In a recent paper, Houchmandzadeh [Phys. Rev. E 72, 061920 (2005)] introduce a correlated bigradient model in order to explain the robust scaling of the boundary of hunchback (hb) expression in the early Drosophila embryo. In particular, they stress that recent experiments by Lucchetta [Nature (London) 434, 1134 (2005)], where embryos whose anterior and posterior halves develop at different temperatures still show excellent precision in the hb boundary, are in good agreement with such a model. We would like to show here that this conclusion is unwarranted. This is because the experiments of Lucchetta were done at different temperatures from those studied in the model. Since in other temperature combinations the model does not produce precise boundaries and there are no systematic trends in these deviations, a comparison to the experiment is not possible. Furthermore, we would like to point out that any correlated bigradient model should also take into account the fluctuations of the bicoid profile within an embryo. When forming correlated bigradients of experimental profiles from an online library, we observe that these intraembryo variations destroy robustness of the hb boundary even in the wild-type situation.