. Given this, it is hard to dismiss the possibility that individual years in the past did exceed the measured 1998 value. These errors on the individual years are so wide as to make any comparison with the 1998 measured value very problematic, especially when you consider that most reconstructions do not include it in their calibration range (curtailed predictor network in recent times) and the usual estimates of uncertainty calculated from calibration (or verification) residual variances would not provide a good estimate of the likely error
Similarly, it is likely7 that the 1990s have been the warmest decade and 1998 the warmest year of the millennium.