Ground Penetrating Radar Investigation of Corvin Castle (Castelul Corvinilor), Hunedoara, Romania

Round 1

Reviewer 1 Report

In order to support the restoration as well as to derive a deeper understanding of the construction phases and histories of the Corvin Castle, the performance and potential of GPR in the subsurface archaeological prospection is  well presented. In addition, several unknown archaeological features and constructions were identified. Generally, the manuscript is well drafted, and the quality of GPR slices and profiles are high enough to show archaeological evidences. My personnel comments for the further improvement of this manuscript is as follows:

1) Minor english style editing to avoid typo errors, such as page 26 Line 465, "one of these features or a combination thereof"

2) Figure 24, Notations of X10 and Y8 are missing in the slice subimage with a depth of 1.25 m.

3) Are there any artificial radar reflections in the GPR profiles, such as the signature reflected from the surface wall corner. I observed potential artifacts in Figure 14-X7, Figure 24-X10, Figure 30-Y4G3*.

4) Section 4. "Discussion" could be replaced by "Discussion and conclusion"

Author Response

Dear Reviewer,

We would like to thank you very much for careful review and critique that helps us improve the manuscript. Our answers to your comments and concerns are provided below.

In order to support the restoration as well as to derive a deeper understanding of the construction phases and histories of the Corvin Castle, the performance and potential of GPR in the subsurface archaeological prospection is  well presented. In addition, several unknown archaeological features and constructions were identified. Generally, the manuscript is well drafted, and the quality of GPR slices and profiles are high enough to show archaeological evidences. My personnel comments for the further improvement of this manuscript is as follows:

1) Minor english style editing to avoid typo errors, such as page 26 Line 465, "one of these features or a combination thereof"

These have been rephrased and fixed, including:

(p.9 line 236-237 )

number of other sources that include the castle or a feature of the castle in either a specific investigation or as part of a more general survey, e.g., [26], [27], and [28].

(p.26 line 493-494)

The reflection in Y4 must be related to one or more of these features.

(p.35 line 621)

though the orientation of this transition is unclear in GPR scans.

(p.36 line 663)

GPR investigations of Corvin Castle will aid the present restoration effort and future efforts by providing important  and more complete reference information about architectural features.

2) Figure 24, Notations of X10 and Y8 are missing in the slice subimage with a depth of 1.25 m.

This has been corrected (page 27, Figure 24):

3) Are there any artificial radar reflections in the GPR profiles, such as the signature reflected from the surface wall corner. I observed potential artifacts in Figure 14-X7, Figure 24-X10, Figure 30-Y4G3*.

In response to this comment, when these kinds of artifacts are present, they are mentioned and discussed:

(p.17) Fig 14-X7 labeled this first appearance of the artificial reflection and referred in text to it as such:

(p17, Lines 391-392):

The artificial air reflection at the exterior wall is also visible in the lines that begin perpendicular to that wall (Figure 14), accompanied by the strong shallow reflections from the paving and leveling surface of the floor

Figures 24, 30: mentioned in text:

p.26, lines 484-485:

Also note the “ringing” effects on the radargram (localized columns of black and white stripes) from a strip of iron installed in the floor and the carpet between the apse and nave  and the artifacts from corners, as in Figure 14 (Figure 24, line Y8).

p.31-32, lines 556-557:

Note the artificial reflections as in Figure 14 from the air wave inside the corner of the room.

4) Section 4. "Discussion" could be replaced by "Discussion and conclusion"

This suggestion has addressed by separating the discussion from the conclusions (p. 35-36, lines 625-663); additional discussion of future work to address challenges has been added (p.35-36, lines 638-642, 657-659, and 663):

