We thank the referee for valuable comments and suggestions. We have responded to all comments, as detailed below. The resulting changes to the paper are marked in blue. We also note that our old Fig.17 was incorrect and has now been replaced. This led to modifications in the corresponding discussion on page 16, which is also marked in blue. > The paper is rather dense and hard to read, since many > variables and quantities are defined and used. To help the reader navigate through the paper, we have now added a new appendix A on pages 20 and 21, where we have added another table defining additional time scales and also a table summarizing various wave numbers. We refer to this appendix at the end of Sec.II.E, just after Eq.(35) on page 6. > 1. The particular value of the D5 coefficient in (5) is not addressed. We noticed that also the value of nu was not specified. We have now added a comment at the end of the paragraph after Eq.(9) on page 3 that we keep the ratio of viscosity to magnetic diffusivity, nu*sigma, i.e., the magnetic Prandtl number, unity in all cases. There we also now say that we chose nu/D5 to be unity in all cases. For simplicity, these ratio were never varied, but this could be done in future. > 2. Perhaps naively, I was thinking of a kinetic energy spectrum E_K(k) > increasing with the wave number. From Fig. 2 I can see this is not a > monotonically increasing function of k. What am I missing in the > definition of E_K(k)? After Eq.(15) on page 4, we wrote that \int E_K(k) dk is the kinetic energy, so a monotonically increasing function of k would diverge at large k, so the spectrum must go down. In the second paragraph of III.B on pages 7+8, we now discuss the evolution of E_K(k) in Fig. 2 in more detail. > 3. In Fig.3 for eta=0.3 and k=10 it looks like that the positive > helicity component is larger than the magnetic energy, which seems not > in accordance with (16). Thanks for pointing this out. We have now checked that this problem originates from the discretization error when taking the curl of A. Discretization errors are generally low for high-order codes, but they always exist and show up when comparing A.B with |B|^2 at high wave numbers. A.B is more accurate than |B|^2, which can be underestimated by up to 50% near the Nyquist wavenumber. This is usually not a problem, because it only shows up at very early times. We have now shown two later times instead where the problem is not visible. The present paper is not the right place to discuss this unphysical technicality. Therefore, we have now explained and discussed this in the Pencil Code Newsletter 2023/2; see https://github.com/pencil-code/website/tree/master/NewsLetters, where we have verified that the error becomes smaller when using a 10th-order scheme (about 33%), and much larger when using a 2nd order scheme (230%). > 4. In Fig.4 the decay of at eta > eta_I seems to follow a power > law decay. Is its numerical exponent close to -2/3, as indicated in > (28)? If so, it could be mentioned in the text. Yes, we have done this now and it matches well the eta^{-2/3} prediction; see Fig.4 on page 8 and the blue text. Thanks for this suggestion. > 5. In the caption of Fig.5 it is said that a scaling of eta**-1/3 > describes the spectrum better than eta**-4/9. Some runs also present > this exponent in Table II. This seems to invalidate the expectations > in (27). How can this be understood theoretically? We note that this was commented upon in the text, where we wrote "It also has a shallower scaling of the correlation length, xi_M ~ k_I ~ eta^{1/3}, which seems to be an artifact caused by insufficient scale separation, i.e., the value of k_1 is not sufficiently small." This is on the right-hand column of page 9 in the paragraph *before* the one that starts with "The evolution of the peaks of the...". We discussed this on page 10, right column, in paragraph 2 of Sec.III.C, where we wrote "The reason for this is that for large values of eta_CPI, it became necessary to decrease the value of k_1. This decreased the Nyquist wave number since N remained unchanged, which can cause artifacts in the values of k_+/-. Small values of k_1 also facilitates the eta^{4/9} scaling of xi_M and related length scales; see the comparison between Runs~N and N' in Table II." > 6. In Fig.8 the authors explain how eta_{mu_M}^+ is estimated but not > eta_{mu_M}^-. Is it exactly the maximum position of the negative > magnetic helicity evolution? Yes, eta_{mu_M}^- is based on the maximum of the negative magnetic helicity spectrum. This is now said at the end of the first paragraph of Sec.III.D on page 11, lower left column. > 7. For the case with spin flipping processes, is Gamma in Table 2 > equal to Gamma_{f0} in (21)? Yes, we have now replaced Gamma in Table 2 and several other places by Gamma_{f0}. > 8. Why did the authors only consider one simulation with Gamma > different from zero? Is it perhaps not interesting a case with Gamma > finite and |mu_{50}| > k_0 ? Yes, the idea was to examine explicitly the case where ACC and chirality flipping take place at the same time. The parameter for Run F is chosen in such a way. In the case |mu_{50}| > k_0, the ACC washes out the chirality and helicity already at the time of turning on the chirality flipping interaction, and hence such a case is not so interesting. We have now added an explanation on this issue at the end of the first paragraph of Sec. III G on pages 12+13, where we now also refer to Appendix B on page 20 for corresponding details. > 9. For the evolution of in run F of Fig. 11 the authors say > that for eta>eta_flip the value of declines at a much smaller > rate, but this is not evident from the plot (the rate seems similar > before and after eta_flip). Can the authors superimpose the result of > a similar run but with Gamma=0 to actually see the difference in the > decrease of ? We have now overplotted Run N, which is the same as Run~F, but without spin flipping; see Fig.11 on page 12. We have now referred to this also in the text just after the place mentioned by the referee. > 10. Is the run F with a finite eta_off physically relevant, or is it > made for illustration? In other works, have the authors in mind some > physical scenario in which the spin flipping processes are only > transient? Yes, we introduced the turning off of the spin flipping mainly for illustration to understand the physics. However, a sudden change of spin-flipping rate is realized, e.g., at the time of right-handed neutrino decay, which is studied in the context of the wash-in leptogenesis (though the dynamics is not a simple turning off). It would be also possible to construct a model of new physics where the smooth turn-off is realized. Going to the technical details is beyond the scope of the paper, and we briefly explained it at the end of Sec. II C on page 5, upper left column.