We present evidence in the Schwinger model that rooted staggered fermions may correctly describe the m < 0 sector of a theory with an odd number of flavors. We point out that in QCD-type theories with a complex-valued quark mass every non-chiral action essentially "borrows" knowledge about the θ-transformation properties from the overlap action.The staggered fermion operator D st describes N t = 2 d/2 flavors ("tastes") of quarks in d dimensions. In order to obtain a single quark flavor in dynamical fermion simulations, it has become customary to include the N t -th root of the staggered determinant as a weight factor.In a recent preprint [1] Creutz has argued that this procedure makes it difficult to obtain the right continuum limit. His argument is based on the observation that the staggered determinant is strictly a function of m 2 , whereas the determinant of a true single flavor theory also contains odd powers of m, arising from the unpaired chiral modes of the Dirac operator.In a reply [2] Bernard et al. have noted that the rooted determinant always corresponds to a theory with positive fermion mass, regardless of the sign of the staggered mass term. In addition, they point out that rooting is a non-analytic procedure. Accordingly, they discuss how taking the continuum limit first can lead to odd powers of |m|.In this note we want to address how two observables, the chiral condensate and the topological susceptibility, behave for non-positive quark masses. We demonstrate in the Schwinger model [3] that even with rooted staggered fermions these quantities can be measured for m < 0, by simply reinserting the phase information which gets lost by starting with a doubled action and taking the square-root. This "recipe" can be justified by reference to the overlap action, but it is beyond the realm of a purely Lagrangian approach with staggered fermions.The numerical data presented below are based on a sample of 1000 gauge configurations at β = 9.8 on a N ×N = 28 2 lattice. We compare (rooted) staggered to overlap fermions [4], both actions defined with one step of APE smearing. For details see [5,6]. Throughout, we use the coupling e to define the physical scale and set the lattice spacing to a = 1.First, we consider the scalar condensate. We are interested in its continuum limit in a fixed physical volume (cf. [6]), and for N f = 1 its presence is due to the anomaly [3] and does thus not signal spontaneous symmetry breaking. For overlap fermions and m > 0 its bare version is(1) where the sum runs over all eigenvalues of the massless operator D ov , except for the doublers (on topologically non-trivial backgrounds) at 2ρ. Here, ρ is a parameter in the overlap construction which in our runs is set to 1. The measure in (1) 1