Discussion

Features identified with GPR provide insight into lingering questions and unresolved historical debates about the castle’s past, including the history of the lapidarium area (Figure 30).  This room is much lower in elevation than the rest of the castle.  An additional hypothesis arising from these investigations is that the lapidarium and adjacent courtyard with the same floor level were part of the pre-Bethlen phase of the castle.  Beginning with Bethlen, later modifications may have sought to provide a larger and more even footprint for the structure.  The accuracy of earlier plans of Corvin Castle has been substantiated by GPR surveying and analysis.   In scans of the northeast portion of the castle (surrounding the Well Courtyard), there are reflections from wall foundations running East-West and North-South (Figure 31). This feature is shown in the Moller map, however no longer present above ground or indicated in other plans. The data attest to the presence of the crypt and the original fortress wall foundation through the chapel, as in some sources (e.g., Moller, Figure 24). The jagged path of a wall running through the Knight’s Hall in Moller’s map, possibly indicating the uncertainty of the 14th century wall, was verified as a continuous and curved foundation in GPR scans without the jog (Figure 29). Further investigations into the substructures and history of these areas of the castle are an excellent candidate for future targeted investigations aimed at resolving these questions and validating the results of this work.  These investigations could be performed as part of restoration efforts using a variety of methods, including small targeted excavations, higher frequency GPR survey, or other methods.

Conclusions

Even though historical accounts of Corvin Castle are sparse and can be inconsistent or difficult to obtain, GPR scans were able to confirm features reported in both plans and textual sources, especially the foundation of the 14th century fortress walls (e.g., Figures 15, 22, and 24).  In addition, GPR expands our knowledge of known features, corrects previous assumptions, and even identifies unreported substructures and architectural features, especially in the White Tower and the chapel (e.g., Figures 18,19, and 24). The extent of the bedrock foundations and transition to built-up masonry foundations was previously unknown and has been mapped using GPR in the scanned areas (Figure 34).  Moisture ingress identified in the Knight’s Hall and the lapidarium can be localized and investigated (Figures 29 and 31).  Unidentified anomalies can be sought out and clarified, increasing our knowledge of Corvin Castle and its complex history while preserving the integrity of the site.

GPR surveys help critically evaluate the quality of the information provided by historical documents, reveal previously unrecorded information (such as the locations of some of Bogdan’s backfilled 1969 trenches, figure 21) and provide clues to the internal architecture of previous phases (as it did in the present Noble’s Man Kitchen [Figure 16], the courtyard [Figure 23]and lapidarium [Figure 30]). The data also serve as a baseline for more detailed study; for this reason, both selected features and typical scans are presented here. At Corvin Castle, where invasive tests have been the primary (if not exclusive) means of investigation for the last two centuries, nondestructive methods like GPR are critical to the survival of this heritage site and our overall understanding of its history. GPR investigations of Corvin Castle will aid the present restoration effort and future efforts by providing important and more complete reference information about architectural features.

Author Response File: Author Response.docx

Reviewer 2 Report

Adding some references, citing in text the first figures, and cleaning up or describing better the processing of the GPR results (bidirectional survey not mentioned in method, yet evidence for it found in figures and scan lines crossing). See docx for few more details.

Comments for author File: Comments.docx

Author Response

Dear Reviewer,

We would like to thank you very much for careful review and critique that helps us improve the manuscript. Our answers to your comments and concerns are provided below.

Heritage Peer Review of Ground Penetrating Radar Investigation of Corvin

Castle (Castelul Corvinilor).

Adding some references, citing in text the first figures, and cleaning up or describing better the processing of the GPR results (bidirectional survey not mentioned in method, yet evidence for it found in figures and scan lines crossing). See docx for few more details:

These comments are very helpful and the suggestions for clarification, additional references, and expansion have been incorporated in the text according to the following specific comments:

None of the early figures are actually cited in the text until second half of article. I believe they are close to appropriate discussion points in the first part of the text as presented, but authors need to list where they apply specifically.

This has been fixed in the following lines of the introductions:

(p.2, line 33):

When it comes to a challenging and worthwhile conservation project, Corvin castle is a ready example  (Figure 1)

(p.2, line 39):

though the exact date is subject to scholarly dispute over the veracity of property documents drafted in the medieval period ([1] vs [2], as well as [3]; see also Figure 2).

(p.3, line 53):

Brandenburg sold the castle to the Török family in 1526, and during that time there is little documented alteration besides the discovery of the foundation of a small structure in the SE corner of the castle possibly built in the early 16th century [7][1][6], see Figure 3.

(p.3, line 64):

Subsequent imaginative restoration works executed between the end of the 19th and early 20th centuries may have caused more harm than good, as original elements of the Western palace were destroyed and “faithfully” reconstructed by Ferenc Schulcz, while other elements were “restored” more creatively [1], see Figure 4.

