This paper introduces for the first time—and computes explicitely for various cases—the two-photon correlation spectrum (or "two-photon spectrum" as per the nomenclature of the paper), that is, the 2D landcape of photon correlations when retaining their frequency degree of freedom.
We made a "video abstract" to explain the main idea:
This is, for instance, the 2PS (two-photon spectrum) for the Mollow triplet:
We would understand the triplet of red antidiagonal lines right away (they are leapfrog processes) but it would take another decade for us to figure out the blue circles!
The two-photon spectrum has been measured experimentally for the first time by Peiris et al., which is the most interesting case, and shortly after that by Silva et al. but for something with much less quantum structure.
Such 2D structures remain ignored by the bulk of quantum opticians, who fail to understand that to look at two-photon observables, one must look at two-photon spectra. In this way, peopled remained oblivious to the strongly-correlated emission away from the spectral peaks that, if you Purcell-enhance it, gives rise to a new regime of quantum emission.