|The authors’ changes have addressed most of my concerns in the first draft. The additions starting at line 275, in particular, help clarify the scope of the article. Nevertheless, the discussion of historical causation could still use improvements. In particular, the authors face a central problem that the correlation between the timing of eruptions and the timing of uprisings in Ptolemaic Egypt fits two scenarios equally well: (1) the occurrence (rather than non-occurrence) of eruption-induced droughts was necessary (and/or sufficient) for the occurrence (rather than non-occurrence) of uprisings, or (2) the timing of eruption-induced droughts in some years (rather than droughts occurring in other years) was necessary (and/or sufficient) for the timing of uprisings in some years (rather than uprisings that would have taken place anyway in other years). This is not a fatal problem with the article. However, it needs to be openly admitted and addressed. The authors do not need to determine which scenario was the case, and they may clarify how the findings of this study add some credibility to scenario (1). However, they need to acknowledge that both scenarios remain possible and that some historical evidence points to (1) while other historical evidence and circumstances point to (2). |
I would particularly encourage revisions in two sections:
The authors’ discussion of causation and correlation starting at line 183 is confusing. It is true that historians frequently resort to the platitude “correlation isn’t causation.” That platitude often applies to situations in conventional history, where a correlation between phenomena A and B might be explained away by some set of factors (C, D, E, etc.) that influence both A and B, rather than any connection between A and B themselves. However, in much climate history, where there are no hidden variables influencing, e.g., both social unrest and volcanic eruptions, that problem doesn’t apply and we need to drop the platitude altogether.
In this study would be more accurate to say that this correlation does imply causation—yet the nature of that causal linkage remains unclear. Absent further research and clarification, we cannot say, for example, whether these uprisings might have occurred without the eruptions but perhaps in a different manner or in a different year. Moreover, even if the eruptions and Nile failures were a necessary condition for the uprisings, we would need further information and arguments to determine whether the eruptions should be deemed “the cause of” the uprisings. For example, if the Ptolemaic regime were especially vulnerable to Nile failures at this time, while another regime would have endured similar natural shocks without popular unrest, then it may be more appropriate to label those societal vulnerabilities “the cause of” the uprisings instead.
The discussion of causal “pathways” does not necessarily address this problem and could be even more misleading. After all, almost every causal pathway could be broken down into additional causal mechanisms ad infinitum. The length or complexity of the “pathway” per se doesn’t actually change how we determine causation. I can’t run over someone in a car and then claim that my actions didn’t cause their death because really there was this whole complex pathway between pressing my foot on the accelerator, the motor running, the car moving, the impact, injury, blood loss, etc. What matters, as discussed in the review of the first version, are determinations of the appropriate contrast set in cause and effect and the strength of causal necessity and sufficiency.
I would also discourage the use “proximate” vs. “ultimate” causation in lines 1303-1305 for the following reasons. First, the terms themselves are confusing and ambiguous. Without further explanation, many readers would assume that an “ultimate” cause should be somehow more fundamental than a “proximate” cause—exactly the opposite of how they are used in Gao et al. Second, the long time-series and abundant data for disasters and political events in imperial China enable inferences and analysis that just aren’t possible (yet) for a case like Ptolemaic Egypt. Third, and most important, the spectrum between “ultimate” and proximate” fails to capture the central problems regarding causation in the case of Nile failures and uprisings in Ptolemaic Egypt. As explained in the review of the first draft, these problems are essentially two-fold. On the one hand, we cannot say (yet) whether these Nile failures were necessary or sufficient for the uprisings to occur at all, or only whether they were necessary or sufficient for the timing or perhaps character of social and political turmoil that was going to occur sooner or later anyway. On the other hand, even if eruption-induced Nile failures were necessary or sufficient for these events, we cannot say (yet) which condition was more exceptional and relevant and therefore appropriately labeled “the cause of” the uprisings: the degree of Nile failure or the particular vulnerabilities of Ptolemaic regime.
I would strongly encourage the authors to spell out these issues of historical causal explanation plainly and clearly.
Finally, I would encourage the authors to tighten the language to improve readability, particularly in sections that have been added since the first draft (e.g., lines 180-183). The article is still unusually long. While the complexity of the topic does merit extra space, I believe it could be at least a page shorter simply by improving the style and removing ambiguities and redundancies.