(p.4, line 85):

Though the results were never officially published, some of Floca’s 1956 archaeological report was later analyzed by Bogdan in an attempt to gather information on the evolution of the castle’s building phases shown in Figure 3 [6].

Section 1.2. Nondestructive Methods – a number of good noninvasive techniques are mentioned, but no references are given to examples of there use in conservation. The authors should have at least one example to reference for each technique discussed.

These references have been added as follows (p5. Lines110-114 ):

investigations of historic buildings.  These include portable XRF (X-ray fluourescence) analyzers [9], ultrasonic testing [10], etc.  In structural applications (for feature detection and resolution of targets larger than a few centimeters) methods like GPR, magnetometry, and photogrammetry are common [10][11][12].  These methods have the potential to provide information about a structure, ranging from the composition of pigment layers in paintings [9] to the mapping of foundation walls [12].

Figure 14 to 19. No sure if survey was run in perpendicular manner. Both figures have a lot of star or “t” shapes that I normally see in processed X plus Y direction surveys. If it was run both directions, didn’t see that noted in the methods. Would also be good to diminish one direction (1X + 0.6Y) so the star shapes don’t appear. Also, background matching for the different grids is poor (very sharp color change between grids).

This omitted detail about the bidirectionality of the surveys is added (p.7, line 175): 

…in bi-directional (X and Y) orthogonal grids so that depth slices could easily be generated to

aid in interpretation.

We agree that there are star artefacts from the bilinear interpolation (it might have been less pronounced with nearest neighbour interpolation or closer linespacing), however, at this stage we are unable to reprocess the data, but we introduced appropriate discussion in the manuscript so that it explains this particular aspect of the data (p.8, lines 198-205):

Initial interpolation between lines is done bi-directionally; this interpolation scheme produces the two-directional smearing of strong amplitude reflections into plus- or star-shapes (e.g., Figure 19).  Other interpolation schemes reduce these artefacts of data processing, though they were not available or necessary for this project.  Interpretation of some grids is improved by using only one direction of lines; single-direction depth slices are noted and presented for some grids.  These reasons include bias from different line spacing in each direction, orientation of features in the scans like floor tiles or columns, and directionality of the surrounding environment, which might include anything from electrical wiring to agricultural furrows.

Figure 28. Is the spacing / processing / gridding of this different from other slices shown? Something very different happed to these scans.

These scans are indeed fundamentally different than the others.  The floor of the Diet Hall is the ceiling of the Knight’s Hall, so this is the only ceiling that is scanned.  In addition to the high contrast reflections from the top and bottom of the vaulted ceiling, the construction is also more modern and reflections indicate that it also contains metal reinforcing.  In combination, these factors indicate that the scans should contain much higher amplitude reflections than the other rooms.  The changes include discussion about the remodelling and altered appearance  on p.29 lines 509-515:

Though the scans of this room appear markedly different from other scans of the castle, the survey parameters and collection method are the same in the Diet Hall as other rooms.  The differences are primarily caused by the nature of scanning a ceiling, where there are clear reflections from a single integrated level with air on both sides.  The ceiling was restored in 1874 by Schulcz and later by the Historical Monuments Directorate (1966-1969) [6], [1].  The extent of the latest restoration is unclear, though the reflections indicate that the ceiling may be made (at least partially) of reinforced concrete (Figure 28, especially X26).

Author Response File: Author Response.docx

Reviewer 3 Report

Very good paper with great cultural interest.

My suggestion is: accepting in present form.

If you want to improve something, and the Editor accepts this:

Figure 10. Chapel excavation stratum as reported in [27] ->
Figure 10. Chapel excavation stratum as reported by Roman, Diaconescu and Tiplic in 2004 [27].
The same in the text and in Figure 17, 34.

Figure 17. Add a scale.

Author Response

Reviewer 3:

Very good paper with great cultural interest.

My suggestion is: accepting in present form.

If you want to improve something, and the Editor accepts this:

Figure 10. Chapel excavation stratum as reported in [27] ->
Figure 10. Chapel excavation stratum as reported by Roman, Diaconescu and Tiplic in 2004 [27].
The same in the text and in Figure 17, 34.

Figure 17. Add a scale.

