Thank you for your thoughtful and fairly comprehensive responses to reviewers' comments. I believe your revised paper is much improved and will be of significant interest to the hydrologic community. In reading all the reviews, your responses and the revised manuscript, however, I have two concerns that I believe you can address fairly easily. (1) The statement on p. 19 line 9: "This does not imply that physical dispersion does not exist, only that physical dispersion is accounted for by numerical dispersion." is problematic because you do not even assert that the magnitude of numerical dispersion might similar to that of the sub-grid scale (200 m) dispersion. It sounds like you hope that the numerical dispersion will mimic the real thing, but you do not really know. Unless you have some evidence that the numerical dispersion might fortuitously mimic the real thing, I suggest you remove this sentence and discussion of sub-200-m scale dispersion, and rely on the argument that your resolution of 200-m scale heterogeneity accounts for the dominant dispersion on the scale of this problem and leave it at that. You could cite Weissmann et al. (2002) on this, as well as LaBolle & Fogg (2001) which is cited by Weissmann. The obvious fix would have been to use the more accurate MOC or TVD solution schemes in MT3DMS, and apply actual grid-scale dispersivities, rather than hoping numerical dispersion will cover it for you. If those methods created some other problems, such as excessive execution times, making them impractical, you should say so in the methods. (2) As suggested by reviewer 2, the low clay porosities that you estimated will look quite implausible to most reviewers. A true clay will indeed always have higher porosity than a sand, unless perhaps the clay has been heavily lithified, say by compaction due to glacial loading, in which case they may be more like claystone formations than clay. Another possibility is that the clays have heterogeneities within, possibly including fractures, that provide preferential flow paths that would increase the apparent effective porosity. So if it really is plausible for the clay effective porosities to be that low, you should explain it through geologic arguments concerning the actual nature of those clays. My first thought, however, was that the calibration was forced to dramatically lower the clay porosity to compensate for too-high K values from your flow model calibration. The latter is also quite plausible because calibrating aquitard K values in a flow model is tough unless you have really good 3D head data. At this point, I just suggest you give one or two geologically plausible explanations for how the clay porosity could be so low.

Congratulations on publishing your manuscript!