Revised answer summary
My original answer provided evidence that the trp repressor is a dimer, and that conclusion stands, especially as regards its functional form. I still regard the statement in Wikipedia that it is a tetramer as incorrect, although I now see how it might have arisen. It seems to be a misinterpretation of results in vitro that show an equilibrium can exist between dimer and tetramer (not tetramer only). I have expanded on this in an additional section.
A functional dimer
The functional form of the is certainly a dimer, as originally reported by Schievitz et al. (1985). The summary of this paper includes the statement:
“The crystal structure of the Escherichia coli trp repressor has been solved to atomic resolution. The dimeric protein has a remarkable subunit interface in which five of each subunit's six helices are interlinked.”
It was crystallized in combination with the operator DNA, and the structure deposited in the Protein Database (PDB) as 1TRO. I reproduce an image from the PDB entry below. Many other DNA-binding proteins also have dimeric structures.
A dimer–tetramer equilibrium observed in solution
The statement in Wikipedia probably arises from a report by Fernando and Royer (1992) — subsequently confirmed — that in solution there is an equilibrium between the dimer and higher oligomers (the tetramer predominating). Tryptophan was found to shift the equilibrium to the dimer. The authors claim that this has physiological relevance, although this still seems a moot point.