Dear authors,

I think that this is a very interesting work that is dealing with a lot of different issues regarding solar physics and geometry, instruments and measurements and modeling at "difficult" in terms of available signal situations like this of the eclipse.

I think that the answers to the reviewers are complete.

I have some comments on the manuscript in addition to what the reviewers have already commented on.

Concerning the statements about the diffuse to direct ratio. Reading the paper it is not clear to me from the measurements if this is increasing or not during the eclipse. Looking at figure 10, what it looks like is that direct to global ratio is slightly decreasing at all wavelengths close to the eclipse time. This means that the diffuse to global ratios has to slightly increase. Is the above statement more or less correct ?

There was a lot of arguing against the non use of direct sun measurements and the choice of the 305nm and 340nm ( well, it was not actually a choice but what was available..). In addition, there is a number of other processes like the channel width (10nm FWHM) procedure to retrieve the 1nm spectrum and finally the FOV. FOV of the shadowband is difficult to simulate. And for sure it is larger than an “ideal” FOV of 0.5 degrees or even a 1.5-2.5 degree FOV that most of direct sun pointing AOT measuring instruments use. Correction techniques and calibration principles can improve a lot the AOΤ results but in the case of the eclipse things are much more complicated, especially at (low signal) lower wavelengths.
Could you comment on the above ?

Finally, the ozone issue is a big puzzle for me. Kazadzis, Kazantzidis, 2 papers of Zerefos and Groebner studies all pointed an artificial drop of total column ozone. On the contrary you find an increase. the limb darkening corrections and explenations of all the previous papres can be improved as you say. But the fact remains that a numner ( about 10) of instruments at different locations ( brewer and NILUs) performing automatic measurements are providing the same artificial decline of ozone.
Would it be possible to explain why this is not the case here ?
Is it the wavelength choice ?

To summarize, I think that the paper is a great contribution to the spectral behavior of solar radiation components during the eclipse. It provides new information from measurements and “new” models and opens up also new questions for future investigations. It provides evidence that the Nickel and Pierce models are helping towards explaining some open questions that raised from previous studies.

best regards
SK

I just have a minor comment on table 1 "by us" - > "in this study"

Also the captions of some of the figures are relatively large paragraphs that in principle they contain information that could hev been in the main body of the document.

However, I personally do not have a problem with this option. It is a bit not common but for me it is informative ot read it as I look at a figure.

congratulation on an interesting work
SK