The authors would like to thank Reviewer 3 for the positive evaluation of our paper and very helpful comments on referring to references and figures 10, 17, and 34. The following changes to the captions have been made and carried though the manuscript:

p.13:

Figure 10. Chapel excavation profile as reported by Roman et al in 2004 and 2012 [31] and [2].

p.20:

Figure 17. Bogdan’s reported profiles (with author translations) from the white tower and entrance

hall. H1 is in the tower (east), H2 is in the entrance hall (west) (figure from Bogdan (1970, [ 6 ] used with

permission).

p.35:

Figure 34. Compilation of scanned and identified foundations. The foundation composition in the chapel is still unknown, but likely contains both built-up and bedrock foundations. Note the unscanned areas and comparison to the solid black 14th century extent walls, overlain on map from Bogdan (1970, [6].)

Scale is added to Figure 17.

Author Response File: Author Response.docx

Reviewer 4 Report

This is a quite interpretation-detailed GPR survey about a highly modified and little documented castle. Below, some suggestions to help to improve the paper:

Add the location/at least country in the title

Line (25-26) "From limited time…": Not sure what is this challenge about...can you explain this further?

Line 98: why is "Ground" capitalised only? This may be the moment to introduce the acronym.

Line 103-104 “it is not usually possible to conduct invasive tests, collect samples, and performmore excavations, so all methods of inquiry must be noninvasive.”: In the   field of cultural heritage management, at least in Europe, invasive methods   are used (e.g. test trenching, field evaluation… ) in regular bases. Non-invasive techniques, like geophysical techniques are no mandatory in most cases in   Europe. You sentence, at least on its reference to CHR and archaeo-geophysics, it’s not correct.

Line 128 “The size of the features…”: The size of the features does not depends on the freq…this statement is not correct/clear…I believe that what you mean is that resolving small features (and energy depth penetration) is a function of frequency used. Please rephrase this. 

Line 144: The selection of a geophysical technique or combination of them is a function of many things but very importantly, upon the characteristics of the site to survey (including terrain, soil/geology) could you expand a bit more on justifying why you selected GPR and no other geophysical technique (or combination of techniques) according to the specifics of your site?

Line 170-172 "For example, vertical scans can locate a former staircase embedded in a wall by identifying reflection patterns of different construction material and technique in the shape of stairs.”: Provide a   reference

Line 186-187, 352-353 "nominal velocity of 0.1 m/ns, which was verified when187 possible": How was   this verified?

Line 377: It may be helpful to scale the B-scans to the photo

Several lines/pages: Fig 18, 19 and other: there is a lot of banding in the Bscans. For the sake of data presentation it’d be good to reprocess these ‘busy’ datasets (also depth slices). Also, from Fig 18 on-wards, there is a tendency to crowd and reduce figures and b-scans making difficult to see the data. I suggest to make them bigger and select key results 

Line 592-595: It’d be good to refer again to figures

The paper lacks of a conclusion. Please, add this section and ideally, separate discussion & conclusion. Also, could you please add a bit on the limitations/problems you encounter during your survey. Further work and ideas for data validation could be also interesting.

Author Response

Reviewer 4:

Dear Reviewer,

We would like to thank you very much for careful review and critique that helps us improve the manuscript. Our answers to your comments and concerns are provided below.

This is a quite interpretation-detailed GPR survey about a highly modified and little documented castle. Below, some suggestions to help to improve the paper:

Add the location/at least country in the title

Title has been changed to “Ground Penetrating Radar Investigation of Corvin Castle (Castelul Corvinilor), Hunedoara, Romania”

Line (25-26) "From limited time…": Not sure what is this challenge about...can you explain this further?

Here we are referring to limited time available to spend working on a conservation project (limited by budgets, open hours of the site, etc.).  Page 1, lines 25-26 have been modified:

 However, they often face an onslaught of challenges-- from limited work time, funding, and personnel, to the complicated relationships between stakeholders, funding sources, and legislative bodies.

Line 98: why is "Ground" capitalised only? This may be the moment to introduce the acronym.

