Unveiling the structure of the Broad Line Region (BLR) of AGNs is critical to understand the quasar phenomenon.
Resolving a few BLRs by optical interferometry will bring decisive information to confront, complement and calibrate
the reverberation mapping technique, seed of the mass-luminosity relation in quasars. BLRs are much smaller than the
angular resolution of the VLT and Keck interferometers and they can be resolved only by differential interferometry
very accurate measurements of differential visibility and phase. The latest yields the photocenter variation with λ, and constrains the size, position and velocity law of various regions of the BLR. AGNs are below the magnitude limit for
spectrally resolved interferometry set by currently available fringe trackers. A new “blind” observation method and a
data processing based on the accumulation of 2D Fourier power and cross spectra permitted us the first spectrally
resolved interferometric observation of a BLR, on the K=10 quasar 3C273. A careful bias analysis is still in progress, but
we report strong evidence that, as the baseline increases, the differential visibility decreases in the Paα line. Combined
with a differential phase certainly smaller than 3°, this yields an angular radius of the BLR larger than 0.4
milliarcseconds, or 1000 light days at the distance of 3C273, much larger than the reverberation mapping radius of 300
light days. Explaining the coexistence of these two different scales, and possibly structures and mechanisms, implies
very new insights about the BLR of 3C273.