OPT-IN: Beam-spin asymmetry Σ for Σ− hyperon photoproduction off the neutron

[David Ireland]

Dear Nick, et al.,
 
Looks like you have done a nice job in analysing and writing up this work. I have just a few comments that need addressing:

* line 59: "complete experiment": whilst this is often referred to in the literature, it is only a mathematical construct when talking about a set of observables. Observables are not what are measured; what is measured are cross-sections and asymmetries, and it has been demonstrated in a number of papers that the critical requirement to enable the extraction of amplitudes is to have polarized combinations of the components of the reaction (beam, target, recoil). Each measured asymmetry is then a combination of several observables. This is described in the recent review by myself, Eugene and Igor (https://doi.org/10.1016/j.ppnp.2019.103752). I am not angling for an additional citation (actually I am!), but this is also a compendium of results for all meson photoproduction measurements in the last few decades, and is probably worth citing.
A suggested phrase to substitute lines 59 – 64 in this case could be:
"Values of the four complex amplitudes may be extracted up to an arbitrary phase, given data from a suitable combination of polarization measurements of sufficient accuracy, which would therefore provide a maximum constraint…"

 * line 87: “Recent measurements”: you might want to cite also the CLAS g8 measurements reported in Paterson et al. (http://dx.doi.org/10.1103/PhysRevC.93.065201), from which the findings in the Anisovich et al paper are derived.

* The above Paterson et al reference also gives, in the appendix, the derivation of likelihood that you show in lines 200 onward. You probably had subtle differences, but essentially your likelihood is a special case of the more general extraction method.

* line 434: For this paper, simple stating that the variant of Bonn-Gatchina with the smallest $\chi^2/ndf$ is fine, because that is how the choice has been made. For the further publication, I would caution that this is not the best way to compare models. It is a good statistic for assessing goodness of fit, but can always be improved by adding extra resonances (= more fit parameters), so does not really mean anything unless you also penalise for number of parameters.

Best wishes, 
Dave