This typo has been corrected and the acronym introduced (p. 5, line 99)

historical information, apply scientific techniques like ground penetrating radar (GPR), and integrate…

Line 103-104 “it is not usually possible to conduct invasive tests, collect samples, and perform more excavations, so all methods of inquiry must be noninvasive.”: In the   field of cultural heritage management, at least in Europe, invasive methods   are used (e.g. test trenching, field evaluation… ) in regular bases. Non-invasive techniques, like geophysical techniques are no mandatory in most cases in   Europe. You sentence, at least on its reference to CHR and archaeo-geophysics, it’s not correct.

The sentiment of these statements has been shifted from mandatory stipulations to a discussion of the merits of noninvasive tests to supplement or improve existing practices (p.5, lines 103-120): 

Regular site evaluation and investigation practices include excavation, sample collection, and other invasive testing.  Though these practices are accepted, invasive testing permanently alters the site or object being conserved and may not be desirable. For example, as excavation procedures and documentation have improved, they can reveal more and more information about a site; however, it is not possible to retroactively re-excavate an important site that was first excavated with very different means.  This motivation has developed the field of nondestructive testing and evaluation into an increasingly robust set of techniques and tools to perform assessments and investigations of historic buildings.  These include portable XRF (X-ray fluourescence) analyzers, ultrasonic testing, IR imaging, etc.  In structural applications (for feature detection and resolution of targets larger than a few centimeters) methods like GPR, magnetometry, and photogrammetry are common.  These methods have the potential to provide information about a structure, ranging from the composition and arrangement of pigment layers in paintings to the mapping of foundation walls.  They are best used in concert with one another, as many of the methods provide complementary information that can be used to gain a fuller understanding of the investigated work.  Increased applications of these tools can improve the methods and techniques themselves, as well as increase the appeal and familiarity of these methods to a broader audience.  

Line 128 “The size of the features…”: The size of the features does not depends on the freq…this statement is not correct/clear…I believe that what you mean is that resolving small features (and energy depth penetration) is a function of frequency used. Please rephrase this. 

You are correct; this has been rephrased (p.6, line 137):

The size of the smallest resolvable features and the depth of signal penetration depends on the frequency of the antenna.

Line 144: The selection of a geophysical technique or combination of them is a function of many things but very importantly, upon the characteristics of the site to survey (including terrain, soil/geology) could you expand a bit more on justifying why you selected GPR and no other geophysical technique (or combination of techniques) according to the specifics of your site?

Discussion about the selection of GPR for this particular environment is added in this section:

p.6, lines: 151-159:

As such, GPR is a natural choice to gather and verify important details about Corvin Castle’s history and physical structure.  The castle is an enclosed, electrified, and paved space open to tourists; as such, the environment has particularly high background noise and confined spaces that impede equipment maneuverability over highly modified ground.  Other non-destructive methods like electrical resistivity tomography (ERT) were not appropriate because of the solid floors, while potential field surveys (i.e., magnetic gradiometry) inside buildings are relatively absent from the literature and do not provide useful results in setting such as the castle.  For these reasons, GPR was selected as an appropriate method of investigation over other ground-based techniques for subsurface investigation in complex urban environments.

And p.6, lines 162-163:

GPR was chosen because of its ability to provide the desired information, the interpretability of the results, and the availability of the equipment and resources to perform the study.

Line 170-172 "For example, vertical scans can locate a former staircase embedded in a wall by identifying reflection patterns of different construction material and technique in the shape of stairs.”: Provide a   reference

The staircase is visible in the Neumann map (figure 4) and in the photograph of the well in figure 13.  These references are added (p.7-8, lines 189-190):

For example, vertical scans can locate a former staircase embedded in a wall by identifying reflection patterns of different construction material and technique in the shape of stairs (refer to Figure 4 and the wall in the background of the image of the well in Figure 13).

Caption for Figure 13 (p.16):

Features in the northeast sector, including the Bear Pit and the repurposed frame and embedded staircase in the Well Courtyard.

Line 186-187, 352-353 "nominal velocity of 0.1 m/ns, which was verified when187 possible": How was   this verified?

This velocity is verified where depth information (mostly approximate) is available from excavations, in the Chapel, White Tower, and Entrance Hall.  The text is modified to read (p.8, lines 209-211):

…a nominal velocity of 0.1 m/ns, which was verified when possible using approximate depth information from excavation reports (in the Chapel, Entrance Hall, and White Tower).

