A simple model for describing two-point far field correlation patterns due to a statistically axisymmetric point quadrupole embedded in a plug flow is derived. The model is employed to interpret existing experimental data of jet noise far field correlations, which refer to the average phase angle of simultaneous emissions in different directions. Correlations with azimuthal or polar angle separation are discussed. It is shown that regions of negative correlations are expected, mostly due to the fact that, as pointed out by Lighthill, the radiating part of the quadrupole source term associated with incompressible turbulence should have zero trace. This property is used to model the source cross-correlation terms, which contain phase information, using the autocorrelation terms. It was verified that the experimental data, mostly obtained in the 1970’s and 1980’s for Mach number MJ = 0.4 and MJ = 0.75, are consistent with the model. For the lower Mach number case, the negative cross-correlations are fully supported by the experimental results; for MJ = 0.75, on the other hand, the strong non-compactness affecting the polar correlation data (which implies a significant loss of coherence), does not permit one to be fully confident with respect to a negative cross-correlation term, which is, however, supported by existing limited narrow-band data for MJ = 0.8. Comparison of the two main sets of data suggests that the source correlations vary with Mach number, perhaps also with shear. The model is also used to discuss the changes in the far field azimuthal correlation plots for an acoustically excited jet with MJ = 0.4. It is shown that these changes are compatible with a new source associated with a sequence of vortex rings which deform azimuthally as they travel downstream and that, due to the phase locking attained by excitation, maintain significant axial coherence which affects the directivity. Recent data are also briefly discussed, it being pointed out that some features of the sound field of non-excited jets which have been ascribed to strong source non-compactness can also be explained by mean flow effects acting on point quadrupolar sources. The importance of having both far field directivity (i.e., amplitude) and correlation (phase) data for a complete analysis is emphasised. The discussion also points out that usual hypotheses on the statistics of turbulence do not permit obtaining negative cross-correlations terms and, therefore, seem to be inadequate for jet noise modelling.