Comparison of Hyperspectral Imaging and Fiber-Optic Reflectance Spectroscopy for Reflectance and Transmittance Measurements of Colored Glass
Round 1
Reviewer 1 Report
For me the current rationale of the paper is not that convincing. It seemed that the authors wanted to compare the performance of two different detectors (point measurement spectrometers versus a spectral camera). This aspect is not so innovative. Both devices have the same detector technology and a similar wave diffraction component so I would not expect a difference there. However, where I do see an aspect that might be interesting to the heritage community is the fact that a comparison is made between transmittance and reflectance. And in addition, comparing here the two measurement geometries which are linked to the Ocean Optics and the hyperspectral camera. This can indeed be of help when glasses need to be studied in-situ. Although in that case I don’t see the point of positioning an additional reflectance tile or backing paper behind the glass. In practice this is not feasible. In the current version the authors are focusing on the analysis of colour differences while I think intensity differences can never be eliminated (R versus T) but to my opinion this is also not needed when you want to analyse which chromophores are present. Obviously, it all depends on what you want to sense, a quantitative colour or an identification of the chromophores. A final remark is about the XRF data. I understood XRF data were collected to evaluate the sample uniformity. A better link must be made between the measurement area/location of the spectroscopic and the XRF measurements. For the XRF measurements, are these calibrated against standards (such as the corning standards?).
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The topic of the paper is quite interesting. Unfortunately, in my opinion, there are many critical issues.
First of all, English language that is really poor; there are a lot of inaccuracies both as terms and as grammar.
There are some paragraphs that result unclear, as the description, at rows 260-268, of the production of red glass. It seems that the glassmakers have to add a thin layer of copper nanoparticles on top of the glass, while they have to add a thin layer of red glass, colored with Cu nanoparticles on top of an colorless glass or to alternate multiple colored and colorless layers.
In the title the authors state “Comparison of hyperspectral imaging and fiber optic reflectance spectroscopy for reflectance and transmittance measurements of stained-glass”. Reading this, one would expect to find analysis done on stained glass, but then all the paper deals with analyses on samples of new and modern glasses supplied by a restorer and nothing about real stained-glass artworks. This would have been very interesting, given that making measurements on old glass, often corroded and degraded, is not as simple as working on new glass and the data obtained are usually more complex to interpret.
Figures are usually very small, and it is difficult and sometime impossible to see the small bands that are reported in the text (for instance, at row 212 or 271). Furthermore, there is no unit for Y axis.
Regarding the text, there are often discrepancies between what is stated and what results from the data presented.
Discussing about green glasses (rows 223-224) they state “The spectra of the Green 3 (Fig. 3) sample on the other end is very different from the others. In this case, the color is given only by iron and manganese”.
But a few rows below they state “Since Mn2+ is uncolored, it was probably used as a decoloring agent.”, so if it is uncolored how can it gives color to the glass?
At row 254, they state “The absorption band centered at around 780-790 nm in sample Blue 1 and 2 suggests that copper could have been used as an additional coloring agent as Cu2+ [3, 15, 16, 255 20], and probably with higher concentration in sample Blue 3.” But if you give a look at the XRF spectra in supplementary materials you find that the Blue 1 has the highest counts of Cu, in agreement with what is shown in the FORS spectra.
At row 273, the authors state “In Red 2 (Fig 7b), only the band at 565 nm is visible, but it is significantly less intense and sharp than in Red 1” but both reflectance and transmittance spectra are surely comparable.
Rows 276-282: All this I really confused, “In addition to copper, the presence of Fe2+ in both samples can be inferred by the presence of its absorption band at 1000 nm, more intense in sample Red 1 [10, 19, 20], even if XRF detected a higher amount of iron in sample Red 2.”. Then “The fact that its absorption band is more intense in Red 2 could depend on the higher transparency of the red layer of Red 1.” So is more intense in R1 or in R2???
