CIE 15:2004 names the blue region as the site of CIELAB's residual non-uniformity, and states the CIEDE2000 rotation term exists to correct it

Section 8.3.1 of CIE 15:2004, "CIEDE2000 total colour difference formula," states plainly that the correction exists because CIELAB is non-uniform: "The CIEDE2000 total colour difference formula corrects for the non-uniformity of the CIELAB colour space for small colour differences under reference conditions." It then identifies the mechanism and its location: "Visual colour-difference perception data show an interaction between chroma difference and hue difference in the blue region that is observed as a tilt of the major axis of a colour-difference ellipsoid from the direction of constant hue angle. To account for this effect, a rotation function is applied to weighted hue and chroma differences." The report then gives the rotation term itself (R_T, built from a hue-rotation function Δθ) as Equations 8.43-8.45.

This is the primary-source confirmation claim-cielab-lacks-perceptual-uniformity-in-blue-hues needed: not a Wikipedia paraphrase but the CIE's own technical report naming the blue region as the location of a specific, documented perceptual distortion — a hue/chroma-difference interaction tilting the color-difference ellipsoid off the constant-hue direction — addressed by a dedicated correction term in the CIE's own recommended formula. An independent peer-reviewed implementation-notes paper supplies the specific hue-angle figure for the correction's centre (~275°, the constant of the formula's own Δθ function), though it is mathematical-only and does not second the perceptual data: see claim-ciede2000-blue-hue-rotation-localized-near-275-degrees. entity-ciede2000 collects the formula's other appearances across the vault.