Line 377: It may be helpful to scale the B-scans to the photo

The scans are now scaled and the alignment is indicated by labels in the photo:

Several lines/pages: Fig 18, 19 and other: there is a lot of banding in the Bscans. For the sake of data presentation it’d be good to reprocess these ‘busy’ datasets (also depth slices). Also, from Fig 18 on-wards, there is a tendency to crowd and reduce figures and b-scans making difficult to see the data. I suggest to make them bigger and select key results 

We agree that there is a lot of banding; however, the banding does not obscure the features. Because the authors wished this paper to serve as a reference for future works, the scans were left minimally processed.  That is, we did not wish to over-emphasize the features we interpret (and minimize features that were less apparent) in hope that minimally processed data will retain aspects pertinent to future studies and present a reference or confirmation for future investigations. Unfortunately, at this stage we are unable to reprocess the data, but we introduced appropriate discussion in the manuscript so that it explains this particular aspect (p.16-17, lines 372-381 and few other lines as shown below). 

 (p.16-17, lines 372-381 (Results)):

GPR surveys have provided important results for each of the areas presented here.  In general, scans of the floors successfully reveal pavement levels, changes in floor covering, moisture patterns, and myriad substructures.  In some areas, such as the courtyard and bear pit, features were more difficult to interpret.  This is largely due to the uneven ground and surface debris in these areas. Asterisks (*) indicate that GPR scans of the area include a material change in the foundation of the indicated room; the foundations are discussed separately.  In most rooms, the penetration depth of the antenna was estimated at between 1 - 1.5 m (using a velocity of 0.1 m/ns, which is appropriate for most of the castle and verified where possible using depths reported by excavations). Note that features are visible in very lightly processed data; this data is presented in an effort to provide a reference for continued study of the castle. 

(p. 36, line 663 (conclusion)):

GPR investigations of Corvin Castle will aid the present restoration effort and future efforts by providing important and more complete reference information about architectural features.

(p.36, lines 658-659):

The data also serve as a baseline for more detailed study; for this reason, both selected features and typical scans are presented here.

As for density and number of figures, this study serves as an important baseline filling in our gaps of knowledge about the castle.  It is the authors’ aim to provide a complete reference for the scans of the castle, showing at least one scan containing both typical reflections and feature reflections.  Some figures have been expanded/reorganized to become less crowded (bear pit scans, ne sector scans), and the others have been rescaled so that they are a more appropriate size.  

Line 592-595: It’d be good to refer again to figures

These references have been added as follows

(p.34, lines 608-610):

It is proposed that this represents the transition between foundation on bedrock and the foundation laid on top of bedrock using supplemental construction materials (see Figure 34).  This trend is seen in multiple rooms across the scanned areas of the castle (e.g., see Figures 18, 29, and 30).

(p.36, line 648):

GPR scans were able to confirm features reported in both plans and textual sources, especially the foundation of the 14th century fortress walls (e.g., see Figures 15, 22, and 24).

(p.36, lines 646-652):

In addition, GPR expands our knowledge of known features, corrects previous assumptions, and even identifies unreported substructures and architectural features, especially in the White Tower and the chapel (e.g., see Figures 18, 19, and 24). The extent of the bedrock foundations and transition to built-up masonry foundations was previously unknown and has been mapped using GPR in the scanned areas (see Figure 34).  Moisture ingress identified in the Knight’s Hall and the lapidarium can be localized and investigated (see Figures 29 and 30).

(p.35, line 625-626):

Features identified with GPR provide insight into lingering questions and unresolved historical debates about the castle’s past, including the history of the lapidarium area (see Figure 30).

(p.36, lines 655-658):

And provide clues to the internal architecture of previous phases (as it did in the present Noble’s Man Kitchen (Figure 16), the courtyard (Figure 23), and lapidarium (Figure 30)).

(p.35, lines 636-638):

In scans of the northeast portion of the castle (surrounding the Well Courtyard), there are reflections from  wall foundations running East-West and North-South (Figure 31).  This feature is shown in the Möller map, however no longer present above ground or indicated in other plans. The data attest to the presence of the crypt and the original fortress wall foundation through the chapel, as in some sources (e.g., Möller, Figure 24). The jagged path of a wall running through the Knight’s Hall in Möller’s map, possibly indicating the uncertainty of the 14th century wall, was verified as a continuous and curved foundation in GPR scans without the jog (Figure 29).

