Using four well-defined rules, Bahcall and Goldshith (1971, Astrophys.J., 170, 17) (BG) have proposed eight absorption redshift systems for the QSO 4C 05.34. These systems identify 47 lines out the 93 observed. BG have also analysed 10 nonsense spectra; they find that for the 'average' nonsense spectrum, the number of acceptable redshifts is only 1.4, and the total number of line identified is only 7. It is pointed out that BG have missed out a vital point, namely, the density distribution of lines, in generating their nonsense spectra. Following the basic ideas of Russel and Bowen (1929, Astrophys.J. 69, 196), ten nonsense spectra are generated which are similar to the real spectrum in all its statistical characteristic features. Such spectra are named ghost spectra. These are analysed using the rules of BG, and the results are as follow. For the 'average' ghost spectrum, the number of acceptable redshifts is 8.4 ± 0.92, and the total number of lines identified is 35.5 ± 5.39. The problem of multiple identifications for an observed line is discussed, and the results obtained from the chance-coincidence hypothesis are compared to the actual ones. It is concluded that the number and properties of the absorption redshift systems proposed by BG for 4C 05.34 are insignificantly different from those that would be expected from chance coincidences. The foregoing results further strengthen the suggestion that there is no redshift in QSOs (Varshni, 1974)