Context :

Aims : We aim to use microlensing taking place in the lensed quasar QSO 2237 + 0305 to study the structure of the broad line region and measure the size of the region emitting the C IV and C III ] lines .

Methods : Based on 39 spectrophotometric monitoring data points obtained between Oct. 2004 and Dec. 2007 , we derived lightcurves for the C IV and C III ] emission lines .
We used three different techniques to analyse the microlensing signal .
Different components of the lines ( narrow , broad , and very broad ) were identified and studied .
We built a library of the simulated microlensing lightcurves that reproduce the signal observed in the continuum and in the lines provided only the source size is changed .
A Bayesian analysis scheme is then developed to derive the size of the various components of the BLR .

Results:1 .
The half-light radius of the region emitting the C IV line is found to be R _ { { C \textsc { IV } } } \sim 66 ^ { +110 } _ { -46 } light-days = 0.06 ^ { +0.09 } _ { -0.04 } pc = 1.7 ^ { +2.8 } _ { -1.1 } 10 ^ { 17 } cm ( at 68.3 % CI ) .
Similar values are obtained for C III ] .
Relative sizes of the carbon-line and V-band continuum emitting-regions are also derived with median values of R ^ { line } / R ^ { cont } in the range 4 to 29 , depending on the FWHM of the line component . 2 .
The size of the C IV emitting region agrees with the radius-luminosity relationship derived from reverberation mapping .
Using the virial theorem , we derive the mass of the black hole in QSO 2237 + 0305 to be M _ { BH } \sim 10 ^ { 8.3 \pm 0.3 } M _ { \sun } . 3 .
We find that the C IV and C III ] lines are produced in at least 2 spatially distinct regions , the most compact one giving rise to the broadest component of the line .
The broad and narrow line profiles are slightly different for C IV and C III ] . 4 .
Our analysis suggests a different structure for the C IV and Fe II+III emitting regions , with the latter produced in the inner part of the BLR or in a less extended emitting region than C IV .

Conclusions :