**************************************************************************** * Albert's email: **************************************************************************** Hi Jeff, I have a few comments, but first, this is a very good paper (compared to most in this little area of activity). Other comments: (1) I would like to ammend the title to include "in the Energy Range of Neutron Beta-Decay". >>>>>>>>>>>>>>>>>>>amended (2) The sentence beginning "The range for 43.5 keV electrons in plastic" is a little confusing. Are you trying to say that the mean range is 3 orders of magnitude larger than the Al layer? I have a few comments on this if that is what you mean: i) For backscatter, half the mean range is the relevant scale for backscattered electrons. ii) dE/dx and backscatter in Al will be higher by roughly a factor of 2 iii) This makes the factor of 600 quoted pretty close to 100 iv) Perhaps we should soften our position on this to: "Backscattering from Al itself, or rescattering from the Al, is therefore suppressed at a level below 1%, even for the lowest energy incident electrons reported in this work. >>>>>>>>>>>>>>>>>>>softened (3) In both Sections A and B concerning the measurements, I always feel that a table summarizing the systematic uncertainties are needed. I know this will clearly push us beyond brief report status, but probably we should just make this a normal article anyway. >>>>>>>>>>>>>>>>>>>put back. Those tables were removed previously because they were nearly identical to the ones in our previous paper. The information in the tables is taken from Mike's SURF report (see http://nuclear.uwinnipeg.ca/ucn/bs/ under "SURF Report"). I also included a subtable on extrapolation (which was not done before), and snuck in a whole bunch of references to these tables all over the place, basically whenever a numerical or comparative statement about systematics is made. Please check and make sure I didn't screw it up!!! (4) As before, I do feel we should put a quantitative number on the time variation of beam currents during a typical run (due to charging). Didn't we say that we had access to these data? >>>>>>>>>>>>>>>>>>>modified in the following way for silicon detector mode: No such effects could be seen with the Al-coated scintillator target. On leaving the beam on the target for long periods of time, no change in the sensed current was seen at the 1\% level. To also search for charging, the electron beam could be switched off and on rapidly by inserting a Faraday cup upstream of the chamber, which would not cause the beam to vary in intensity. Upon restoration of the beam, the current sensed by the target was found to agree with the value before intercepting the beam with the Faraday cup to the level 1\%. >>>>>>>>>>>>>>>>>>>also added the following for current integration mode: A dependence on beam current at the level 3\% had been seen for our previous current integration mode data on Be and Si target~\cite{bib:prcme}. For those data, the current dependence was attributed to charging of various components on the chamber. As our new studies were done generally at lower beam currents, we expect this contribution to be smaller, but retain the pessimistic upper limit of 3\%. (now followed immediate by systematics table). (5) A point of curiousity: How do the extra systematic uncertainty (associated with extrapolation to zero energy in integrals over q) vary with energy? >>>>>>>>>>>>>>>>>>>see http://krl.caltech.edu/~jmartin/ucn/work/backscatter/ under "Report on the other older systematics studies..." Table 3 on the last page which states: 15-20% at 40 keV and <4% at 120 keV. Yes, this is included in all plots where integrals of silicon detector data are shown. I never made a plot of that systematic uncertainty by itself - it was always just added in quadrature to the rest. A little more info is included in this in the paper now that the systematics tables have reappeared. (6) At this point, I'm pretty satisfied with the chi^2 treatment. It's still hard for the reader to understand at what level the mean square deviations between experiment and the models are coming in, but of course our data figures are pretty reasonable for them to eyeball this quantity. Given that we expect the point-to-point uncertainty to be much smaller than our overall quoted uncertainties, wouldn't it be reasonable to state (at least) that the absolute value of the chi^2 associated with all of our fits (except Be for Geant) indicate general agreement between the models and theory when the uncertainty in these fits is taken to be our overall uncertainty? >>>>>>>>>>>>>>>>>>>I have temporarily inserted a statement about absolute chi-squareds that I find acceptable. I have placed it in the "Quantitative" section. Please read and see if you really want a statement like that: I.e., when taking the point-to-point uncertainty to be equivalent to the total normalization uncertainty, the value of the reduced $\chi^2$ was generally significantly less than unity. The exception to this was the comparison of the Geant4 simulation to $\frac{1}{N_e}\frac{dN}{d\Omega dq}$ for Be targets, where an absolute reduced $\chi^2$ of 2.1 was seen. Comments (3) and (4) might require a bit more work (I don't know if you all feel they are really important, but I do think they will add signifcantly to the qualtiy of this publication). I believe I suggested these on earlier versions of the paper as well...I don't think we should cling to the Brief Report format if it detracts from the value of this publication. Thanks Albert **************************************************************************** * Junhua's email: **************************************************************************** Hi Jeff, In the caption of Fig1, You were still using the old factors. Junhua >>>>>>>>>>>>>>>>>>fixed **************************************************************************** * Brad Plaster's email: **************************************************************************** Hi Jeff, Your paper looks pretty good. I guess if you were looking for ways to shrink the text, you could eliminate the two paragraphs on p. 2 beginning with "Fig. 1(a)..." and "Fig. 1(b)...", as the information contained there is essentially repeated in the caption for Fig. 1. Fig. 2 could also probably be "compressed" such that it is only as "tall" as the top panel of Fig. 1 (as there seems to be a lot of dead, white space in its interior). But, whatever... Brad >>>>>>>>>>>>>>>>>>>eliminated repetitive text, viz: Fig.~\ref{fig:120} compares Monte Carlo simulations based on Geant4 and Penelope. In each case, a scale factor is applied to the Monte Carlo, in order to allow a more accurate visual comparison of the curves. The scale factor was determined from a fit to the data, which is described in Section~\ref{sec:fit}. As for our previous work, Geant4 systematically underestimates the peak in the data near $q=0.95$. However the positions of the low-energy and elastic peaks are rather well-described by Geant4. In the case of the Penelope simulation, when the Monte Carlo is rescaled, it is apparent that trends in both energy and angle are well represented by Penelope. This result is discussed quantitatively in Section~\ref{sec:fit}. >>>>>>>>>>>>>>>>>>>Fig. 2 left as is for the time being. Fig. 1 will likely have to be changed, as the text is way too small or font wrong or something (compare to our previous article). I will make a note of this to Junhua.