Dr Freyssinier from Lighting Research Center (LCR) of Rensselaer Polytechnic Institute presented at SIL 2012 the last experiment covering tint perception along all CCTs, a long needed study. Although results has been in the past suspected by experts, this is the first time it is empirically confirmed on a scientific study:
"Pure White" perceived by people, meaning light has no tint (hue), is not on the blackbody locus nor daylight phases locus (although the latter is closer and almost exact from 4100 to 5000K).
The study was performed with 20 persons (likely, more color research groups will replicate it soon) and found that when presented a white stand illuminated with 7 lights for each CCT (6 in total), answers clearly pointed out they perceive as untinted white not lights with chromaticities off the planckian locus.
For the three cooler CCTs analyzed, 6500K, 5000K and 4100K, untinted white was perceived over the BB locus, thus for people the BB locus has a purple tint at these CCTs. Not huge, untinted white selected was always the first option over the BB locus (the closer to it, but BB itself was perceived as tinted).
But for the 3 warmer CCTs (3500K, 3000K and 2700K), untinted white was perceived below the BB locus, meaning BB locus is perceived as with a green tint. So really, using BB locus or, in the practice, incandescents, as reference for untinted white has no any support, and poses and interesting dilemma about references used for color rendering metrics.
The page at LCR explaining it, with more links at the bottom: http://www.lrc.rpi.edu/programs/solidstate/assist/whitelight.asp
This is the presentation used at SIL: http://www.lrc.rpi.edu/programs/solidstate/assist/pdf/SIL-2012_FreyssinierRea_WhiteLighting.pdf
And LEDs magazine article talking about this at SIL 2012 report