Referee A has unconditionally approved publication. Referee B says:

This paper presents new data on the sideward flow in heavy-ion collisions in a particularly interesting energy range...... and justifies that these results be published in PRL. Before final approval there are, however, minor points which I would like the authors to consider.

The text of the paragraph at the bottom of page 3 is somewhat confusing. In the second sentence, the models may lead to different flow amplitude but those still correspond to the same observables, the observable being v1, F Fy, not their amplitude. The next sentence beginning "It has been argued...." tries to say too many things in the same sentence; I am not sure I follow its logic.

In the first paragraph on page 5, the author says that there are distortions below Pt = 0.3 GeV/c caused by track reconstruction inefficiencies and breakdowns of the pion-proton separation. Then they discuss the fact that the mean Px is independent of Py. I fail to see the relation. If tracks are not reconstructed, this will change the value of mean Px of the reconstructed tracks. Also if the sample includes pions which have a smaller flow than proton, this will reduce the measured flow amplitude. The authors should discuss their estimates of these systematic errors.

In the bottom paragraph on page 5, the authors should define y' the first time it appears on the second line of the paragraph.

Similarly, in the same paragraph at the end, it would be clearer if they would write "the steady decrease in the azimuthal anisotropy of the particle distribution v1 (defined below)."

In the sentence beginning by "an earlier version of BEM" on page 7 it seems that the arrow of time is going backward. It gives the impression that the predicted flow has changed going from the present version to an earlier version of the model. This can be improved.