About Red 3: “The glass composition could have played an important role. A particular concentration of antimony (Sb3+) and tin (Sn2+), detected by the XRF both in sample Red 2 and 3, could be the reason behind the different coloration of Red 3, as well as a different annealing temperature” In XRF spectra Sb is higher in Red 2 than in Red 3, I don’t how they can state from XRF that Sb3+ and Sn2+ are present.
Which contribution they could give to the color of the glass? Sb is normally used as bleaching or opacifying agent.
At row 302: “However, the data obtained with the hyperspectral camera seem to have an opposite trend, with the two values of L* (reflectance and transmittance) quite high, and almost comparable. It is likely that this arises from the different optical design of the instruments.” How can the authors define quite high L values of 5.71 e 3.74? This is something near to black!
Rows 329-333: “From the chemical point of view, according to the available literature, the orange/amber color in glass is given by the ferric iron-sulfide (Fe3+-S) complex. This chromophore can be recognized by a broad absorption band at around 410 nm [4, 10, 12, 13, 16, 20, 40, 46]. XRF analysis seems to confirm this hypothesis, as both the elements were identified in all the samples.” No evidence of S is highlighted in the XRF spectra in the supplementary materials.
In the Discussion paragraph, at row 395:” in general, the intensity of the spectra obtained by the FORS are consistently higher than the hyperspectral one, apart for few exceptions.” This seems not so true if we consider the spectra reported. But in general all the Discussion paragraph is quite confused:” It could be possible that the different distance of the detector from the sample (in contact/not in contact), combined with the surface characteristics (presence of bubbles, scratches, uneven areas…) affected the results for the darkest glass pieces more.” What is the meaning of this sentence?
Conclusions are poor.
These are only the most important things but there are many other inaccuracies and questionable claims in the paper.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 3 Report
This is a very interesting work, considering the number of studies conducted in glass, from mosaic tesserae to stained glass and glass equipment and jewelry. In every case, spectrophotometric measurements on both T% and R% modes are considered as “painful” as many aspects should be taken under consideration, such as transparency, background material used -for reference, and for R backing material-, illumination and thickness of sample. In any case, this work provides a comparative and comprehensive study of the aforementioned spectrophotometric measurements, trying to optimize measurement protocols.
The manuscript is of scientific soundness as the authors do not just provide the collected spectra, but they analyze them also, with extensive references. The use of language is good and the manuscript has an easy-to-follow structure. Some points to consider:
11. The two different papers used as backing materials should be mentioned, in the case of readers with no access to used ISO
22. Please, also insert ΔΕ formula, for the same reason.
33. Consider adding your elemental analysis measurements, even in a Table. Considering the fact that the journal has no table, figure and length limits, elemental analysis should not be restricted only in the supplementary section, as the results can confirm the spectral interpretation
Some minor language and style comments:
11. L.34. on the materials used
22. L. 44. “glass is transparent”
33. L.146. Fig. 3
44. Whole manuscript. Please, check the journal’s template regarding mentioning of figures in the manuscript.
55. Please, put a single space between values and units in Experimental section
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The manuscript can be published in its present form.
Author Response
Thank you
Reviewer 2 Report
The paper was highly improved but I have still few considerations.
First of all the title. The authors state "measurements os stained-glass", but this is only a future prospect for the use of these techniques since the work only talks about new glass samples, quite different from old and often corroded glasses in real stained glass windows.
In many cases, when results are not clear or not in agreement with what expected, they simply say that further investigation would be necessary. This might be acceptable if unclear things were few but in this paper there are too many obscure points and hypotheses not supported by reliable data.
Author Response
Thank you for your comments. Below are the responses:
1) First of all the title. The authors state "measurements os stained-glass", but this is only a future prospect for the use of these techniques since the work only talks about new glass samples, quite different from old and often corroded glasses in real stained glass windows.
1)We change the title with "colored glass".
2)In many cases, when results are not clear or not in agreement with what expected, they simply say that further investigation would be necessary. This might be acceptable if unclear things were few but in this paper there are too many obscure points and hypotheses not supported by reliable data.
2)Some sentences have been rephrased. Nevertheless, the main scope of the paper is to compare the performance of the FORS and HSI and we think that specifying other techniques is out of the scope of the work.
