In an experiment performed some time ago, it was reported that the photon statistics of the 7th harmonic radiation of the MARK III free-electron laser (FEL) was sub-Poissonian. If FEL light exhibits such non-standard behavior, our understanding of the FEL needs to be radically modified. In this talk, I will present a re-examination of the data analysis of that experiment. We find that the data could be explained within the standard FEL theory if one combines the detector dead time effect with photon clustering arising from the FEL gain, while it seems unlikely that the photon statistics for an FEL oscillator starting from the noise could be sub-Poissonian according to the standard FEL theory. We propose an improved experiment for more definitive measurement of the FEL photon statistics.