We have reanalyzed the 260 ksec XMM-Newton observation of 1E1207.4-5209. There are several significant improvements over previous work. Firstly, a much broader range of physically plausible spectral models was used. Secondly, we have utilized a more rigorous statistical analysis. The standard F-distribution was not employed, but rather the exact finite statistics F-distribution was determined by Monte-Carlo simulations. This approach was motivated by the recent work of Protassov et al. (2002) and Freeman et al. (1999). They demonstrated that the standard F-distribution is not even asymptotically correct when applied to assess the signficance of additional absorption features in a spectrum. With our improved analysis we do not find a third and fourth spectral feature (Bignami et al. (2003), De Luca et al. (2004)) in 1E1207.4-5209, but only the two broad absorption features previously reported (Sanwal et al. (2002), Hailey and Mori (2002)). Two additional statistical tests, one line model dependent and the other line model independent confirmed our modified F-test analysis. For all physically plausible continuum models where the weak residuals are strong enough to fit, the residuals occur at the instrument Au-M edge. As a sanity check we confirmed that the residuals are consistent in strength and position with the instrument Au-M residuals observed in 3C273.