Iron atoms were formed in an all-glass shock tube from mixtures of Fe(CO) 5 in Ar. Observations of the reflected shock region have led to the following gf-values (within ±20 per cent) for selected low-excitation lines of multiplet 13: 1·05 × 10 -4 for the line at 6280·620 A, 1·70 × 10 -4 at 6358·690 A, and 6·70 × 10 -5 at 6498·950 A. The following results (within ±30 per cent) were obtained for gf-values of high-excitation lines belonging to multiplets 1032, 843 and 928, respectively: 1·05 × 10 -1 for the line at 5187·92 A, 3·00 × 10 -1 at 5242·50 A, and 1·10 × 10 -1 at 5379·58 A. The new measurements correspond to an increase of the more recently accepted f-number scale by about a factor of two. Curves of growth for the strong, high-excitation lines have been used to determine half-widths for collision broadening of Fe by the gas mixture existing under equilibrium conditions. The most frequent collision partners are argon atoms and electrons. At room temperature, the broadening coefficient is 0·15 cm -1 at 1 atm, which is about five times larger than the values normally observed in nonreactive collisions, but is appreciably smaller than the line-widths expected for hydrogen-atom broadening of iron lines. Particular care was exercised in these investigations in order to define the gas composition in the reflected-shock region. Large systematic errors were encountered when a conventional static mixing procedure was used in place of a dynamic procedure, which is required for adequate composition control of strongly-adsorbed gases [Fe(CO) 5] in a large excess of inert diluent (Ar). The consistency of our observations was demonstrated by showing that, although measured intensity-time data were strongly affected, final estimates for gf-values were not influenced by the following systematic variations in experimental technique: changes in the mole ratio of Fe(CO) 5 to Ar produced by temperature alterations in the liquid Fe(CO) 5-bath which served as an equilibrium source for Fe(CO) 5; changes in the flow rate of Ar through the isothermal Fe(CO) 5 supply source; changes in the volume of the liquid Fe(CO) 5 supply source; admixture of excess CO to the Ar which served as carrier gas for Fe(CO) 5; changes in the slit functions; changes in the concentration of the dominant collision partners for the iron atoms.