The paper lacks of a conclusion. Please, add this section and ideally, separate discussion & conclusion. Also, could you please add a bit on the limitations/problems you encounter during your survey. Further work and ideas for data validation could be also interesting.

This suggestion has been implemented; most discussion exists in the relevant sections of the results.  

Challenges have been added to the results section (p.16-17, lines 374-381):

GPR surveys have provided important results for each of the areas presented here. In general,

scans of the floors successfully reveal pavement levels, changes in floor covering, moisture patterns, and myriad substructures. In some areas, such as the courtyard and bear pit, features were more difficult to interpret. This is largely due to the uneven ground and surface debris in these areas. Asterisks (*) indicate that GPR scans of the area include a material change in the foundation of the indicated room; the foundations are discussed separately. In most rooms, the penetration depth of the antenna was estimated at between 1 - 1.5 m (using a velocity of 0.1 m/ns, which is appropriate for most of the castle and verified where possible using depth reported from excavations). Note that features are visible in very lightly processed data; this data is presented in an effort to provide a reference for continued study of the castle.

additional discussion of future work to address challenges has been added to the former “Disucssion” section: (p.35-36, lines 625, 648-652, 656-658, and 662):

Discussion

Features identified with GPR provide insight into lingering questions and unresolved historical debates about the castle’s past, including the history of the lapidarium area (Figure 30).  This room is much lower in elevation than the rest of the castle.  An additional hypothesis arising from these investigations is that the lapidarium and adjacent courtyard with the same floor level were part of the pre-Bethlen phase of the castle.  Beginning with Bethlen, later modifications may have sought to provide a larger and more even footprint for the structure.  The accuracy of earlier plans of Corvin Castle has been substantiated by GPR surveying and analysis.   In scans of the northeast portion of the castle (surrounding the Well Courtyard), there are reflections from wall foundations running East-West and North-South (Figure 31). This feature is shown in the Moller map, however no longer present above ground or indicated in other plans. The data attest to the presence of the crypt and the original fortress wall foundation through the chapel, as in some sources (e.g., Moller, Figure 24). The jagged path of a wall running through the Knight’s Hall in Moller’s map, possibly indicating the uncertainty of the 14th century wall, was verified as a continuous and curved foundation in GPR scans without the jog (Figure 29). Further investigations into the substructures and history of these areas of the castle are an excellent candidate for future targeted investigations aimed at resolving these questions and validating the results of this work.  These investigations could be performed as part of restoration efforts using a variety of methods, including small targeted excavations, higher frequency GPR survey, or other methods.

Conclusions

Even though historical accounts of Corvin Castle are sparse and can be inconsistent or difficult to obtain, GPR scans were able to confirm features reported in both plans and textual sources, especially the foundation of the 14th century fortress walls (e.g., Figures 15, 22, and 24).  In addition, GPR expands our knowledge of known features, corrects previous assumptions, and even identifies unreported substructures and architectural features, especially in the White Tower and the chapel (e.g., Figures 18,19, and 24). The extent of the bedrock foundations and transition to built-up masonry foundations was previously unknown and has been mapped using GPR in the scanned areas (Figure 34).  Moisture ingress identified in the Knight’s Hall and the lapidarium can be localized and investigated (Figures 29 and 31).  Unidentified anomalies can be sought out and clarified, increasing our knowledge of Corvin Castle and its complex history while preserving the integrity of the site.

GPR surveys help critically evaluate the quality of the information provided by historical documents, reveal previously unrecorded information (such as the locations of some of Bogdan’s backfilled 1969 trenches, figure 21) and provide clues to the internal architecture of previous phases (as it did in the present Noble’s Man Kitchen [Figure 16], the courtyard [Figure 23]and lapidarium [Figure 30]). The data also serve as a baseline for more detailed study; for this reason, both selected features and typical scans are presented here. At Corvin Castle, where invasive tests have been the primary (if not exclusive) means of investigation for the last two centuries, nondestructive methods like GPR are critical to the survival of this heritage site and our overall understanding of its history. GPR investigations of Corvin Castle will aid the present restoration effort and future efforts by providing important and more complete reference information about architectural features.

Author Response File: Author Response.docx